AMD Demos Dual-Core Athlon 64

DigitumDei writes "Dual core chips came closer to reality as AMD demonstrated their Athlon64 dual-core offering. The 90nm technology chip will use the same 939-pin infrastructure and cooling solutions as the current Athlon 64 chips, meaning that upgrading to a dual-core chip from your current AMD64 will require little more than a BIOS update. Available in the second half of this year, the chip will be added to AMD's current line (Athlon64, Athlon FX, Sempron)."

Comments

Am I Missing Something?

[Dragoon412]

I don't understand the hype about dual core CPUs.

As I understand it, they work almost identically to a SMP setup, meaning they don't offer much of any performance benefit in most apps (particularly games). They draw more power, they run at higher temperatures, etc.

Is there something I'm missing? Or is this whole dual-core mess really just SMP on one CPU? Because from what I've read on the likes of Extremetech, Anandtech, and so on, I'm not finding any reason to be impressed.

Re:Am I Missing Something?

[Anonymous Coward]

Well, during games one proc can run the game and one can run the OS...assuming you can dedicate one processor to a process? IANAAMDE.

Re:Am I Missing Something?

[DrXym]

they don't offer much of any performance benefit in most apps

And how many apps & other processes is your system running at the moment? Mine's running 58 with 518 threads.

Re:Am I Missing Something?

[teg]

I don't understand the hype about dual core CPUs.

As I understand it, they work almost identically to a SMP setup, meaning they don't offer much of any performance benefit in most apps (particularly games). They draw more power, they run at higher temperatures, etc.

SMP without the mess (extra CPUs, cooling, expensive/complicated motherboards) and cost is definitely something to be impressed about.

It should give a big performance boost to a multi app and multi thread environment.

Re:Am I Missing Something?

[forkazoo]

Yup, pretty much. But, some of us do more than play games. Also, as multiprocessing hardware becomes more common, game makers will begin to take advantage of the benefits. For me personally, when I want to use my box for general-purpose stuff, and it is running the mythtv backand and transcoding some files into MPEG4, and I am rendering a 3D animation, and so on... Well, having SMP sure isn't a bad thing!

Re:Am I Missing Something?

[drwtsn32]

Sure, benchmarking a single app on an SMP system often makes little to no performance difference, but SMP is fantastic if you are a heavy multitasker and work with several apps at once.

My first SMP system was a dual Pentium 133Mhz box. After that I never went back to a single proc... until the Pentium 4 came out. It's disappointing that this chip does not support SMP (except for the Xeon line). P4 hyperthreading helped bring back some SMP goodness, but it's still not as good as two real chips.

Personally I can't wait for dual core CPUs!

Re:Am I Missing Something?

[spac3manspiff]

I dont get it either,
If the main benifit is a "decrease in power consumption", shouldn't it be implemented on a laptop rather than a server??

Re:Am I Missing Something?

[Anonymous Coward]

The OS is responsible for distributing the workload over the CPU's, and the OS can even distribute one process over several CPU's, since each process is running at least one thread. So in the end, how well it runs over multiple CPU's depends on how well it makes use of parallell threads. The better it does that, the better the OS will be able to distribute the process' work over multiple CPU's.

Re:Am I Missing Something?

[i41Overlord]

As I understand it, they work almost identically to a SMP setup, meaning they don't offer much of any performance benefit in most apps (particularly games). They draw more power, they run at higher temperatures, etc.

The reason most games don't get a performance boost from dual CPU's is because they aren't programmed to take advantage of the other CPU. How many end-users' home systems have dual CPU's? Hardly any of them. There was no reason for game makers to go through the effort programming for something that 99.99% of their customers can't use.

With the new dual core chips, technically it isn't anything groundbreaking but it will ensure that there's much more widespread adoption of multiprocessor systems. With more of the userbase using dual core CPU's, game makers will have a reason to program to take advantage of it, and you'll begin to see games that do see a performance increase when using dual CPU's (or dual cpu cores).

Re:Am I Missing Something?

[Junta]

The big deal is that it is a way to get *cheap* SMP, motherboard sockets x2, the downsides being cooling such densities (read: reduced clock per core), and sharing a memory controller per two cores (which is what Intel SMP has been doing forever, AMD used memory controllers per processor in a NUMA fashion full time, and Hyper Transport to access memory not associated with the current processor).

Theoretically, the dual core clocks will add up to more cycles overall than a single core, but the single core will have more clocks per individual thread, so unless a game leverages threading very nicely in the processor intensive segments, a multi-core may be slower than a single-core for the high-end gaming scene, however for workstation/server/HPC fields, it is very exciting.

Re:Am I Missing Something?

[Shalda]

Effectively it's SMP on one CPU. Something I very much look forward to as I always build my desktops SMP. It also incorporates some of the overhead of SMP into the CPU driving down the system price a little bit.

While you're right that SMP offers little performance to most apps, I tend to run a lot of CPU hogs at the same time. Watch a DVD while waiting for a project to finish compiling or whatnot. It can also help keep runaway processes from sabatoging your system. I used to have a program that set its priority to 'AboveNormal' and would from time to time it would hang up in a loop. Since it was running at a higher priority, you sometimes couldn't bring up Task Manager to kill it off as the higher priorty thread always took precedant. And if all else fails, you can set up 2 SETI @ Home clients and process twice as many packets. But do your self a favor and set their priorty to 'Low'.

Also, you've got a chicken and egg problem. The reason there's so few programs that benifit from SMP is that there are so few computers that are SMP. When I was running some computer labs at a Big 10 university a few years back I was insisting on SMP workstations so that the CS students could learn to program multithreaded apps and see the benifits of it when it ran on 1 vs. 2 processors.

Lastly, my basement is very poorly insulated and gets a bit chilly in the winter. Anything to help warm it up and keep my fingers working properly is a good thing(tm)!

Re:Am I Missing Something?

[magarity]

they don't offer much of any performance benefit in most apps (particularly games)

Ah, rejoice irresponsible youth! Right now a $25,000 8 CPU machine that can no longer keep up with a decent sized corporate database needs to be replaced with a $60,000 16-CPU machine. After dual core hits the market, it can be upgraded for the price of 8 new dual-core CPUs and a BIOS flash. Less money for hardware == more money for bonuses... W00T! W00T!

Re:Am I Missing Something?

[SpongeBobLinuxPants]

meaning they don't offer much of any performance benefit in most apps (particularly games). They draw more power, they run at higher temperatures, etc.

from TFA:

For example, a processor with dual 2.0-GHz cores can deliver performance not all that different from a single-core 3.5-GHz part. More important, such a dual-core part will hold down power dissipation to a figure closer to that of a standalone 2.0-GHz CPU, allowing processing throughput to effectively double for not much more power.

and

At such speeds, single-CPU processors can often dissipate more than 150 W.
The dual-core Athlon 64 runs at a clock-speed of 2.4 GHz and has a maximum power dissipation of 100 W.

Re:Am I Missing Something?

[lsmeg]

Dual core is basically SMP as far as I know, so you're not missing anything, except maybe some of the benefits of running SMP. From my experience, SMP can go a long way to making a system feel less "sluggish". Tasks that will essentially lock up a single processor system can run in the background without making a system useless on SMP.

Now this doesn't usually add up enough to warrant a normal user to spend the extra money on SMP, but if dual cores become the replacement for the desktop line, they should (eventually?) be more affordable than SMP. Plus, if dual cores become ubiquitous, more and more software will no doubt be written to take advantage of SMP.

I can't see any downside to the push for dual cores, as we seem to be slowing down in how far we can push a single processor in clockspeed and performance.

Re:Am I Missing Something?

[dsginter]

And how many apps & other processes is your system running at the moment? Mine's running 58 with 518 threads.

But what's the processor utilization? On most systems, its usually less than 10 percent. So when a user does something, the bottleneck is usually not the processor. Its usually the hard drive.

Money would be better spent on RAID, rather than dual core or dual processor.

Re:Am I Missing Something?

[madprof]

Does the software you're running take advantage of the dual cores for multithreading though?

Re:Am I Missing Something?

[roach2002]

Some instructions on the x86 hardware have a delay; let's say I'm asking for something from ram or cache. This takes more than one cycle (3 from cache if I remember correctly). On a normal, non hyperthreading processor, the processor sits idle until that memory value comes back.

On a hyperthreading processor, the processor can do an instruction on another process in that time. So a second process can come in and do a couple ADDLs.

It may not make your game run any faster, but if you have something in the background that can complete what it's doing more quickly.

SMP with n processors can theoretically run n instructions in one cycle. Hyperthreading still limits you to one instruction per cycle per processor, but it reduces the number of cycles where the processor does nothing because it's waiting on other hardware.

So for you, it probably isn't worth it. But for some workloads, ie: some servers that are constantly delaying on memory, it can speed secondary processes up

Re:Am I Missing Something?

[ceeam]

There must've been some similar reasoning when ICs first came out. "What do they offer compared to the soldered circuits we know and love?". Well, for once, integration.

Re:Am I Missing Something?

[Rhys]

Don't offer much benefit... for games that aren't written to use them.

However lots of games could be very easily written to use them. Not to mention they offer immediate (small) performance benefits to gamers of being able to offload everything but the game onto the second processor. Windows, the mp3 player, the virus scanner, web browser showing a stuipd flash ad, the (speedhack... er I mean) "stats" addon you have running for your favorite MMO/FPS...

"But writing parallel software is hard!" is the usual refrain I hear. Oh really now? Every bot in a match can be it's own process, talking to the game server just like a player. That's trivially parallelizable.

Re:Am I Missing Something?

[oldmanmtn]

As I understand it, they work almost identically to a SMP setup, meaning they don't offer much of any performance benefit in most apps (particularly games).

I suppose if your idea of "most apps" is games, then this probably isn't an area that would be of interest to you

If you have any multithreaded app that is even remotely competently written, then it will benefit from dual cores or (possibly) hyperthreading. If your multithreaded app is full of "big locks", then dual cores won't help, and the application designer is a failure.

If you have a workload that has multiple processes running simultaneously, then it is also likely to benefit from dual core. It gets more interesting with business/server workloads, but "home users" can benefit two. Even something as simple as running xmms and gcc at the same time should go faster. Or running two instances of lame.

The real win with dual core comes from increased throughput. A single job/application/process isn't likely to go any faster, but a full workload of multiple, reasonably parallelizeable, tasks will be faster.

Re:Am I Missing Something?

[amaprotu]

Actually power consumption can be more of a concern for a server in a large server farm than it is for a laptop. It is one of the reasons blade servers don't actually do that well - you can't power a full rack of them with out some very extreme power (and cooling) solutions.

When building a large server farm power consumption very quickly becomes more important than any other factor including processor speed.

Re:Am I Missing Something?

[Waffle Iron]

And how many apps & other processes is your system running at the moment? Mine's running 58 with 518 threads.

Typically 517 of those threads are asleep waiting for IO or a signal, and the one piece of information that you are currently waiting for is being processed in the single remaining active thread.

Re:Am I Missing Something?

[JustNiz]

It depends on the app. Specifically if it is multithreaded or not. if it is, different threads can run on different cores.

Most current desktop apps don't use many/any threads therefore aren't as able to really capitalise on SMP architectures.

As dual-core gets more generally widespread, there will be more pressure/benefit for developers to write multi-threaded apps.

Re:Am I Missing Something?

[drinkypoo]

You can use an affinity tool to put your process on a single processor alone, and everything else on the other. This all but eliminates context switches in your program's context, while all other processes can continue to run on the other core(s). The only thing that consumes a lot of CPU that is not typically multithreaded is game software. Eventually, this too will be multithreaded, as we see more multi-core systems out there.

Re:Am I Missing Something?

[TheRaven64]

I only do two things that really tax my CPU. Compiling and video editing. Compiling is embarrassingly parallel, and make programs (including GNU make) have been able to take advantage of this for ages. Generally, the best performance can be achieved by running make with number of CPUs + 1 way parallelism. Video editing is similarly parallel, since most CPU intensive things are effects that need to be applied to a large number of frames, making it trivial to split the workload. I would certainly see a large performance benefit from SMP.

Before I abandoned the desktop in favour of the laptop, I had an SMP system, and it was nicer to use than my faster UP system, since single-threaded computationally expensive things could be run on one CPU leaving the other one free for UI-related tasks.

Re:Am I Missing Something?

[c.r.o.c.o]

SMP without the mess (extra CPUs, cooling, expensive/complicated motherboards) and cost is definitely something to be impressed about.

The motherboards supporting dual core CPUs should be identical to those running single core CPUs. I guess this is where having the memory controller integrated into the CPU really pays off for AMD since it further simplifies mb design. But in the past SMP motherboards weren't THAT much more expensive (at the most $100 extra) than similar single CPU motherboards. The main cost associated with SMP setups were the very expensive SMP CPUs, which were anywhere from 1.5 times or more expensive than regular CPUs. The pricing of dual core CPUs remains to be seen, but I think it'll still be cheaper than 2 separate SMP-enabled CPUs.

However I completely agree with the rest of your post. Not having separate heatsinks, large motherboards, etc is a definite advantage. Just because of that the market acceptance will increase very rapidly. I wouldn't be surprised if games and other CPU intensive apps started supporting dual core CPUs soon.

Re:Am I Missing Something?

[stecoop]

Does the software you're running take advantage of the dual cores for multithreading though>

So Intel is doing something crazy by making dual core processors because application haven't had to think about multiprocessors right?

Well, change is inevitable and developers can't stick their head into sand and stay that way forever. Intel recognizes that change has to occur in the development community to enhance performance their product line. What better way then introducing dual cores? This will force programmers to start thinking about programming their application with multi - user, threading, layer, etc' thus over time, the application will be better utilized for the future.

Re:Am I Missing Something?

[Anonymous Coward]

Yes, you are missing something. Many of us use computers for more than games. Java applications for instance are nearly always threaded and benefit greatly from multiple CPUs (at least on Linux, I can't speak for Windows). Or how about working on a PowerPoint presentation while your spread sheet recalc's. Most any app that has a GUI will be more responsive if other activities are going on as well.

Whether the price difference is worth it is another question. Also you can expect many threaded apps to break. If you haven't debugged a threaded app on a multiple CPU system, then you haven't debugged the threaded app. And how many developers have multiple cpu systems laying around?

Re:Am I Missing Something?

[drinkypoo]

I trust you are aware that you could have gone dual AMD? And that in many cases HT actually results in poorer performance?

Re:Am I Missing Something?

[Taladar]

What makes you think the database is CPU-bound instead of I/O-bound? And what makes you think the manufacturer of the two machines will allow this software patch instead of selling lots of expensive new systems?

Re:Am I Missing Something?

[Chris Kamel]

One cpu makes it less hassle for end-users to run and some shared resources should make it cheaper to produce than 2 cpus So you could call it SMP for the masses.

Re:Am I Missing Something?

[mattyrobinson69]

it does help:

one core for playing a game, the other for the OS

or one core for ripping a dvd, the other for everything else

as long as you dont only run one app at once you will greatly benifit, and even when you do (when you play a game for example), you still benifit.

Re:Am I Missing Something?

[Vaystrem]

I've had the pleasure of using a number of friends Dual CPU systems across the years.

The most relevant one to this discussion is my friends Abit BP6, Dual Celeron, setup. At the time he was running Dual Celeron 300s. Not that impressive right? Except that he was able to host our Unreal Tournament Server - and then join it with no lag for any of the players. Running Unreal Tournament by himself showed a 50% load on each processor. He was able encode MP3s, burn CDs, and play games simultaneously. Something I was not able to do, and wanted to for the sake of time saving. I did not want to choose one activity over another - a 'leisure' productivity issue if you will.

Fast forward to now. A nice dual cpu system would allow me to play games, encode movies / my audio files, simultaneously while running Distributed.net. Are dual core's absolutely necessary? No, but when you are doing some very intensive applications you can't do anything else. As well even though not all games are multi-threaded, various aspects of the game may be, the networking code/the sound code etc, meaning that the game may run somewhat better on a dual cpu setup.

Re:Am I Missing Something?

[plague3106]

You know, I've been hearing this alot on this thread, and I don't understand the thinking. You do NOT need threaded apps to take advantage of SMP. A SINGLE app will run faster (potentially) on SMP if it is threaded, but running a SINGLE application isn't the big benefit SMP gets you.

Its that Firefox can be run on one processor, while your MP3 player is running at the same time on the other. This in turn will speed up BOTH applications, since Firefox does not ever have to yield to the player, and visa versa.

Since there's more going on then just those two apps (various system process, etc), your machine should be faster as each process now only has to worry about HALF the number of processes it did before.

I don't understand why a largely tech audience misses that point. We're not in DOS anymore; the OSes we are using all run more then one process at a time.

Re:Am I Missing Something?

[DarthVain]

Just a couple of things. I used to run an SMP system before I fried it. yes dual core is supposed to be very much like SMP (symetrical Multi-Processing, I think is what it means).

One its is on the same chip. That means shared cache memory with I would think would be much faster than having to pass operations across busses and the like.

Two, you are correct in that running an SMP (and likely a dual-core) will have very little effect on gaming or single applications... 4-6% probably in most cases. However this is NOT its primary purpose as far as I am concerned. It is about sharing proccesing power. So if I wanna rip a DVD, play and mp3, surf the net, play a game, all the while processing vertor equations in the back ground it will help me do that. It will allow to multi-task much more effectly. My crappy SMP system I built could do more things at the same time than my newer faster DELL

Three, to my knowlege having 64bit rather than 32bit architechure in current present day applications is pretty much for all intents and purposes useless. Should be go back to the single core 32bit, 1ghz processor? No.

All it will take is someone to create some killer app using all these technologies that will blow current practices out of the water (dual-core, 64bit). They will make a bazillion dollars (probably MS), and then everyone else will jump on the band wagon and harold in a new generation of computing. However it seems to me that the software is where the bottleneck seems to be, it has to catch up with the hardware. Both sides must be profitable, they both have their models, this is how it works. Hardware will always come first it is the nature of the beast. Hell AMD may be looking at the server or serious workstation market (currently dominated by Intel probably), and want a piece. These are usually much more expensive anyway. Once the process it down pat it will be cheaper, or whatever they think the market will bear.

In the short term, dual-core is just a meaningless at 64bit, for the mainstream user or gamer. There is a neish for it, but really it is the developement process is what is important here.

my 2 cents.
DarthVain

Re:Am I Missing Something?

[ivan256]

You can't really take TFA seriously, since it's InformationWeek, and not a reputable source. Just look at your first quote. There are plenty of 2.0Ghz single core chips out there that can deliver beter performance than a 3.5Ghz P4...

When they talk about single core chips using 150 watts, they're talking about intel chips (almost certainly the Itanium). 100 Watts for a dual core 2.4 Ghz Opteron is a 60% increase in power usage over the single core version.

Of course the parent to your post is just plain wrong about not offering a performance benefit for most apps. Now that developers are starting to try and use Hyperthreading to their advantage, those apps will see some serious benefit from multi-core chips; benefit that users could only wish for with a P4HT.

Re:Am I Missing Something?

[The_K4]

These could really be helpful in video editing and encoding. If I could pull my 2 xeons and replace them with 2 dual core xeons i suspect i would see a big time decrease in compile times....

Re:Am I Missing Something?

[MindStalker]

Also with dual core memory controller on chip has the distint advantage of being able to share that on chip memory. Meaning core to core communication is significantly faster than chip to chip communication.

Re:Am I Missing Something?

[utlemming]

deliver performance not all that different from a single-core 3.5-GHz part.

Agreed...but that is generally. When it comes to games and other programs which rely on the higher clock speeds, it doesn't matter that the overall, aggregate performance is simular to a 3.5ghz machine. A single threaded app will run at the clock speed of the chip. So if the chip is clocked at 2.0, then the single threaded app will only run at 2.0, and not the aggregate 3.5. Then there is the issue of an operating system that can utilize the dual core. Right now making the leap to dual-core will do you little unless you have an OS that is dual-core aware (i.e. does SMP). It would be like running Windows on an Athlon 64 -- it might be fast, but you're not going to use the main benefit and features.

Re:Am I Missing Something?

[convolvatron]

yes. exactly, its just SMP on a single die. nothing to see.

except that its a very concrete example of the trend that started when intel decided not to chase after 4GHz.

this year, its duals. but next year it will be quads, and number-of-cores is going to be the dominant growth mode for commodity parts.

which means that performance sensitive applications are finally going to have to start exploiting concurrency. which is a big change indeed.

as pointed out elsewhere, it also means that vendors and customers are going to see a much longer lifetime for their system designs. which is going to have all sorts of interesting second order effects. cheap, no margin systems. truly modular architectures (ht, pci-e)

Re:Am I Missing Something?

[LWATCDR]

"Most current desktop apps don't use many/any threads therefore aren't as able to really capitalize on SMP architectures."
They don't? The desktop app my company sells is multi-threaded. The app we wrote to manage our support calls is multi-threaded. Frankly any program that has ability to have more than a single view should be multi threaded. The Phone call manager uses a thread for to check what the status for all the support techs is. A thread to refresh the display of the waiting calls. and a thread for the main user interface.

Re:Am I Missing Something?

[JustNiz]

But the all the processes you're talking about really don't consume much processor time.

Even on a single core system, the CPU load of running an os, firefox and an mp3 player would be pretty minimal, especially since firefox isn't usally doing anything unless its actually loading a page.

It seems to me that the real benefit of dual core is for continuous high-load applications, such as hardcore gaming or bulk-serving database queries or web pages. The latter is probably more likely a to be running on an SMP server box than a desktop anyway.

Unfortunately the forthcoming dual core cpus probably won't make much sense for gaming either as their clock speeds will be at lot lower than the bleeding edge single cores. Most game engines are still optimised for single core performance.

Re:Am I Missing Something?

[Donny Smith]

>Money would be better spent on RAID, rather than dual core or dual processor.

You're right about that.

Unlike CPUs which become worthless in less than 2 years, RAID h/w last a bit longer.

Some five years ago I bought an Ultra2Wide SCSI 320 card and a (at the time big) 8GB HDD - I paid $400 for the card and $250 for the HDD.

I still use the card - I haven't checked but it should be as fast as SATA II I guess - and the SCSI disk works too (although it's quite useless - I use it as dedicated swap disk).

In the meantime I went thru 3-4 generations of motherboards and CPUs (consecutive 100% wipeouts) and my RAID stuff still rocks...

By year's end I'll go not for a dual core CPU system but for what's today top of the line nForce4 system. Screw the hype.

Re:Am I Missing Something?

[fitten]

Yes, a dual-core CPU is functionally equivalent to a dual processor CPU. One of the largest differences is cost. Last year, if you wanted a dual CPU machine, you had to buy a dual socket motherboard. Dual socket motherboards are more expensive to produce than single socket motherboards. The obvious difference in cost is that you have to buy (and charge for) another socket. But you also have to have all of the voltage regulator circuitry (x2) and the motherboard traces to memory, between CPUs and all of that. This makes dual socket motherboards more expensive... more traces running all over require more area, more area = more cost. It also makes the boards more complicated because there are lots more traces that can be bad when it's made. So, a dual socket motherboard can easily cost 2x that of a single socket motherboard and it is probably larger too.

For examples, go to the Tyan web site and look at their boards. The cheapest dual socket motherboards you'll find are in the $250 range and the better boards are around $500. They are also E-ATX (extended ATX) so they require bigger cases. Also, they have more power requirements so you have to buy big power supplies with special connectors. Pretty much, the cheapest you can put together a dual Opteron rig (with all new parts) with even slow CPUs is around $1500.

Now, with dual core, the motherboards are no more complicated or costly than single socket boards because, well, they *are* single socket boards. The power requirements are lower than dual socket boards so you can use "regular" power supplies with them. You can also use the regular ATX cases. The cost for having a dual core computer compared with a single is simply the difference in the costs of the two processors... Considering, say, $700 for all the parts minus the cost of the CPU. Suppose there's a $50 difference in the cost of the CPUs, so maybe $1000 will get you into a decent dual-core/single-socket machine.

Also, as many folks have pointed out, there's MUCH more stuff done on computers than play games. Currently, I'm working on a project that takes about 20 minutes to compile on this single processor machine. Having a dual(core|processor) machine would boost my productivity greatly during the times when I have to do a complete build because I could use 'make' with '-j #' where '#' would be 3 or 4 instead of 1 or 2 as we do now on single cpu machines.

Re:Am I Missing Something?

[Anonymous Coward]

Because you can already do that on SunFire V880s? Replace the UltraSPARC IIIs with UltraSPARC IVs.

Re:Am I Missing Something?

[Thagg]

Multiprocessing, both discrete and multicore, will accelerate all compute-bound applications in the future. Right now we haven't reached a critical mass yet -- where programmers feel it's worth the effort to multi-thread all of their applications -- but we will get there soon. It's not an easy change, and there are a whole world of problems that programmers haven't had experience with yet, but either these programmers will learn or they will not be competetive any longer.

Face it, the days of increasing clock speeds is over. It's done. Finished. Kaput. The low hanging fruit has all been eaten.

On the other hand, the multiprocessing benefits are huge and practically untapped. There is every reason to expect that in ten years we have 64 or 256 processors on a chip. People who hope to be working in ten years better learn how to write for these systems.

Compare the P4 to the Cell. The P4 goes to unbelievable lengths (even literally, in pipeline lengths!) to run at a high clock speed. Its contribution to global warming is substantial. It's expensive. And, it's an absolute dead end. Intel has already abandoned it.

The Cell has eight much-simpler processors along with its Power core. It can, and will, compute 10 times as fast as a P4, if programmed correctly. The game programmers are going to be pulling their hair out for the next couple of years, but they are going to be the high-demand programmers of the next decade as they are the first over the wall of significant multiprocessing.

Thad Beier

Re:Am I Missing Something?

[krayfx]

things can only get complicated with sfware. if i am not wrong - people have always relaxed with more precessing power- if i can refer to quakr 3. with quake 3 it was isae level of optimisation so that the game could run better with a slow internet connection , as the connects improved, as with the time doom 3 got released - ID relaxed thier stringent optimisation - and adrian's even stated that for doom 3 - he's not working that hard, just asking users to upgrade thier connection. while this might sound unfair - i know the effort needed for optimisations - it speaks the truth - albeit, the 'generic' truth. you give better resources and a better environment to work with - we tend to relax. i am sure, we will live to consume more and more powerful cpus. despite none of the legion of software not adapting the x86 64 platform - people continue to buy for the hope of upgrading to the latest and the greatest - fighting obsolescence(...heck, even me). so whats with the dual cores? plenty!! why? dont ask me! i run an amd 64, hope to use a linux 64 bit ditro sometime soon. but does it really matter? everyone is pushed to the next level automatically! everyone ugrades despite the fact that people can do plenty with a 1-2 ghz machine.

Re:Am I Missing Something?

[hattig]

There are plenty of 2.0Ghz single core chips out there that can deliver beter performance than a 3.5Ghz P4

There aren't many 2.0Ghz P4s that deliver beter (sic) performance than a 3.5GHz P4!

Which was the point!

The fastest Athlon 64 you can currently get is either a 2.4GHz 4000+ with 1MB cache, or a 2.6GHz Athlon FX57 with 1MB cache. The latter has a 95W power draw already, but it is 130nm. So within the next few months, AMD will be selling a dual-core 2.4GHz Athlon "FXD" with a max power consumption of 100W. Even in a single threaded environment, you'd hardly lose any performance.

There's a rumour of a 2.2GHz dual core Athlon 64 at 55W! Sure, this will be a "low power" edition at $$$$. It will still outperform 98%+ of the computers out there at the moment in single threaded stuff.

Re:Am I Missing Something?

[flocculent]

Money would be better spent on RAID, rather than dual core or dual processor.

Actually RAID doesn't noticeably speed most things up in a desktop environment in my experience.

http://faq.storagereview.com/tiki-index.php?page=S ingleDriveVsRaid0 has this to say:
Dont assume RAID 0 offers increased performance for all or even most applications... and dont assume that transfer rates reflect application-level performance.

Perhaps the money is better spent on more RAM? :)

Re:Am I Missing Something?

[Just Some Guy]

But what's the processor utilization? On most systems, its usually less than 10 percent. So when a user does something, the bottleneck is usually not the processor. Its usually the hard drive.

That's most systems, but certainly not all. I wrote a web application in Zope that acts as a portal to our scanned document warehouse. Whenever a customer wants to access some of the data we're storing for them, we fetch a few TIFFs from a Samba filesystem, convert them into a PDF with ImageMagick, and send them out. A RAID wouldn't make a bit of difference to our setup, since even the comparatively slow network file retrieval is much faster than the image processing which is the real bottleneck.

Our system is idle probably 95% of the time. In fact, it currently has a 5-minute load average of 0.04. But in that other 5% of the time, we want it to respond NOW and not 30 seconds from now. This is a pretty common situation for server machines - relatively long periods of inactivity punctuated by short periods of frantic scurrying - and it seems reasonable that AMD is offering their 64-bit server chip with this feature.

Re:Am I Missing Something?

[SIGPUNKT]

What makes you think the database is CPU-bound instead of I/O-bound?

If he's hosting his database on an 8-way server, I would imagine he's got the performance monitoring tools to figure out what it's doing and where the bottlenecks are. If you've got a decent RAID set-up (0+1, or "RAID 10" in some circles, hooked up with FibreChannel or the like), then the burden almost always shifts back onto the CPU.

Re:Am I Missing Something?

[hattig]

Both AMD's dual-core and Intel's dual-core (Smithfield) will have separate caches.

AMD's dual-core will connect at the on-die SRQ which will mean very low latency connections between each core. Performance scaling could be very very good.

Intel's dual-core will connect via the FSB, which will mean higher latency and lower bandwidth connections between each core. Intel's dual-core is most like SMP therefore. You can guess the performance of Intel's dual-core processor by running a non-HyperThreaded 1MB L2 SMP machine at the likely clocks of 2.8,3.0 and 3.2 GHz.

64-bit is useful for the next generation of graphics cards with 512MB of onboard memory, simply for addressing purposes. Yeah, you'll be waiting for Windows XP 64 too. Also it does speed up a lot of applications when they are written for 64-bit.

Re:Am I Missing Something?

[ColdGrits]

Absolutely correct.

Indeed, you can do that across all Sun's UltraSPARC-III - based SunFire range, replace the single-core US3 with dual core US4 CPUs. Something Sun deliberately decided to ensure was possible right from the outset.

Re:Am I Missing Something?

[badmammajamma]

"They draw more power, they run at higher temperatures, etc."

Did you even bother to RTFA?? The reason they are using this technology is to REDUCE power and heat dissipation.

Re:Am I Missing Something?

[That's Unpossible!]

But what's the processor utilization? On most systems, its usually less than 10 percent.

Regardless of the fact that you pulled that number out of your ass ;-) there are many applications that are processor, not data, intensive. Also, in the case of servers that run multiple services, if one of your services has a problem and pegs out a CPU, your site is not completely crippled, and it is also much easier to remotely connect to the server to fix it. Ever try fixing a machine with one CPU that is completely pegged out due to one broken app?

Re:Am I Missing Something?

[amanpatelhotmail.com]

You definately raise a good point. But you need to understand that the real news here is that SMP will finally be available for the desktop users.

As a consequence more and more developers will be interested in writing SMP enabled apps. The way I see it is as a Chicken and the Egg problem. It'll be a catalyst for programmers and corporations to produce code for SMP systems.

Re:Am I Missing Something?

[magarity]

What makes you think the database is CPU-bound instead of I/O-bound?

CPU load versus disk queue length versus memory page accesses. But no worries about my off the cuff example being the only case; a very popular move these days is to run VMware (or similar) on an 8 or 16 way machine in order to provide 20, 30, or more "servers" doing light to modest loads so introducing dual core CPUs will allow even more virtual servers hosted on a single physical box. The cost saving potential is TREMENDOUS.

And what makes you think the manufacturer of the two machines will allow this software patch

It only takes one manufacturer to advertise "Buy our 8 way single core now and upgrade to dual core on the cheap later!". You can buy an 8-way now with only 2 CPUs and add later; there's no reason why you won't be able to buy 8-way dual core capable with only a few single cores and upgrade/add CPUs for years. Big corporate servers costing serious money are upgraded and/or assigned to different roles for a long time compared to desktop PCs. It's a whole different world. One company *finally* decommissioned a quad PPro server and donated to a nonprofit I know. That's been in service 24/7 for what, 14 years? It was probably over $50K when new and they have to get their money's worth out of it. Anyway, the manufacturers know that most IT budgets are not unlimited and customers always like less expensive alternatives.

Re:Am I Missing Something?

[ziandra]

"Is there something I'm missing? Or is this whole dual-core mess really just SMP on one CPU? Because from what I've read on the likes of Extremetech, Anandtech, and so on, I'm not finding any reason to be impressed."

The likes of you are not going to be impressed. A majority of computer users are far better off buying a faster single processor than a multi processor system. Even if it only costs 20% more for a dual core chip, most computer users would see more improvement by spending that 20% to buy a faster single core chip.

This may change as dual core processors become common and developers write multi threaded games. But until then, gamers are much better off spending that extra money on a better video card, more memory and rarely a faster processor.

There are not a whole lot of applications out there that are processor limited. Most are I/O limited. People who run applications that are processor limited know it and rejoice multi core processors. The rest of you can rest assured that you will see benefit from it in a couple of years and not worry about it until then.

Re:Am I Missing Something?

[sgant]

Waiting for someone to come along now and quote some article about how RAM doesn't noticeably speed things up in a desktop environment.

Pretty soon we'll talk ourselves into a corner where nothing speeds anything up and we were better off with a C-64.

(btw, I'm joking)

Re:Am I Missing Something?

[SpaghettiPattern]

But what's the processor utilization?

I recently ran a realistic stress test using standard business software on Xeon (dual cores) CPUs. The processes are very CPU intensive and require minimal I/O. I got both cores to 100% usage and I had no iowaits whatsoever. The result was that I could run two times the processes I could on a single CPU system. This is of no use for Joe Average but in a business situation the double CPU bandwidth makes sense.

Re:Am I Missing Something?

[Anarke_Incarnate]

pssst, this story is about AMD not intel.

Re:Am I Missing Something?

[LurkerXXX]

Running Unreal Tournament by himself showed a 50% load on each processor.

Your friend should have locked that process to one CPU. He lost a bit of performance doing context switching.

Why oh why did dual cores come out so long after BEOS is gone. Now that was a system built from the ground up for multiple cores. :(

Re:Am I Missing Something?

[adachan]

This is somehting that pops up everytime someone posts anything regardig multi-cpu setups. The Quake 3 engine made use of multi-cpu technology. It ran very smoothly on a nice dual athlon system. However, the real benefit comes in video editing and compression. Try it and you will see the benefit. Dont just ask why.

Re:Am I Missing Something?

[yorkpaddy]

SMP ready cpu's weren't always more expensive than there UP breatheren. The CPU's are almost exactly the same, but the manufacturers just disable the SMP ability and sell them for more.

Re:Am I Missing Something?

[stecoop]

S#!t I was going to say AMD/Intel and kept reading Hyperthread and the brain refused to say AMD part...

Re:Am I Missing Something?

[magarity]

If he's hosting his database on an 8-way server

Well, the database I work on is on an 8 CPU Xeon but the management of the thing has several layers, none of which involve me monitoring the actual CPU load versus IO. There's someone in charge of the physical care of the servers and someone else in charge of making sure the DB is up and running properly. I talk to them from time to time but all I know is that the thing performs decently right now. As soon as we get more users and more data (coming by mid-year) it might not keep up so well. It'd be nice on the budget to have it upgraded to dual cores without buying a whole new 16 way. Anyone wanting to donate a spare Superdome, please let me know.

Re:Am I Missing Something?

[Stonehand]

...or, rather, threads stall waiting for other transactions to relinquish locks. Regardless of how many cores you have, there are likely going to be limits on concurrency unless you have a strictly read-only database or you simply enjoy aborts and rollbacks.

Data cache misses are also a nontrivial issue.

Re:Am I Missing Something?

[kotj.mf]

It does when the purchase of, say, a used $4k H80 forces you to start taking bids on a new $10k 120A generator/transfer panel and $5k UPS. Not to mention the new AC to suck out some of that 90F heat that's building up behind the racks.

And that's for a couple of RS/6000's wto IO drawers and half a dozen 7133's.

Re:Am I Missing Something?

[default luser]

Where's the surprise in this? They havn't pulled any amazing numbers out of their hat, 55w is roughly %70 more power than a single core Winchester. The other %30 is probably accounted for by a new more efficient core revision. The .09 micron Athlon 64 Winchester parts already sip power AT FULL LOAD (27w for 3000, 30w for 3200+, ~33w for 3500+).

The design spec for the Winchester core says 63w TDP, but that's just so designers can eventually drop in a 2.4-2.6GHz+ Winchester core. These higher frequencies will most likely require higher voltage, and thus have substantially higher power consumption than the 3000-3500 Winchesters.

Or perhaps you could just, say, drop in a 2.2GHz dual core. So many wonderful options...

Re:Am I Missing Something?

[ivan256]

Where's the surprise in this?

There's no surprise. I think AMD does an amazing job at keeping power consumption low. I'm mostly annoyed by the 'apples to armadillos' comparisons TFA kept making.

Re:Am I Missing Something?

[drwtsn32]

Yes, I did try a dual Athlon setup (1800MP procs, IIRC). Unfortunately it was not a stable setup so I ditched it for a single P4.

I have not seen HT cause poor performance with how I use my system, and it does improve multitasking between several apps. I can't wait to get back to true SMP though.

Re:Am I Missing Something?

[drwtsn32]

I completely agree. I don't understand why people don't understand this. I think they would appreciate true SMP if they actually used it for a few days!

Re:Am I Missing Something?

[Atzanteol]

Mmmm. Imagine a game like Neverwinter Nights or some such spawning a separate thread to handle "world events" while the main thread controls the main action? Dual CPU's could be a new boon to gaming!

Re:Am I Missing Something?

[twiddlingbits]

RAID's whole purpose was NOT speed it was REDUNDANCY. Most RAID controllers are just an upgrade over the standard ones in most PCs, that's where the performance upgrade comes in. They move data faster but if your program can't process the data as fast as it comes in you really don't see a benefit. And even with RAID the disk size makes a difference, smaller is better in many cases.

Re:Am I Missing Something?

[pclminion]

Its that Firefox can be run on one processor, while your MP3 player is running at the same time on the other. This in turn will speed up BOTH applications, since Firefox does not ever have to yield to the player, and visa versa.

Only if those applications were maxing the CPU to begin with. An MP3 player on a modern processor only utilizes around 1% of its capacity. Firefox a similar amount. They can easily share a CPU with 98% of its capacity to spare. They might run imperceptibly faster due to better cache utilization, but the reality is that almost every application spends 99% of its time waiting for something slower, like disk or network.

The only sort of application that a typical user (i.e. a non-developer) uses that's actually capable of maxing the CPU is, say, video editing, or a high-performance game.

Re:Am I Missing Something?

[pclminion]

Get real. The only time I ever even notice that I have an SMP system is when I'm doing something crazy in a VMWare and a compile at the same time. My load average over the past few days is 0.02.

Yeah, I can build the kernel almost twice as fast as on a single-CPU system. So what.

Your average user simply isn't capable of maxing a single CPU, let alone two of them.

Re:Am I Missing Something?

[MightyMartian]

Well, I thought that any multithreaded app would benefit from SMP, as the scheduler could round-robin threads between processors/cores (or whatever method is the best for popping threads on to which processor).

Re:Am I Missing Something?

[MightyMartian]

Precisely. We run a web server hosting several domains as well as a busy little MySQL server. At some point we're going to have to move the SQL server on to another box, but our dual processor system does an awesome job of juggling.

Re:Am I Missing Something?

[pclminion]

If you have any multithreaded app that is even remotely competently written, then it will benefit from dual cores or (possibly) hyperthreading. If your multithreaded app is full of "big locks", then dual cores won't help, and the application designer is a failure.

You assume everything can be trivially multithreaded. That's utterly wrong.

First of all, multithreading is not always done for performance. Sometimes it makes the code easier to write. It's nice to be able to compartmentalize tasks into threads and control them with a boss.

Second, in the field of parallel programming there is a term called "embarrassingly parallel." This means that it's obvious how to multithread the task, in such a way that you get maximum benefit from each CPU. However, real-world tasks are rarely "embarassingly parallel." It's absolutely unfair to call the application designer a "failure" because the application can't be effectively multithreaded. Most things can't be.

There are many real-world super parallel applications (the kinds of things that run on 10,000 CPU clusters) that have properties like, to double the execution speed you must have 8 times the number of CPUs. This doesn't mean the application designer is an idiot, it means the problem is hard.

It's easy for people who've never done parallel programming to run their mouth off, however.

Re:Am I Missing Something?

[DrXym]

CPU utilization on my box is always 100% or nearly since I run SETI & Climate Prediction through BOINC. Dual CPUs means they don't have to be put to sleep as often so they run faster. But it's not just about BOINC.

For example, I build lots of source code. Things like Mozilla take 30 minutes or more to build. On my current system drags down other tasks such and the whole system is stilted and slow. With a dual CPU, one of processors can be doing the the intensive task while the other keeps the desktop running. Or I could possibly throw a -j4 onto GNU make and benefit from builds in a fraction of the time.

Or if I'm playing a game that plays background music, it might spawn a thread to play it. It might spawn another thread to be loading scenery as I walk through a level. A dual CPU means that those tasks can be offloaded leaving the other dedicated to handling the gameplay. The net result is less jumps, freezing and stutters and a better framerate.

Or I can be compressing / burning a DVD while getting on with something else without fear of having a buffer underflow on the recorder.

While my PC is a single CPU, I also own a Mac G4 500Mhz dual CPU system. It's not the fastest thing in the world, but what impresses me is that I can be building something on it (e.g. Mozilla) and the desktop is still responsive and fast. That's what a dual CPU gets you. I recently tried a Mac mini (clocked at 1.25Gz) and the different in performance was very noticeable - if I launched a video or something, and ran the mouse over the dock the desktop would start getting all jerky as it tried to service both tasks. My venerable G4 has a slower clockspeed than a Mac mini, but I've never seen anything like that before.

Re:Am I Missing Something?

[Zork the Almighty]

Hahaha, I guess that's the theory, anyway. Wait till Longhorn ships.

Re:Am I Missing Something?

[BFaucet]

I do a lot of 3D rendering and compositing... My 3D renderer can take advantage of multiple processors.

I also like to have a lot of big programs running at the same time. Like my 3D package, Compositing package, image manipulator, web browser, audio player, instant messaging app, and email client.

Currently I modify priorities in order to say, render a scene (low priority) while setting up the composite (high priority) it's going into.

It isn't at all uncommon for my system to be at 100% proccessor utilization for hours on end while I'm working on it.

There are plenty of folks out there who have a use for this power.

Re:Am I Missing Something?

[DrXym]

Not all the threads are asleep when I'm building something, running a virus checker, burning a DVD / CD, listening to music, watching a video or playing a game. Which is the times when you need a dual CPU. The rest of the time it doesn't matter that much, unless of course you're running something like SETI which I am as well.

Re:Am I Missing Something?

[c.r.o.c.o]

SMP ready cpu's weren't always more expensive than there UP breatheren. The CPU's are almost exactly the same, but the manufacturers just disable the SMP ability and sell them for more.

So in other words, UP CPUs were just as cheap to make, but they were marketed at higher prices. The end result to the consumer was that UP CPUs were significantly cheaper than MP CPUs.

It's the same in any field. I'm a fan of mountainbiking, and bike component prices show the same trends as CPUs. The manufacturing cost difference between a high end component and an average component is very small, but the final price difference can be quite large (50-100% more).

Re:Am I Missing Something?

[RailGunner]

Loop parallelization (Intel Compiler setting) makes a huge difference.

What is it? Let's say you have this:

for (int i = 0; i < 1000; i++)
{
...
}

The Intel Compiler will try to run i=0-4999 on one CPU (or core) while running 5000-9999 on the other. It makes a HUGE difference in a lot of image processing (TIFF G4 encoding, for example).

Re:Am I Missing Something?

[platos_beard]

Think about it this a minute. The heat/power problems mean CPU advances can't follow Moore's law by increasing the clock speed as they have in the past. Instead, CPUs are going multi-core.

Two CPUs may not be worth emailing home about, but look several years down the road and if Moore's law results in doubling the core count instead of clock speed, before too long you see a computing environment where dozens of cores are the rule on PCs.

Hell yeah that's a big deal.

Re:Am I Missing Something?

[stuartkahler]

Like MMX and SSE, this will benefit very few people for the first year or two. You'll only see the benefit if you beat the hell out of your machine with multiple apps at the same time and want smooth performance.

As the share of multi-core systems becomes dominant, developers will code and compile for the additional cores. Games stand to reap huge benefits; so do video encoding apps. Benchmarking current apps to find the improvement is futile. It's like benchmarking Doom 1 on a computer with a Radeon X800; the results are meaningless.

Re:Am I Missing Something?

[PSC]

As I understand it, they work almost identically to a SMP setup

... except that the two cores have to share the memory bandwidth, whereas with AMD Opteron SMP systems, every CPU gets its own dedicated HyperTransport link to its memory.

It really depends on what your bottleneck is. If arithmetics is the bottleneck, well, a dual core CPU will get you approximately twice the throughput.

If, however, your application can hardly utilize a single core CPU because it is limited by memory bandwidth, the second core will not help you a bit. My condolances. Switch it off.

(Unlike the XEON, which uses a shared memory controller in SMP systems, the Opteron uses a non-uniform memory architecture (NUMA) where each CPU has its own memory and controller; if a CPU wants to access another CPU's memory, access is routed through that other CPU's on-die controller.)

Re:Am I Missing Something?

[anandrajan]

Is there something I'm missing?

You can run a second OS on a desktop dual core CPU via VmWare, Xen etc. with one CPU set aside for the second OS. Intel is pushing forward Vanderpool (virtualization tech.) and AMD just announced that they were working with the Xen guys to presumably achieve the same goal.

Re:Am I Missing Something?

[doormat]

(at the most $100 extra)

Yea, but when the ASP (avg selling price) for a retail mobo package is about $100, thats a 100% increase in price!

Re:Am I Missing Something?

[router]

This reminds me of the SCSI-IDE debate; if you have used multi processor boxes as your primary workstation you will always want one from that point on. Same with SCSI back in the day, doesn't make sense until you use it and get used to how much better it is, going back is like getting your hair pulled. Netscape could crash and I wouldn't know it, I could encode mp3s without worrying if playing another one at the same time was going to get hokey, CD burning was not failure prone, etc. Just makes the experience better if you are actively doing more than one thing at a time. If you don't agree, don't buy one, I don't care.... I love duals tho, if I used a desktop instead of a laptop now it would be a dual anything (PIII 500ish+) with scsi.

andy

Re:Am I Missing Something?

[ArmorFiend]

I kinda still don't get it. I mean, I've looked at a lot of apps that were underpowered, but from a software standpoint it was impossible (or prohibitively expensive) to rewrite them to take advantage of multiple cpus. (I'm THINKING OF YOU, Allegro Common Lisp!).

Therefore I'd rather have one super-duper chip that runs 150% faster than an average chip, than two average chips. (regardless of whether the two average chips are on the same physical chip, or plugged into different sockets on the mobo). Is the 150% faster super-duper chip not possible with the same resources as the two-on-one chip?

Re:Am I Missing Something?

[lsmeg]

The parent's example of running firefox and mp3s probably wasn't the greatest example. I'll give another one.

I know someone who has an Athlon 2600+ system, gig of ram, etc... When he encodes and burns a movie to DVD, his system becomes unusable until the DVD is done. When I do the same thing on my dual Opteron system, my system simply becomes "sluggish".

It's not like most people are going to say "OMG my system is so much better with SMP!", but when you get used to using it, you don't want to go back. If dual cores become mainstream, it's a win for everyone.

And as far as games go, they're mostly GPU bound (there are exceptions of course). But even if they were all heavily CPU bound, we're slowing down on cranking up the clockspeed anyway. If abandoning dual cores meant Intel and AMD could churn out 5+ GHz procs this year, then that might be the way to go. But as it stands, both seem to think there's more potential in dual cores.

Re:Am I Missing Something?

[pclminion]

I use a dual Athlon 2400 SMP box at work AND at home. At least until last week, when I "downgraded" to a Mac Mini, and I can't tell the damn difference.

If you think SMP is somehow helping you as you browse the web and play some MP3s you're deluding yourself. I'm not saying you shouldn't do it, though. By all means have fun, like I said, I have a similar system myself.

By the way, if you're playing an MP3 and it starts to skip or lag because of other stuff that's going on, that's a sign your operating system sucks, or is misconfigured. There's this thing called priority. Boost it for the MP3 player, or lower it for the encoder or whatever else is going on.

Re:Am I Missing Something?

[ArmorFiend]

The only sort of application that a typical user (i.e. a non-developer) uses that's actually capable of maxing the CPU is, say, video editing, or a high-performance game.

Lets not forget the age-old benchmark suite: Dragging or resizing windows overlapping other windows. Xterm has no ghosting and only takes 5% cpu when I wiggle the mouse at max speed. Xterm is a highly optimized application. Firefox visibly lags and has recognizable garbage in the framebuffer before it refreshes, and it takes 70% cpu.

Its ironic that basic window manager ops continue to lag 2ghz systems.

Re:Am I Missing Something?

[pclminion]

While you are dragging a window you obviously can't do anything else productive, unless you've found a way to compose an email while simultaneously dragging another window with the mouse, and dragging the window only takes a second at the most, so what difference does it make what the CPU load jumps to during that operation? So a background app gets slowed down for a second.

Re:Am I Missing Something?

[MindStalker]

Exactly, currently we are maxing out at around 4GHz, expecting a 6.5GHz chip to appear soon is unrealistic, btw I think you meant %50 percent faster, because 150% faster than 4GHz would be 10GHz, while the 6.5GHz would be approimatly compariable to performace with a dual core 4GHz.

Though it will be interesting to see how AMD markets a dual core 4GHz, would they call it a 6500+ if it preformes what would be comparible to a single 6.5GHz chip?

Re:Am I Missing Something?

[bored]

What makes you think the database is CPU-bound instead of I/O-bound?

Because this is slashdot and he is running linux? Every stock (non SGI kernel) linux box i've run in the last year and a half can't sustain more than ~200megs a second without consuming 100% CPU because there is something wrong with the core VM in linux.

Re:Am I Missing Something?

[MindStalker]

Also from a single app standpoint you may not see a significant increase in speed. But when running several apps like most desktops do nowadays, you should see a good increase.

Re:Am I Missing Something?

[Bombcar]

IANAAMDE ?? Man, next time use spellcheck!

I Am Not An AMD Engineer, I guess.

Re:Am I Missing Something?

[theVP]

The only benefit that I have found thus-far is that you can divert one core to one process, and another core to a different process. Basically, it'll assist with multitasking. But any one program will get nearly no benefit from this whatsoever.

Re:Am I Missing Something?

[InvalidError]

whereas with AMD Opteron SMP systems, every CPU gets its own dedicated HyperTransport link to its memory

I was not aware there was such a thing as "Hypertransport" DRAM... last time I checked, the Opterons and Athlon-FX had on-chip dual-channel DDR controllers.

BTW, single-core Opterons also share some RAM bandwidth with the other CPUs over the HT links and multi-core Opterons would also share RAM/HT bandwidth along its cores. Single-core Athlon64s have only their own DDR controller so dual-core Athlon64s will have to share RAM bandwidth somewhat like the Xeons do - without the extra package/bus/chipset/others latencies.

Since I just upgraded my PC and laptop, dual-threading dual-core CPUs should be standard fare by the time my next upgrades are due. As a moderate-heavy multitasker, I am looking forward to that and quad/octuple-channel FB-DIMM.

Re:Am I Missing Something?

[rikkus-x]

I have a dual Athlon too, and I ripped the second processor out recently. I hardly noticed the difference. I used to use this box for compiling C++, which takes a huge amount of CPU time with gcc. Now I write .NET code, which compiles so quickly that I'm often left wondering if the compile has actually happened.

For general use (web browsing, playing games, listening to music, using office apps) SMP only makes a tiny difference. It makes the apps and the shell feel very slightly less sluggish when you're starting, stopping and switching between applications, but I'm not even slightly tempted to put the second CPU back in.

Rik

Re:Am I Missing Something?

[cynyr]

yes but what if i already have a dual 939 system can i upgrade to a 4way system but updating my bios? yes i know this would be over kill for just about everything(currently) but it would allow expansion/upgrades to be very easy/cheeper, go buy a dual core processor, update bios, install new dual core chip, sell old one on e-bay, and have a 3 way box, repeat for the other when you need/want to. i hope someone thinks of this update/upgrade scheme so that i can use it, it would almost be like buying a 4way board and only putting 1 chip in it now.

Re:Am I Missing Something?

[ckaminski]

Yup, try burning a DVD and a CDROM at 8x and 52x respectively, while copying a 6.5GB disk image over the network via Samba, browsing on firefox reading up on all the latest w3c Xml craziness trying to reeducate oneself, do a disk ghosting on one of my vmware hosts, and intermittently typing on gaim, or building and testing my latest Java craftiness.

I could definitely use another core or three.

Re:Am I Missing Something?

[Short Circuit]

If you have a couple working "worthless" 1.4GHz Athlons, drop me an email; I'll gladly pay for shipping.

Re:Am I Missing Something?

[aichpvee]

Wouldn't this be similar to a multiprocessor setup where the processing load could be distributed over multiple CPUs? Because there are a number of games that support that and I don't see why you wouldn't get a similar benefit from having dual core.

Re:Am I Missing Something?

[TheRaven64]

Bullshit. Simple cut and splice may not be CPU intensive, but add any effects, transitions, overlays etc. and the CPU usage on this machine jumps to 100% and stay there.

Re:Am I Missing Something?

[runamok1]

I think it is important for two reasons.

1. If few general consumers have a dual processor system (or a dual core CPU in this case) then there is no motivation for programmers to create programs that take advantage of them. So if everyone DOES have one of these then perhaps these programs WILL be written.

2. Chip manufacturers really can't add a hell of a lot more MHz at this point. Intel's strained silicon technology has not made the Prescott much faster in clock speed then chips that were out last year. They are now incorporating features to make them use power more wisely and increasing cache sizes. I think that was the point in changing the chips to "model #s" rather than clock speed. IMHO, this was a tacit agreement between AMD and Intel because they have to give SOME reason to consumers to buy the best and the newest.

I personally have had several dual Pentium III cpu machines and was very pleased with them. Even if one processor was cranking at 100%, the machine was still very responsive.

That being said, I think most general consumers could have bought a computer 2 years ago and be fine unless they are heavy gamers or into video editing.

Re:Am I Missing Something?

[Apro im]

mmm... RAID striping is meant to be a speed upgrade. It is very possible to have a RAID setup with no redundnacy, but about twice the throughput, by alternating which words get written to which drive. Of course, if one drive fails, you're fubar, but then, that's true on a single-drive system too. With a striped drives, you're just (roughly) twice as likely to have it happen.

Re:Am I Missing Something?

[lakeland]

I accept that is your experience, but what I did is save the money from raid and double the RAM. Over the three generations it probably ended up costing about what your raid setup cost.

In my experience, linux handles extra ram extremely well, while windows handles it awfully. A linux machine with a decent amount of ram will cache so many of your disk accesses than it feels like the disk is a blazing SCSI drive even when it is a 5400RPM sloth. Plus it copes with opening lots of programs, etc. in a way the SCSI drive machine won't.

Oh, I do buy the drives with 8MB cache, but that's because they're the only ones with 5yr warranties. I don't know if they really are faster since I don't have 2MB drives to compare against.

Finally, I will be buying one of the dual-core machines rather than the 3800+. Not because of any benchmarks, but because I've used true SMP machines in the past and they feel so responsive.

Re:Am I Missing Something?

[Apro im]

I don't think anybody disagrees that they'd rather have one faster chip, but really, wouldn't you rather have two faster chips? Let's think of a not-uncommon situation for me when I was a first year that made me glad I had two processors:
I played a lot of q3a, I liked a soundtrack while I blew up my friends. I ran both q3a and xmms simultaneously, no bottleneck (besides ram). Of course, a lot of the rendering got done on my then state-of-the art AIW Radeon 32 MB, but the AI (wwhen there were bots), etc all took up one of my blazing fast 1 GHz PIIIs while my audio thread never jumped.

Obviously, dual cores/SMP are not going to help you on a single, un-parallelizableprogram, but the likelihood of your average college user, for example, having a single program running at once is slim to none - and if you're running enough of them, they're not always waiting on I/O. Of course, I would've traded my 2 1 GHz processors for one 2 GHz processors anyday, but back then 1.3 GHz was pushing hte envelope.

Re:Am I Missing Something?

[ckaminski]

Properly designed database schemas will prevent such from happening, since (in all the models I've worked on), changes happen to a small subset of data per user, mostly not the same data, but sometimes overlapping, and most database accesses are read, not write.

Decent databases have row-level locking and opportunistic locks which help prevent most of these problems. Not all, most.

But you are 100% correct. You can never have 100%*Ncpu concurrency. There's always going to be some loss

Re:Am I Missing Something?

[ckaminski]

Um, the PPro came into production use in ~1995, so no more than 10 years.

Re:Am I Missing Something?

[Arghdee]

Lastly, my basement is very poorly insulated and gets a bit chilly in the winter. Anything to help warm it up and keep my fingers working properly is a good thing(tm)!

Surely you mean your Mother's basement... :D

Re:Am I Missing Something?

[Hes Nikke]

While my PC is a single CPU, I also own a Mac G4 500Mhz dual CPU system. It's not the fastest thing in the world, but what impresses me is that I can be building something on it (e.g. Mozilla) and the desktop is still responsive and fast. That's what a dual CPU gets you. I recently tried a Mac mini (clocked at 1.25Gz) and the different in performance was very noticeable - if I launched a video or something, and ran the mouse over the dock the desktop would start getting all jerky as it tried to service both tasks. My venerable G4 has a slower clockspeed than a Mac mini, but I've never seen anything like that before.

It's all about the RAM. Notice that i'm comparing a 500 MHz single G4 to a 1000 MHz single G4, and calling the faster one slow as ass. this is because the faster one only has 25% of the RAM! the faster machine also has a 2.66x faster bus (100 MHz SDR vs 266 MHz DDR used as SDR) and a loads better GPU, but it's still a PITA to run because it doesn't have enough RAM to run a web browser, (Safari) and a [word processor (Pages) || email program (Entourage)] at the same time without swapping.

How much RAM do you have in your Mac mini? how much in your PM G4? i'm willing to bet that you have a similar ratio i do. try upgrading the RAM on that mini and see what happens.

I build lots of source code. Things like Mozilla take 30 minutes or more to build. On my current system drags down other tasks such and the whole system is stilted and slow.
You should give gentoo a try. That metadistro does a good job at staying responsive while you build code. even with a single CPU/Core.

Re:Am I Missing Something?

[ckaminski]

not necessarily. If the AI was running on one CPU and the rendering engine on the other, than no, he wasn't losing performance at all. Context switching kills you when one thread needs results, but you have to pause it to go run the results on another thread. On two CPU's, if threadA's results are ready when threadB requires them, the period of time that Threadb is suspended is less than if threadA had to go generate those results because it didn't have time before it last got preempted.

In the best case, ThreadB returns within a few hundred CPU cycles. In the worst case, it's still waiting for ThreadA. Reality will place the scenario somewhere between these two extremes.

Re:Am I Missing Something?

[pla]

Unlike CPUs which become worthless in less than 2 years, RAID h/w last a bit longer.

Obviously this depends on your particular needs, but I would say exactly the opposite.

I keep a Linux fileserver on my home LAN, and roughly once per year I upgrade its oldest HDD (four total), by which time I couldn't even give away the one I swap out, a mere three years old.

OTOH, I built my current main desktop machine just under five years ago, using a Tyan S1834 motherboard and a pair of PIII/933s, all of which already counted as slightly older, fairly cheap parts ($150 for the MB, $70 per CPU, IIRC). Compared to my machines at work (a Pentium M 2GHz laptop and a loaded P4/3.2GHz desktop), my home desktop still "feels" a hell of a lot faster. I'll probably upgrade in another year, once the dual core Athlons come down in price a bit, but for now, I really have no need.

To those who call dual core (or older SMP) nothing but hype - Try using one before you criticize it. They don't "just" run more things at the same speed, they run a lot of things quite a bit faster. Anything multithreaded, my 5YO home desktop can actually keep up with the P4/3.2 at work. Anything single threaded and CPU-sucking, the P4 will do it in less total time, but the difference in responsiveness while running flat-out (even at low priority) absolutely amazes me. 5-10s to open an explorer window? No thanks! I can have a half dozen CPU-intensive apps running at once and explorer still pops open in less than two seconds.

Re:Am I Missing Something?

[ckaminski]

64 bit is USELESS for graphics cards. Twice the data space for zero payoff. 512MB is less than 1/8th the addressable value of a 32bit pointer. We're further away from 64 bit graphics cards than we are from 64bit home desktops.

Also, I can tell you from experience, that a simple port of 32 bit software to 64 bits on a single generation (AMD64, for instance) often results in a performance LOSS. (PTC had this issue with Pro/ENGINEER on DEC OSF in 1994)

For applications which shuffle data in and out of memory because 32 bits is not enough, you are correct that 64 bits will speed things up, just as moving from 16bit segmented memory sped things up on 32bit platforms. Not having to deal with complex memory management certainly speeds things up.

Why wait? I can have Linux today. :-)

Regards

Re:Am I Missing Something?

[Evil_Timmy]

The next iteration of the Unreal engine is specifically designed to take advantage of multiprocessor (and thus, multicore) setups, have a look at the latest press release.

Re:Am I Missing Something?

[ckaminski]

You actually HAVE dual core XEONs?

Re:Am I Missing Something?

[twiddlingbits]

One drive FUBAR on a single drive you mightbe able to do data recovery. If your data is striped it's harder to get things back. If you use RAID 5 and the parity drive goes you are in deep trouble. Even with RAID, you should do regular backups. The laws on complex systems actually say the reliability decreases with the more working components you add since you multiply the reliabilities (i.e .99*.99 = .98). A completely redundant system is really 2X as reliable, but striping isn't the same as redundancy. RAID orginally meant Redundant Array of Inexpensive Drives.

Re:Am I Missing Something?

[tbcpp]

The P4 doesn't support SMP? I know of two friends of mine who both have dual P4 systems. One is a Sony, the other a dell. Most of the programs they use don't even use SMP. It's enough to make a guy want to cry..

Re:Am I Missing Something?

[Silverlancer]

Well, I play EVE-Online, and thats written in Stackless Python, is multithreaded, and gets a near 100% boost from dualproc... ;)

Re:Am I Missing Something?

[RzUpAnmsCwrds]

"2.6GHz Athlon FX57 with 1MB cache. The latter has a 95W power draw already, but it is 130nm"

There is no Athlon 64 FX-57. The fastest available Athlon 64 is the Athlon 64 FX-55, at 2.6GHz / 1M / 130nm.

All 130nm Athlon 64 CPUs, excluding notebook parts, have a Vcore of 1.5v and thermal design power of 89W. Measurements of the Athlon 64 FX-55 actually place it at around 70-80W.

All current Opteron processors (excluding the HE and EE variants) have Vcore of 1.5V and TDP of 89W as well.

Winchester-based (90nm) Athlon 64 cpus have Vcore of 1.4V and TDP of 63W. As with earlier K8-based CPUs, the TDP is for the entire line, regardless of clock. Low-end parts consume substantially less power than indicated by the TDP.

Intel, on the other hand, understates their TDP. Their TDP is meant to reflect "typical" usage patterns. Actual P4 Prescott CPUs have been measured to draw as much as 130W - 150W when under heavy load. Intel is relying on the thermal protection built into the CPU to slow down the system if necessary and prevent overheating.

Re:Am I Missing Something?

[JustNiz]

Sounds like your problem there is the amount of disk I/O/bus contention going on rather than anything else....

Re:Am I Missing Something?

[fitten]

yes but what if i already have a dual 939 system can i upgrade to a 4way system but updating my bios?

No, because there shouldn't such a thing as a dual S939 system ;) However, if you have a dual socket S940 system, you should be able to do exactly what you are asking.

Re:Am I Missing Something?

[SQL Error]

The reason for the hype is simple.

You ain't gonna get higher clock speeds. Not any time soon.

You have a few options to make things run faster:

1. Higher clock speeds.
2. More instructions per cycle.
3. More operations per instruction.
4. More processors.

We're maxed out for the moment on 1. 2 requires a complete redesign of the processor. 3 only really works for multimedia apps, but expect to see more energy going into this with SSE4 or whatever.

That leaves us with more processors as the only viable upgrade path right now. And if you don't want to force everyone to buy expensive dual or multi-processor motherboards, multi-core CPUs are necessary.

Current games generally don't take advantage of multiple CPUs, but if game developers want more speed, they're going to have to fix that. They simply don't have any other options. (Even the Cell processor is effectively a multi-core chip; it just has two different types of core.)

Re:Am I Missing Something?

[phrasebook]

Firefox visibly lags and has recognizable garbage in the framebuffer before it refreshes, and it takes 70% cpu.

And having two CPU cores isn't going to help. It'll still use 70% of one cpu while the other will be at 0%. Firefox is an exception of course. It's user interface is... well, the less said about it the better. Why do developers spend so much time making sure their interfaces are skinnable in order to 'look good', only to neglect the way it works when you actually click, move, or resize.

Re:Am I Missing Something?

[mrchaotica]

More importantly, gamers are early adopters. Game companies will add dual CPU support even faster for the same reason they code for $500 video cards -- so they have more whiz-bang features to write on the box.

Re:Am I Missing Something?

[DrXym]

My dual CPU Mac has 512Mb. I believe the Mac mini in the shop was 512Mb too although I'm not sure. I recall thinking it wasn't the basic model.

The mini surely did suck though and totally put me off the idea of "upgrading" even though it was higher clocked. I think I'll wait a year or two until they get around to making one which has some decent specs out of the box such as G5, more memory.

As for Gentoo, unfortunately I'm talking about an XP box. While it would be nice if my main machine were Linux, it isn't practical for the work I do.

Re:Am I Missing Something?

[hattig]

Yeah, I know I made a mistake with the FX57 thing, but there's no edit functionality, or even an Append a Note functionality here, so I didn't bother.

Yeah, I know that AMD overstate the TDP, and that 2.2GHz 90nm Athlon 64s are drawing 33W of power and stuff despite the 67W TDP figure, and that Intel really play it close to the line with their TDP figures.

Re:Am I Missing Something?

[Lord Omlette]

The Free Lunch is Over, same thing you're saying, but now you too can beat people over the head w/ it.

Re:Am I Missing Something?

[Otterley]

I don't think the optimization you speak of will work. What happens if the code in the loop changes the value of i?

Re:Am I Missing Something?

[Miguelito]

I think that dual-core and making things more paralell is being pushed because they're having a harder and harder time increasing the speed of chips.

Look at Intel.. the Itanium hasn't had a speed boost in awhile, still 1.5GHz at the top. I believe there's a slated 1.6GHz bump soon, but they're mostly just cramming more and more cache onto the chips to make things speed up. ANd they're going dual core eventually too.

Pentium4/Xeons were going to be 4GHz according to them not too long ago.. now we know they're having huge problems and have basically given up on that target for now.

AMD is also slowing it's speed increases a lot.

So they only real way they have to show progress until some new processes can be found that increase speed, but don't have such huge losses of power to heat, is to try to paralellize things and do dual and quad core cpus. Unfortunately, in my work, I support chip designers and raw cpu power has been what we've needed. Too many of the EDA apps they use are single-threaded, single-cpu programs. Very few are SMP or threaded yet.. though I think the ISVs are going to have to start changing that ASAP.

Re:Am I Missing Something?

[slashdot1968]

That depend's on what sort of game your running. I like Starcraft, and everything fit's in memory and you never have to use your disk. I'm not familiar enough with the latest first person's, but if you have adequate RAM, what's the use of your hard disk?

Re:Am I Missing Something?

[pipingguy]

Apparently, Intergraph's SmartPlant Review (3D engineering model realtime viewing software) supports multi-CPUs but it runs OK on a single processor AMD 64. I wonder if it would run a lot faster at home on an older dual MP 2800.

Re:Am I Missing Something?

[pipingguy]

If your multithreaded app is full of "big locks", then dual cores won't help, and the application designer is a failure.

After reading that I visualized "vapour lock" while trying to pump gas into a car's tank. Would that be an appropriate analogy for non-expert readers such as myself or am I intuiting off on a tangent? Seems to me that electron flow can be roughly compared to really tiny fluid flow.

Re:Am I Missing Something?

[TheLink]

Do you mean SGI's version of Linux performs better? Which one is that?

What are the differences?

Re:Am I Missing Something?

[pclminion]

You're saying all this to a person who has used exclusively SMP systems since 2000. I have five years of experience telling me it doesn't make any fucking difference for everyday use. It's not "theoretical" in the slightest bit.

Maybe it's your operating system that sucks.

Re:Am I Missing Something?

[Tzarius]

The only sort of application that a typical user (i.e. a non-developer) uses that's actually capable of maxing the CPU is, say, video editing, or a high-performance game.

Don't forget some unoptimised p2p, most Java apps, archive tools, near-full-screen Flash movies, any distributed computing projects, not to mention the horde of handy background apps doing their thing.
All it takes is one spike in one of these apps for the user to notice that their system can't quite take the load. (Also, it gets quite annoying when a background app you want to keep running consistently spikes the CPU load every few minutes.)

Re:Am I Missing Something?

[mi]

I don't think the optimization you speak of will work.

It actually works already. Get yourself Intel's or Sun's (for Solaris) compiler and see for yourself.

What happens if the code in the loop changes the value of i?

A fairly trivial problem given today's state of the art and science of compiler design. I trust, gcc will have this optimization soon too -- if it does not already.

Also compiler errs on the safe side, but there is a standard called OpenMP, which specifies compiler pragmas, with which you can assure the compiler, that certain things can be parallelized, even when it is not obvious to it.

Re:Am I Missing Something?

[andrewleung]

laptops?

i.e. on the road: single proc... jacked in: dual proc...

Re:Am I Missing Something?

[andrewleung]

laptops.

ie. on the road: single processor running
jacked in: dual processor

Re:Am I Missing Something?

[Michael Hunt]

You're missing the full 6.4GBps to main memory which each discrete hammer CPU gets at the moment.

Although, that said, it's probably a win for a simple 2-way machine because there's no need for the attendent overcomplexity of NUMA implementations; all memory is in one zone, and considered 'local'.

Re:Am I Missing Something?

[Yohahn]

You have not done low latency audio recording.
Try running ardour, recording while playing a click track, and adjusting a slider. It takes a bit of tweeking to make that work out.

Re:Am I Missing Something?

[sigaar]

"Oh, I do buy the drives with 8MB cache, but that's because they're the only ones with 5yr warranties. I don't know if they really are faster since I don't have 2MB drives to compare against."

They definitely are if you have multiple things accessing your disc.

Re:Am I Missing Something?

[sigaar]

"But people that HAVE TO HAVE PC3200 vs. PC2700 memory or 10k rpm vs. 7200rpm drives... please, people, please!"

Unless you do CPU intensive stuff (in which case memory speed definitely has an effect - compare the performance of a P4 running PC133 SD-RAM to that of one running dual channel DDR400), more memory generally helps more than faster memory.

But you can't seriously, with a straight face, tell me the different between a 7200rpm and a 10k rpm disc is not significant.

Maybe the secretary won't care, but for the rest of us? Go get used to a 10k rpm disc, and then go back to the old 7200rpm. You'll see...

Re:Am I Missing Something?

[Bert64]

It makes quite a difference on a multiuser system linked to multiple dumb terminals...

Re:Am I Missing Something?

[Bert64]

Which is why the mozilla rendering engine (gecko) can be used outside of the mozilla interface itself.. There are a number of browsers that do this, such as galeon, epiphany and skipstone and all have a massively faster interface than mozilla itself.

Re:Am I Missing Something?

[Bert64]

More ram and a fast scsi controller or two would probably help you more than another cpu..

Re:Am I Missing Something?

[Bert64]

Well your Opteron cpu's on their own are a lot faster than his athlon 2600.
Aside from that, i'm glad to see intel and amd concentrating on multi core, this will result in better multithreaded apps being written.. Multiprocessing can incase your speed TODAY.. faster individual processors will take a lot of time to develop.. You could go to CRAY or SGI today and get a system with 4096 processors, and multithreaded apps running on those systems could quite easily oblitterate any single processor system.
Basically, it offers an upgrade path now (add more cpus) rather than having to wait for new processors to become available, currentl single processor systems are well down in the single digit percentages of the performance levels that exist today.

Re:Am I Missing Something?

[Bert64]

But you made the point about OS, were you running the same OS on both machines? Also PPC processors are very different beasts from x86 chips. PPC is a much more modern design and is designed for multitasking OS's, unlike x86 chips.

Re:Am I Missing Something?

[Bert64]

Well, does it load the whole image before it begins processing it? tiff images can often be very large files, it might make sense to start processing it the moment it gets some data and not waiting for the transfer to finish. Samba isn't the quickest or most efficient of file transfer protocols either..

Re:Am I Missing Something?

[Bert64]

Well if the parity drive goes you replace it, raid5 is good so long as you dont suffer 2 drive failures at once, that's the point of raid5.

Re:Am I Missing Something?

[Just Some Guy]

As a general answer, we're optimizing as appropriate for our profile data.

Re:Am I Missing Something?

[Tower]

Either your loop is to 10000, or the one proc should be doing 0-499 and the other 500-999...

I'd hate to see a compiler do ten times as many loops as I asked...

Re:Am I Missing Something?

[Josepdin]

The principal benefits are related to the manufacturing process (it's cheaper to manufacture a dual core than it is to manufacture 2 single core chips) and to the integration between the 2 cores for access to I/O and leveraging cache at certain levels. The bus between the 2 CPU's in a dual core can be faster because of the 90 nano manufacturing. It's not about gaming, AMD themselves indicated that gamers should stay on the single core. It's about business processing where, for example, Oracle might benefit from a faster cache-coherent environment. I am not an engineer, but in my business, we are always looking for the dual (and beyond ) cores because they are faster, cooler, and consume less resources (power, space, etc.) than their single core counterparts.

Re:Am I Missing Something?

[pclminion]

I'm not denying that there are application where it makes a difference. I was very careful in my statements to say that TYPICAL use doesn't benefit from SMP. I admit that "typical" is a very vague term.

Re:Am I Missing Something?

[pclminion]

But you made the point about OS, were you running the same OS on both machines?

I was running it on the same MACHINE, booting between an SMP linux kernel and a single-processor kernel.

Also PPC processors are very different beasts from x86 chips. PPC is a much more modern design and is designed for multitasking OS's, unlike x86 chips.

Are you SERIOUS? Since the 80386 the x86 line has had significant multitasking features. Task state segments, interrupt context chaining, full memory protection, all these features are explicitly designed to support multitasking. It is true that the x86 instruction set is fairly ancient but the internal design of the processor (at least modern ones) is just as advanced as a PPC.

Before you try to start an Apple/Intel war, let me state that I switched over to a Mac Mini last week for home use and I freaking love it. PPC does rock, but don't try to tell us that Intel chips aren't modern.

Re:Am I Missing Something?

[bored]

Yes, for "high" bandwidth raid 0 disk subsystems. I haven't tested every SGI kernel but the latest kernel sources on their site (which we build for x86) do a much better job of keeping the CPU at reasonable levels during >200 Meg/second writes and non cached reads to/from the disk subsystem. In our case the CPU consumption on a 2x Xeon box goes from 100% and ~200 Megs/second to ~350 megs/second and about 10-20% CPU. That about 80% of the theoretical max for the disk subsystem we are using. In the near future I expect that we will be buying nicer disk subsystems (>600 Megs/sec) which will make the problem even more noticable on stock linux kernels (SUSE, RH, Fedora, kernel.org).

Re:Am I Missing Something?

[tabrnaker]

congratulations. You've spawned a whole slew of comments by people, as well as you, who haven't read the article. They claim lower energy usage and lower temps, for same performance.

Re:Am I Missing Something?

[MatrixXForm]

The motherboards supporting dual core CPUs should be identical to those running single core CPUs.

This is correct. In fact motherboards which are pin-compatible with the dual-core CPU's can be upgraded to SMP simply by upgrading the BIOS and then exchanging the CPU's. (Of course then there's the question of how many mobo vendors would make BIOS upgrades available; I'm sure most of them would rather sell a new motherboard ...).

Re:Am I Missing Something?

[ckaminski]

RAM is certainly not my problem. 2GB of RAM, 150M free, 250m of swap out of 1GB used. Swapping isn't an issue... Disk IO might be. Then again, I've always had issues burning CD's with my distro, and haven't tried others to see if it's any better... I know that XP seems to be a LOT more responsive when I do use it to burn things... :-/

How much power is "reasonable"?

[garcia]

They talk a lot about this being the savior of power-consumption but:

They are seen as the solution to power-consumption problems that have come to the fore as clock-speeds have increased beyond 3.0 GHz. At such speeds, single-CPU processors can often dissipate more than 150 W. In contrast, dual-core parts can reduce power consumption to more reasonable levels. For example, a processor with dual 2.0-GHz cores can deliver performance not all that different from a single-core 3.5-GHz part. More important, such a dual-core part will hold down power dissipation to a figure closer to that of a standalone 2.0-GHz CPU, allowing processing throughput to effectively double for not much more power.

Yeah, great, so it reduces power-consumption to "more reasonable levels" yet in every article I have read on this no one really mentions much more than that. What's reasonable? Telling me twice the speed for not much more power doesn't mean anything to me (other than marketing doublespeak).

What I want to know is how much money these processors will save in power consumption compared to how much more they will cost over their single core cousins... No one has said anything about that yet.

Now, also, how many OSs (and applications) are prepared for dual-core support? Are there any available systems that are stable and do that?

Re:How much power is "reasonable"?

[Triumph The Insult C]

the recent push for smp support in openbsd/i386 and openbsd/amd64 is largely in part of this new trend of dual core processors

Re:How much power is "reasonable"?

[oldmanmtn]

Now, also, how many OSs (and applications) are prepared for dual-core support? Are there any available systems that are stable and do that?

Solaris supports dual cores on both SPARC and x86. The UltraSPARC IV processors are dual core.

Any application should be "prepared for dual-core support". If the application even has to be aware that it's running on a dual core or hyperthreaded CPU, then the OS is broken.

Re:How much power is "reasonable"?

[Jugalator]

Now, also, how many OSs (and applications) are prepared for dual-core support?

I won't go into detail of applications since I have no idea which apps you're interested in, but Windows XP Pro supports dual cores (it runs its multi-core kernel even if you just have a Pentium 4 with hyperthreading).

Windows XP Home will not suffice though, which is a bit amusing since this might be the most common OS sitting in homes of gamers which are often the early adopters of this kind of tech nowadays. Unless they just pirated Windows XP Pro with a volume license key of course. :-)

Windows XP Media Center Edition, and Windows XP Tablet PC will support multiple cores though, probably in the same fashion as Pro.

Another one may give details about the common Linux distros, but I'd be very surprised if this support isn't in by far most modern distros, or can be enabled fairly easily.

Re:How much power is "reasonable"?

[bbrack]

The power increase between a single core processor and dual core processor is probably less than you think: I would not be surprised if it is in the 20% range on average.

Simply running the clock (and not performing any operations) on most processors will draw ~60-70% of the parts max power, which suggests that the loading on the part determines how much of that extra 30-40% is being dissipated.

Working under this assumption, worst case, a dual core processor would draw 40% more power than a single core processor, while effectively doubling your maximum throughput.

As for cost, I would be surprised if these processors debuted at more than 50% more than their single core cousins, and the prices would probably drop fairly rapidly

Re:How much power is "reasonable"?

[SirCyn]

Now, also, how many OSs (and applications) are prepared for dual-core support? Are there any available systems that are stable and do that?

Microsoft Windows 2000 and XP support 2.
Apple OSX supports 2.
FreeBSD support 4 (or more?). NetBSD supports 2 (or more?). OpenBSD is working on it (last I knew).
Linux 2.4.x and 2.6.x support 2+.
Sun Solaris has support 2+ for as long as I know.
AIX, HPUX, SCO Unix and all those support 2+.

Did I miss any?
Almost all OSes for the last several years have supported multiple processors natively. At worst these OSes would need a patch to update their SMP awareness.

Applications on the other hand, well they've been slower to change to a multithreaded moddel. Many server grade programs are ready. Most common desktop programs are not.

I have used a dual Athlon MP system for a long time now. The biggest difference I can tell you between dual 1.6GHz and single 3.2Ghz is that one process can not take over the processor. Even with modern preemption I can tell the difference when I have a second CPU processing my clicks and keystrokes. All I can say is "try one for a while, you'll get hooked".

Re:How much power is "reasonable"?

[TheRaven64]

Yeah, great, so it reduces power-consumption to "more reasonable levels" yet in every article I have read on this no one really mentions much more than that.

Power consumption is roughly proportional to the square of the clock speed (everything else being equal). Doubling the clock speed of a CPU will increase the power usage by a factor of 4. On the other hand, simply adding an extra core will only double the power requirements of the chip.

Re:How much power is "reasonable"?

[Jugalator]

The way I understood his question about app support was how well they're written with paralllell threads etc. I.e. not if they'll take advantage of the dual cores or not (yes, that's up to the OS to take care of), but how well they'll do it due to their design. An OS can't just take a thread and split up the work 50/50 on the two cores after all.

Re:How much power is "reasonable"?

[telecsan]

It helps if you read the content that you quote:
They are not stating that you get 2x the performance of a 3.5Ghz part for the same power. They are saying they can deliver the SAME performance (as a 3.5Ghz part) while REDUCING your power requirements back to the level of a 2.0Ghz part.

Re:How much power is "reasonable"?

[Malor]

If XP works the same way 2000 did, when you upgrade to a dual-core from a single-core, you will have to reinstall the OS for the second core to be activated.

2000 had two entirely separate sets of system files, one each for uni- and multi-processor. Even if you added a second CPU, if you didn't have the multiproc HAL to begin with, it simply wouldn't work.

Because XP is just 2000 with a facelift, I suspect this won't have changed. You are correct that if your initial install was on a P4, which 'looks like' two physical processors, XP would have installed its multi-cpu core.

If, however, you are installing a dual-core Athlon in, chances are quite high that you didn't do your initial install on a P4. So you won't have the multiproc system files, and you'll probably have to reinstall to get the second proc going. (A 'repair' installation may be adequate, and would be much less painful.)

Linux works somewhat similarly, but fortunately you can replace just the kernel, rather than the entire OS.

Re:How much power is "reasonable"?

[Malor]

Argh, hit submit too soon..... "if you are installing a dual-core Athlon", no 'in'. Argh.

Re:How much power is "reasonable"?

[ponos]

What I want to know is how much money these processors will save in power consumption compared to how much more they will cost over their single core cousins... No one has said anything about that yet.

Well, supposedly it will cost cheaper, because the chip companies do that to increase production yield. Building simple dual parts is easier than complex single parts. A dual 2.4 GHz will be much easier to build than a single 3.8 (which represents top of the line). Plus, if any part of the chip fails, they can still market it as "single core" or "lower cache" version by eliminating the faulty part. If the chip companies had good yield and regularly increasing speeds they wouldn't have been forced to invent dual core chips. The simple truth is they just can't build faster AND profitable single core processors. Even if AMD, for example, could build a slightly faster Athlon (e.g. Athlon 4200+ or 4400+) it wouldn't be as profitable.

The power consumption issue is highly significant (a) because it is related to heat dissipation and its consequent problems (b) because it will decrease your power bill. Judging from top-of-the line pentiums (have you seen their stock coolers?) we would either have to go water cooling or accept this thermal density (which is said to approach that of a nuclear reactor in heat/surface) as a practical limit. We are doing the latter and increasing chip area (dual core die is much larger) while keeping heat dissipation the same by lowering frequencies.

Now, also, how many OSs (and applications) are prepared for dual-core support? Are there any available systems that are stable and do that?

Well, any system that supports HyperThreading (which, ingeniously, paved the road for SMP support in everyday applications) should do just fine. To make this plain: windows XP and Linux will support a dual core chip like any SMP configuration. Software support is quite good for many modern applications plus there is always a vast benefit in responsiveness when running multiple applications together.

From a hardware standpoint, there is no magic involved, a simple BIOS flash is sufficient to support dual core chips in modern motherboards (at least in the AMD world). I can't tell about the intel world.

P.

Re:How much power is "reasonable"?

[Alioth]

OpenBSD is now SMP (since 3.6, which is the latest release).

Re:How much power is "reasonable"?

[hawkbug]

Good list, but I'd also list Linux kernel 2.2 since I used to run it on an SMP system back in the day. I can't comment on pre-2.2 because I didn't have an smp system back then, but other than that, good list.

Re:How much power is "reasonable"?

[Chris Colohan]

Good question. If you look at the recent history of CPU design (say, for the past 20 years), you see that the primary concern of architects have changed with the technology over time. In the 80's any design you came up with was limited by how many transistors you could squeeze onto a chip, and so everyone was worried about transistor count. By the 90's the transistors became abundant and small enough that a lack of transistors was no longer a primary concern of designers --- instead, they were much more concerned with the wires: how to pack them in and route the nicely, and how long it took for signals to get from one place to another.

These days, wires are abundant, transistors are abundant, and the main concern is power. Why?

a) GHz == Power. If you increase the clock rate, you increase the power drawn. So every time you want to make the machine "faster", you need to feed it more power.

b) More logic == more power. If you want to make the chip more complex or have more cache, it will need more power. But at the moment, it is cheaper (in terms of power) to add more logic than to add more GHz.

So why is power a problem? How much is too much? First, let's talk about total power: the total amount of electricity you have to feed into the chip to keep it running. This changes depending on how you utilize the chip, but designers have to design for the worst case.

There are two problems with total power: how do you get the electricity into the chip, and how do you get the heat it produces off of the chip. My understanding is that both problems are hard, but the heat one is currently harder. If you put too much power in, you need to get the heat out. More power == more heat. More heat == more expensive cooling solution. For example, if you had a low power chip you could get away with a simple and cheap heat sink. As it draws more power you have to add expensive (and noisy!) fans. Add more power, and you might eventually have to add a liquid cooling system, air conditioning unit, and a bunch of plumbing. Add more power, and perhaps you have to hire a team of workers to constantly feed dry ice into your machine, or some other creative cooling solution. So the fundamental problem is that once your chip draws enough power keeping it cool becomes too expensive. Right now it looks like 100 to 150W is the max power which can be kept cool by fan systems which are cheap and quiet enough that people will actually buy them. Nobody wants to buy a computer with plumbing for their home or office, liquid cooled machines are still only used for special business applications where they have little other choice.

Now total power is not the only problem. There is also localized power. Think "will a portion of my chip melt into a pool of molten silicon since I can't cool it off fast enough"? There is a nice story of how one of the first Alpha CPU prototypes cracked in half due to having too much power running up the middle of the chip... It turns out that this localized heat is a major limitation these days, and is a primary reason why Intel won't sell you a 5GHz chip. They don't know how to avoid this problem in a reasonable way. So, instead, they are going to improve performance by adding cores.

If you have one core, you have one central location in the chip that is constantly working hard, and so it is generating a lot of heat. Getting the heat away from that tiny point is hard. If you build a chip with two cores, then you divide the work over two points, and so each point generates half of the heat. Keeping that cool is much easier. With four cores....etc.

A research paper at last year's ASPLOS suggested that a good way to work this problem is to have a program run on one core until it gets too hot, then move execution to the next core, and repeat. This way you always have one core doing work, and the others are cooling off. We may end up having to do this.

Re:How much power is "reasonable"?

[Hoser McMoose]

I don't know... I expectt that dual-cores trickle down to the "home" market before 2050! Ohh, or do you mean that you actually think Longhorn might be released in a reasonable timeframe? :>

Actually this does tie in to an important point though. Apparently Microsoft will treat processors as "1 socket = 1 CPU" in terms of licensing for all future products, regardless of how many cores.

What that means is that Longhorn "Home Edition" (or whatever they call it) should run just fine on dual-core systems, even if it doesn't support classic SMP setups. Also things like MS SQL server will treat licensing costs on a per-socket basis rather than on a per-core basis, which may be important for multicore processors in servers.

Re:How much power is "reasonable"?

[Hoser McMoose]

Umm... just where are you getting that from?! Standard power formula for switching power is:

P = a * F * C * V^2

Power is directly proportional to both Frequency (F) and Capacitance (C), but proportional to the square of Voltage (V).

Frequency is really the only "easy" one in this formula. Capacitance tends to be kinda-sorta proportional to the number of transistors used, and voltage isn't really changing much on modern desktop processors (though things like Cool'n'Quiet and SpeedStep do dynamically change voltage in desktop chips for this very reason).

Of course, there are two real tricks to this equation that make it rather limited in it's usefulness. First is that "constant" alpha (err, 'a'), which isn't really a constant since there are dozens (if not hundreds) of factors that can affect this. The second is that switching power is making up a decreasing portion of the total power budget of a processor these days.

Even just 5 years ago switching power was >95% of the total power used by a modern processor. However now that number is down around 70% and it's expected to drop to about 50% next year with chips made on a 65nm process. The new problem is leakage current, ie power used even if the processor is just sitting around twiddling it's thumbs. This is a rather trickier problem and doesn't have any simple formula like the one above.

Re:How much power is "reasonable"?

[Erwos]

"Microsoft Windows 2000 and XP support 2."

To be more precise:
Windows XP Home: 1
Windows XP Pro / Windows 2000: 2
Windows 2000 Server: 4
Windows 2000 Advanced Server: 8
Windows 2000 Datacenter Server: 32

I'm not a Windows-user, but I wouldn't sell Microsoft so short.

-Erwos

Re:How much power is "reasonable"?

[Dan Ost]

An article I read earlier this week said that AMD is keeping its
dual-core chips at less than 100W. Maybe someone else who read the
same article would be kind enough to post a link.

Re:How much power is "reasonable"?

[fitten]

What I want to know is how much money these processors will save in power consumption compared to how much more they will cost over their single core cousins... No one has said anything about that yet.

Intel has already announced pricing for dual-core Smithfields. They start at ~$250USD for the dual-core 2.8GHz part. Compare that to your favorite shopping spot to other single-core Intel parts.

Re:How much power is "reasonable"?

[BranMan]

"Linux works somewhat similarly, but fortunately you can replace just the kernel, rather than the entire OS."

Or simply select the correct one when booting - several distros include SMP kernels by default.

Re:How much power is "reasonable"?

[bogado]

A research paper at last year's ASPLOS suggested that a good way to work this problem is to have a program run on one core until it gets too hot, then move execution to the next core, and repeat. This way you always have one core doing work, and the others are cooling off. We may end up having to do this.

Or instead you override this cooling feature, making all cores running at once and imerse in cooled oil, or dismantle an air conditioning unit to cool the chip or some other crazy cooling system. The added bonus, you gain an instant /. history, or who knows maybe two or three?

Re:How much power is "reasonable"?

[fyoder]

1) as far as the OS is concerned, there's not much difference between true SMP and dual-core (and only slight difference between those and HT)

IIRC, XP Home doesn't support dual processor, so if dual-core == SMP then it shouldn't support dual-core. But according to this article, XP Home will support dual-core, so perhaps Microsoft isn't regarding them as essentially the same thing?

Thankfully Windows XP Home and Professional will support dual-core right away, although Home will likely see only two processors even with HyperThreading enabled.

You can count on Linux not to deliberately restrict functionality, but with Microsoft the marketing department has input.

Re:How much power is "reasonable"?

[un4given]

2000 had two entirely separate sets of system files, one each for uni- and multi-processor. Even if you added a second CPU, if you didn't have the multiproc HAL to begin with, it simply wouldn't work.

Fortunately, this is incorrect. As described in Microsoft's knowledge base, a HAL change is all that is required to take advantage of the second processor. Windows NT 4.0, 2000, XP, and 2003 all are capable of this, although NT needed a utility uptomp.exe to accomplish this feat.

It is very common to have to do this on dual-processor capable servers when installing a second processor.

Re:How much power is "reasonable"?

[Blaskowicz]

When I upgraded my pentium pro 200 by adding another CPU, I saw Windows 2000 was seeing only one.. So I went in the device manager and found a option to toggle (it was the properties of the peripheral called "Computer" I think :D).
It's even better than linux, all it cost me was a reboot.

Re:How much power is "reasonable"?

[ckaminski]

There used to be a program on the resource kit called UP2MP that would do this upgrade without reinstalling the OS.

Not sure how this works with XP, since I have yet to use XP on anything other than a VMware guest session or a laptop. :-/

Re:How much power is "reasonable"?

[ckaminski]

Multicore is a stop gap to keep increasing overall computing power while taking extra time to flesh out these design issues in faster CPUs.

Thanks for a great post.

Re:How much power is "reasonable"?

[mako1138]

Sorry, your assumption is flawed. If you look at a picture of a dual-core Opteron die, you'll see that's it's essentially two Opterons sharing the same memory controller -- which is a very small slice of shared chip space. Simply "running the clock" still involves the WHOLE processor, not just a mysterious clock generator. It means refreshing all the cache, letting the execution units run stall instructions, etc. Every gate on the die is going to have some sort of leakage, refreshed each clock. That's why power scales roughly with frequency*voltage^2.

I've heard it straight from the horse's mouth (Kevin McGrath of AMD): a dual core processor has roughly twice the power output of a single core, if both are running at the same frequency. That's why AMD's dual cores will be released at several speed grades below the flagship line, ensuring that they run within the existing 90W power envelope. (The grandparent poster hasn't heard about this, it seems.) Meeting that target will be helped by the die shrink to 90nm.

Re:How much power is "reasonable"?

[JamieF]

>The biggest difference I can tell you between dual 1.6GHz and single 3.2Ghz is that one process can not take over the processor.

Hmm, what if that process has two threads?

<troll>Oh yeah, FreeBSD doesn't have threads. Never mind.</troll>

Re:How much power is "reasonable"?

[MojoStan]

Intel has already announced pricing for dual-core Smithfields. They start at ~$250USD for the dual-core 2.8GHz part. Compare that to your favorite shopping spot to other single-core Intel parts.

Intel will likely cut the prices of current single-core CPUs when the dual-core Pentium 4's (Smithfield) are launched in Q2 2005. So it's a little hard to predict what the price premium will be over single core CPUs. However, the pricing for Smithfield seems consistent with Intel's past pricing for mainstream single-core CPUs whenever they introduced a new core.

According to Anandtech's latest Intel roadmap article, the dual-core Pentium 4's will be launched at the following prices:

• Pentium 4 820 (2.80GHz, 2x1MB L2 cache, 800MHz FSB): $241
• Pentium 4 830 (3.00GHz, 2x1MB L2 cache, 800MHz FSB): $316
• Pentium 4 840 (3.20GHz, 2x1MB L2 cache, 800MHz FSB): $530

The fastest dual-core Pentium 4 is actually priced lower than the top-end single-core Pentium 4's (non-Extreme) which are typically over $600. However, the per-core clockspeed of the fastest dual-core (3.2GHz) will be less than the fastest single-core (3.8GHz).

At those prices, there doesn't seem to be any big price premium for the extra core. At each price point ($241, $316, $530), a Pentium 4 buyer can either choose a slower (per core) dual-core CPU or a faster single-core CPU.

Re:How much power is "reasonable"?

[pipingguy]

At worst these OSes would need a patch to update their SMP awareness.

That reminds me - if you have an older (~2.5-3 years?) AMD SMP MB and want to upgrade the processors you might have to flash the BIOS for it to see the faster CPUs. Maybe this is a no-brainer for most here, but since I'm doing just that now I thought I'd mention it.

Re:If this is anything like

[Junta]

HT is a way of letting one processing unit work with mulitple threads at once. Multi-core technology is identical to SMP, meaning more physical processors actually doing work, so it isn't token.

However, expect lower clockspeeds, two cores in that proximity causes a severe power/heat problem that would mandate reduced clock over single processor solution.

Why not two different clock speeds?

[Gates82]

Rather than using two identical chips, is there any difficulty in putting say a 2.4 Ghz chip with a 500 Mhz one? I would love to have the latest and greatest chip for gaming, and crunching through video and then have a low powered second chip to play my mp3's and surf the web, while the high-end chip is crunching through numbers in the background. Guess that's why I have a laptop, play on that while the desktop is doing its thing.

--
So who is hotter? Ali or Ali's sister?

Re:Why not two different clock speeds?

[Junta]

That is what a good realtime OS scheduler is supposed to do, time multiplex appropriately your processor. Not perfect, but for the kind of stuff you describe, you can get away with far from perfect and not experience a perceptible difference.

Re:Why not two different clock speeds?

[i41Overlord]

Most chips nowadays throttle down when they're not doing intensive calculations. They aren't running full-bore all the time.

Even my video card throttles down. The Geforce 6600GT and it runs at 350 mhz when doing 2D stuff, and speeds up to 500 mhz when playing a 3D game.

Re:Why not two different clock speeds?

[bbrack]

The main problem with this is that the processors share a clock tree and arbitration logic - if the clock multiplier is contained in the arbitration logic, then having one core at one speed and another at a different speed would be impossible.

If the clock multiplier is contained separately in each core, it would be possible - however, having different clock ratios on each core would considerably complicate the arbitration logic, since it would have to deal with different setup and hold timings when sending data to one core vs. the other - this would probably greatly increase your chances of inducing a processor error.

Trying to do this could also require a great deal more design difference between the two cores, which might cause many problems. It also would make it much more difficult to sell single core versions of dual core chips (i.e. one core fails, the other core is good - blow a few fuses to get the chip to look like a single core chip, and sell at as a single core)

Re:Why not two different clock speeds?

[MooseGuy529]

This would defeat part of the purpose of dual-core chips. By sharing the clock generator and distribution circuits between the cores, it saves power. If the clocks are different, they have to be separate and will use more power. (IANACD--I am not a chip designer.)

Re:Why not two different clock speeds?

[PDAllen]

First, because scheduling everything is much easier when you only have to worry about one clock speed (or at least, only a few multiples of one clock speed).

Second, because once you've set up a production process which can fab latest and greatest processor on one half of the die, it's really not much more expensive to fab a second latest-and-greatest on the other half than a four years old core. It's not two chips, it's one chip which has two processors on it.

Re:Why not two different clock speeds?

[wik]

It makes a lot more sense to have separate clock domains for each core. You can take two cores that have already been designed (with their clock trees) and throw them on the die. Why duplicate work?

There are already many clock domains in single-core processors (external bus, sometimes L2/L3 cache, etc..), so this isn't a hard problem.

Re:Why not two different clock speeds?

[bbrack]

While there may be separate clock domains per core, the clock multiplier circuitry may be contained in the arbitration logic, as this could probably improve the arbiter's efficiency.

I think that the design specifics of the arbitration logic (specifically, what speed it communicates with the cores, and how core-to-core communication is handled) is the determining factor in how feasible multi-frequency dual core circuits are.

Re:Why not two different clock speeds?

[toddestan]

What about adding a second, dedicated chip to take care of the stuff that now gets unloaded onto the CPU that used to be handled by seperate hardware? Stuff like the HDD Controller, network card, sound card, Winmodems. That would probably net a 10-20%+ difference in speed right there.

Re:Why not two different clock speeds?

[spectasaurus]

I really don't understand why would you want this. I would seriously doubt if such a configuration would be cheaper than two 2 GHz units. The beauty of having dual cores means that now you're able to have twice as many CPU's crunching away. Why you wouldn't do that if you had the chance is beyond me. I would love to see these pan out so I can get one for my own research.

I'm poor!

[shamowfski]

Does dual core mean dual price? With current fx-55's costing around a grand, what can we expect these to cost? 1,500-2,000? If AMD wants to remain competitive with Intel they are going to have to work on that. Who ever guessed AMD would be the one who had to lower prices to compete??

Re:I'm poor!

[shamowfski]

I got my thunderbird 1.2ghz for like 200 when it first came out, but then i'm assuming you are talking about some chip I have never heard of like an AMD spectotron or something.

Re:I'm poor!

[RealityMogul]

Prices haven't changed in a year on CPUs and RAM. Maybe this will at least force the low to mid level stuff to drop.

Re:I'm poor!

[JustNiz]

They will probably do the same as intel...Dual core chips won't use the leading edge cores, probably for power and heat dissipation reasons as well as marketing ones.

Therefore if you're going to buy a system for raw performance such as hardcore gaming, you'll still be better off for now buying a single-core FX-55. Most current game engines are optimised for single core anyway. It'll be intersting to see how quickly that changes, if at all.

Re:I'm poor!

[Anonymous Coward]

i dont know about amd's pricing but i did see intels a while back. as i recall the dual 2.8 was $241, dual 3.0 something like $300 and the dual 3.2 something like $400

Re:I'm poor!

[ThaReetLad]

According to AMD, no. IIRC they said recently that dual core Opterons will be inline, price wise, with the top end Opterons currently available, while offering much improved performance.

Re:I'm poor!

[MindStalker]

If your talking about the demo chip with the crazy water cooling to get it up to 1GHz, that really isn't a fair comparision. Wow remember when 1GHz was the holy grail.

Re:I'm poor!

[EnglishTim]

If you're so poor, why don't you just buy a Sempron?

Re:I'm poor!

[shamowfski]

I don't remember them being that expensive, but this corroberates (*sp) your story, sorry. Article

Re:I'm poor!

[corngrower]

Competitive? It's Intel that's not competitive. For the highest performance x86 chips, including dual cores, AMD's leadership will not be threatened by intel for at least another year. AMD is in the driver's seat and can charge the high prices because of it. It's something that until recently they've really never had the luxury of doing before.

Re:I'm poor!

[timeOday]

Well, I just don't believe they'll release a dual-core chip with single chip price, in effect doubling performance without much price jump. It just never happens. The best option is always just a little better choice than the second best option. If not, whoever is pricing the best option is leaving money on the table.

Re:I'm poor!

[Hoser McMoose]

More specifically, the chips cost $1,299 when they were first released.

Many people seem to forget that this was not considered abnormally expensive for a processor not all that long ago.

Check your licensing agreements first

[bigtallmofo]

Before you buy one of these dual-core processors for your server, make sure that your software vendor isn't going to double your price on you.

Oracle and others have announced plans to increase their revenue by charging people for multiple cores in their single processor.

Re:Check your licensing agreements first

[martinde]

> Before you buy one of these dual-core processors for your server, make sure that your software vendor isn't going to double your price on you.

I looked in /usr/share/common-licenses and I didn't find anything in there like that. Even if the prices doubled, $2*0 is still OK with me.

Gotta love free software.

Re:Check your licensing agreements first

[Anonymous Coward]

What's gonna happen when you try to install the Intel Dual Core "Extreme Edition" on a Windows XP box?

XP sees the single core + HT chips as 2 CPUs, but if the extreme edition dual cores are HT enabled, will XP allow "4" CPU's??

Re:Check your licensing agreements first

[madmac666]

what about for example mental ray, sold on a per processor basis. Does it mean i'd be able to render on two 'sockets', or two cores ?

Re:Check your licensing agreements first

[Jugalator]

That's an interesting question. :-)

First, I think Microsoft has previously said they'll only charge one license per physical processor. However, the problem is that Windows XP Pro supports only 1 or 2 logical CPU's. I wonder how it'll react. It seems like you'd not make optimal use of your system. And in case it does restrict itself upon working on two logical CPU's, you should hope Windows XP restrict itself upon the two cores instead of one of the cores and the HT feature on it!

Re:Check your licensing agreements first

[ThaReetLad]

Buying Oracle is like buying a Ferrari. If you have to ask how much it costs, you can't afford it.

Re:Check your licensing agreements first

[Wesley Felter]

That's why Smithfield has HT disabled.

Re:Check your licensing agreements first

[Richard_at_work]

From experience, it restricted itself to the first two CPUs it comes across, so logical CPU 1, and HT CPU 1. Discovered this with Windows XP Pro and a Dual Xeon w/HT enabled. To use both physical CPUs, we disabled HT on the motherboard - Microsoft is supposed to have released a patch which removes this limitation tho, it will only count the physical CPUs under licenses, but I havent tested this in the pat 18months or so.

Re:Check your licensing agreements first

[akadruid]

Buying Oracle is closer to buying a 40 ton truck. Not something you do if you can fit your stuff into the back of a car (or mysql).

Re:Check your licensing agreements first

[durdur]

Not hardly.
Oracle SE One Database - $149 per user. Dell will sell you a server plus database pre-installed for about $5500.
2004 Ferrari Modena : $150K and up.

Re:Check your licensing agreements first

[Hoser McMoose]

WinXP Pro, yes. Win2K Pro, no.

Windows XP makes a distinction between physical processors (ie classic SMP or dual-core) and logical processors (ie hyperthreading). That's why you can install dual HT capable Xeons on WinXP Pro now with no trouble (or a single HT capable P4 on a WinXP Home install).

However Win2K does not make this distinction. This isn't such a big problem since there is no Win2K "Home Edition", so a single P4 will work just fine. However if you want to use dual Xeons on Win2K Pro then you need to disable Hyperthreading (or spend the big-$$$ for Win2K Server).

My understanding is that we'll see the same thing with multicore chips as Microsoft moves towards Longhorn. With WinXP it will see dual-core as two separate CPUs, so they will only work on WinXP Pro and not on WinXP Home. However for Longhorn the licensing will be handled on a per-socket basis, so dual-core will only be seen as a single processor and Longhorn "Home Edition" (or whatever they call it) will work fine on dual-core chips.

There's also been a rumor floated around that AMD's dual-core chips will claim to be 1 physical processor with 2 logical cores (a la hyperthreading). I haven't been able to verify the accuracy of that rumor though and I'm personally rather skeptical of it's accuracy.

Re:Check your licensing agreements first

[fitten]

There's also been a rumor floated around that AMD's dual-core chips will claim to be 1 physical processor with 2 logical cores (a la hyperthreading). I haven't been able to verify the accuracy of that rumor though and I'm personally rather skeptical of it's accuracy.


I hope not because this may cause the scheduler to mis-handle the threads. Two compute intensive threads in HT is "bad". Two compute intensive threads on a dually(socket or core) is "ok".

Re:Check your licensing agreements first

[archen]

I used to say the same thing, but I know a guy who was trying to sell his Ferrari (since he bought a new one), and priced it at like $120k (I don't recall the exact ammount). Then some guy offers to buy it but they end up going round and round over $5000 difference in price.

I'm thinking to myself, one guy has a NEW Ferrari so what does he care, and why is the other guy bitching over $5000? Getting the brakes replaced will probably cost that much. But some people will squabble over price at the weirdest points I guess.

Re:Check your licensing agreements first

[ediron2]

Wa-ay off topic, but when I made an offer on a house once, the realtor (renowned locally for his arrogance) and the seller (an attorney, similar 'leet attitude) responded that they had received 2 offers, both below the asking price, and they gave me 2 days to bid it up with the other prospective buyer, subject to a complex, quirky page of additional conditions and auction rules. After my realtor gave the details, she and I blurted out simultaneously that it appeared that they were apparently trying to impress each other.

For some, ego is everything once they've got too much money. Your two guys probably were less interested in the 5k than the win. Once you see someone with this, you can sometimes find ways to tweak this urge in 'em for fun and profit.

Being wealthy doesn't destroy character. Wealth reveals character.

Oh, and the house? Took 4 more months to sell.

Re:Check your licensing agreements first

[advocate_one]

deity knows why this was modded redundant... it's the ONLY reply to the OP that actually mentions that you won't have licensing issues when using Linux... or for that matter BSD... the only people suffering with licensing issues will be those using proprietary apps.

Awesome! 939 Huzzah!

[Anonymous Coward]

I actually just purchased a socket 939 board for this exact reason. I'm extremely pleased with AMD for not forcing yet another motherboard upgrade on us based on chip advancement. I got a cheap Athlon 64 3000+, but two or three years from now I can go dual-core without getting a new motherboard, memory, etc. and I like that.

I understand that sometimes it's necessary to upgrade motherboards instead of just chips (FSB adn so forth), but for those of us who can't afford top-of-the-line, bleeding-edge stuff, it's nice to see upgradability for more than just a few months into the future.

Free Sony PSP from Gratis

Re:Xmas list material

[Anonymous Coward]

MORE POWER ARH ARH ARH!

Poor Tim Taylor... His claim to fame was nothing more than a strange barking while being laughed at by a fat man in a flannel.

If only he realized how hot his wife was. I'd never be in my god damn garage with her running around!

Similarly, no one should be putzing around with more power in their dual-cores when there are women to be had!

Oh wait, sorry, I forgot this is Slashdot (the REAL "tool time") ;)

Oracle and Dual Core CPUs...

[Krankheit]

With dual core CPUs coming from AMD and Intel, Oracle is going to be forced to change their licensing policy. There customers will likely think of a CPU as one chip, not one core, and refuse to pay for two CPUs.

Re:Oracle and Dual Core CPUs...

[Shalda]

The answer of course, is don't use Oracle if you can help it. :) They're just as evil as Microsoft, but they don't give away nearly as many free T-Shirts.

Re:Oracle and Dual Core CPUs...

[ThaReetLad]

I think not. Anyone running Oracle on x86 just isn't a serious player (except perhaps the new Sun Quad Opteron servers at $20k each. Nice), and if you are a serious player then you just pay what you have to because there's just not serious competition at that end of the market.

Microsoft on the other had DO care and have said their licences will be based upon sockets, not cores, so a standard XP Pro licence will allow you to use a dual socket, 4 way system for the same price as a normal 1 way, or even a notional 8-way (using SMT) on intel.

Re:Oracle and Dual Core CPUs...

[amorsen]

AMD has already declared that the CPU's will pretend to do hyperthreading. This should solve most licensing issues.

Re:Oracle and Dual Core CPUs...

[birder]

They are already making this pressure and it will be up to the hardware vendors and companies to push back. Do you pay 2x as much running Oracle on a 2 Ghz CPU than a 1 Ghz CPU? No, but Oracle once tried to charge like that until people pushed back.

Re:Oracle and Dual Core CPUs...

[Wesley Felter]

I'm sure Oracle will declare that you can't pretend to have fewer cores just to escape licensing fees.

Re:Oracle and Dual Core CPUs...

[Hal_Porter]

MS don't give away T Shirts anymore, they give away body armour.

Re:Oracle and Dual Core CPUs...

[PDAllen]

> What is next? Per clock cycle licensing?

That was next a few years ago. People made a huge fuss and Oracle backed down, but they did try to charge per FLOP, IIRC.

Re:If this is anything like

[Kiriwas]

You should probably get your terms right before you comment on it. HT is simply Intel's name for SMT (simultaneous multithreading). They didn't choose an optimal implementation and people shouldn't expect the same performance from it as you would from dual processors. SMT is simply an extension of the superscalar idea. Disconnect the dispatch mechanisms from the execution mechanisms and you can run an out of order processor a lot faster than an in order. Make multiple execution units and multiple fetches per cycle and you now have an n-way superscalar. A few more additions (mostly replication of units in the processor) and you can grab instructions from multiple threads instead of from the same thread (it is difficult to get lots of instructions per cycle from the same thread because of the high frequencies of branches in the code stream - and branch prediction isn't perfect). Dual core is completely different, they simply put two processors on the same chip. Dual core has the problem that it cannot share the same resources between the two threads. The resources (execution units, queues, etc.) are partitioned x amount for thread 1 and x amount for thread 2. The designs are really very different, depending on the use, sometimes dual core is better, sometimes SMT is better. AMD's planning on bring out Dual Core SMT where each core will have 2 threads running through it for a total of 4 thread running simultaneous. If you want more information about this "throughput computing" google for Sun's Niagara chip.

Better Question...

[evolutionaryLawyer]

Is this going to be another AMD innovation that will recieve no support from the software industry because they are not intel? Until processor intensive programs are re-written to be multi-threaded (which is much more difficult than it sounds, so I hear) this just allows AIM, GAIM, MSN, Firefox and Photoshop can be open and functional at once. I see productivity declining...

Re:Better Question...

[maraist]

Is this going to be another AMD innovation that will recieve no support from the software industry

Sorry, but UNIX platforms have been supporting/encouraging multi-threading for well over a decade. Solaris has been a key-player. You can't sell a 16-way SMT server unless your software can actually make use of it.

Hell, even windows NT has been pushing MT for a long time (selling 2, 4 and 8 processor versions). Both Windows and most UNIXes have very well established MT software bases. Even simple things like web browsers, office applications, etc, have long adopted MT since it improves user-response-time when the system is otherwise too slow. Look at the old MAC OS9.. That was the last major OS that wasn't fully MT.. Prior to 9, you had cooperative multi-tasking (you had to manually yield the processor back to the OS); not full blown pre-emptive multi-threading/tasking.

So, servers are fully ready to accept dual-core in any platform.. More threads means more throughput under heavy-load (if you aren't currently heavily loaded, then you're not purchasing a new machine, now are you?). For desktops, you will still see a noticable difference in performance with MT. If you've never had the system seem to hang for a few seconds when doing something heavy (virus check, disk-defrag, etc), there are two things that could be causing it; disk thrashing (no help from CPU here), and CPU-starvation. If the CPU is starved, then context switching back to the desktop still produces noticable delays; especially since x86 designs chips aren't the most efficient at ctx-switching.

While it's true, most games aren't going to take advantage of SMP, there are some which were specifically designed to have multiple independent threads (AI v.s. graphics). Quake 3 comes to mind. dual-core initially is being targetted at the server market ($1k+ chips). As killer-app designers demonstrate real performance benifits to dual core, we'll see more main-stream dual-core implementations... Eventually we'll see the retirement of single-core chips.

The article makes a very important case.. More's law is all about the performance increasing over time through innovation.. We're hitting a road-block with merely continuing in a straight line (shrinking transistor-size, using more exotic materials, adding larger caches, having different clock speeds for different portions of the core, VLIW, etc).. Each of these provides some performance enhancement, but most is provided in the first generation of the technology.. You have to keep coming up with radically different ideas which orthogonally enhance performance. While we can currently make 4GHZ CPUs with 2Meg L2 cache (see Intel), they have very large pipelines and the overhead (performance loss due to design trade-offs) provides tremendous deminishing returns. Moreoever, the power dessipation is a very real problem.. So much of the current technology is centered around managing power.. Too much power load causes voltage variations, and signal latencies (which prevent simultaneous arrival of 64-bit wide busses; same reason we went from parallel IDE to serial ATA).

Re:Better Question...

[Saeger]

You have to keep coming up with radically different ideas which orthogonally enhance performance.

Like a CPU with 3 dimensions instead of 2! Nanotech to save the daaay. heh.

This will be great..

[ein2many]

with that new 512 Meg ATI video card, we will get nothing for more.

Most games are multi-threaded

[wowbagger]

Most games have several potential threads running at once:

• Graphics rendering
• Sound rendering (compositing the various sounds together, and playing music)
• Game logic (monster AI, object movement, physics model)
• User input monitoring
• Network processing

An SMP system can greatly benefit a game designed to be truly multithreaded.

Even if the game is NOT designed to be multithreaded, there is the fact that one core can be running the game, while the other core handles interrupts, operating system processing, and other tasks.

The days of your computer doing only one thing at a time are long gone.

Re:Most games are multi-threaded

[Junta]

The problem here is that the multi-core will inevitably mean slower cores, and a lower speed at which any one thread can possibly be processed. In gaming, for example, it may be multithreaded, but most of the threads are pathetic (user input monitoring is actually really easy and insignificant for example), and one thread (maybe two very rarely) is a resource hog. Therefore even if you have overall 150% more clocks, but each core constitutes 75% of the clock speed of the single core alternative, then the resource-hog thread will actually see about 15-20% reduction in performance (allowing for 5-10% of the processor to handle the threads the other core would have offloaded). The unbalanced nature of gaming threaded tasks historically has been why SMP in gaming systems has been not worth the trouble/expense, and in a multi-core setup could be a liability.

The true benefit is in HPC/server/workstation environments, where there are computational tasks that have much more balanced threading models (*GREAT* cluster nodes, since applications are already written to balance as much as possible work amongst threads.

Now maybe if game developers manage to develop things right, this will change and things will be more balanced, but right now threads are used more as cheap ways to multiplex things that have to happen simultaneously in UP environments, and not designed to leverage SMP configurations.

Re:Most games are multi-threaded

[drw]

In theory this is true, but in reality creating multi-threaded games does not really benefit as much as you would think.

Mostly this is because of the extra over-head involved with synchronizing the threads. An obvious example is making sure the video and audio are synchronized so the 'Boom' happens at the same time as the 'Flash'.

When different threads are running on different processors with different loads, there is no way to guarantee that they will remain in sync with each other without some communication and/or blocking to make sure everything is timed properly.

Re:Most games are multi-threaded

[ckaminski]

Ever here of these things called timers? Yeah, every so often, in millisecond resolution, they fire, releasing a waiting thread for executing to prosecute it's chosen task. If you spend even a small amount of time doing setup you can make presumptions that two timer threads on separate CPUS will be released at nearly the same time (within milliseconds).

But you're right. Without communication, no amount of work gets done. Threads, or no threads.
Single threaded functions can block for any number of reasons. Only more so when you have to assure responsiveness by switching tasks at random periods. There's a reason preemptive multitasking kicked cooperative multitaskings ass, and this is it.

Re:If this is anything like

[drwtsn32]

However, expect lower clockspeeds

Really? From the article:

"The dual-core Athlon 64 runs at a clock-speed of 2.4 GHz ... "

2.4GHz is the speed of an Athlon 64 3400+ processor. I don't see a drop in clock speed here...

Oracle vs. AMD and Intel with Dual Core CPUs

[Krankheit]

With dual core CPUs coming from AMD and Intel, Oracle is going to be forced to change their licensing policy. Their customers will likely think of a CPU as one chip, not one core, and refuse to pay for two CPUs.

Re:If this is anything like

[MtViewGuy]

Actually, if you have dual CPU cores programmers could by default write their programs to automatically run in what amounts of symmetrical multi-processor (SMP) mode. That could be very useful for multimedia file editing, where sheer CPU power is a must, especially video editing.

People are finding out the hard way that when you start to edit videos downloaded from your MiniDV/MicroDV digital camcorder, the system resources used can increase quite dramatically.

Reality?

[AyeRoxor!]

Dual core chips came closer to reality as AMD demonstrated their Athlon64 dual core offering. ...As I understand it, that's about as close to reality as you can get...

Re:Reality?

[p3d0]

"When 900 years old you reach, look as good, you will not." - Yoda

I always thought that was kind of a sick joke. Do you say to your dog "when 80 years old you reack, look as good, you will not"? Or do you find a terminal cancer patient and say the same to him?

Re:Reality?

[AyeRoxor!]

I've always made light of death. Western caucasian cultures (who are the minority of the planet) are one of the few who 'fear' death and skirt around it and pretend they never think about it.

IMHO, Yoda was saying several things in this simple sentence. The first, of course, is, "I have lived longer than you will." A simple fact. Another is, "You are mortal (which is latin for 'subject to death'), and your time to find and accomplish your purpose is running out as we speak." Finally, but only as a postscript, he makes light of an otherwise serious statement by pointing out that he looks better than you will when you are his age.

Making light of that which is otherwise scary or unknown is a human tradition. It's psychologically healthy. If we can laugh at somthing, then it loses some of its mystique; it becomes more containable. We can 'wrap our heads around it' a little easier.

Cancer is not a fact of life for everyone of our species (yet). And my dog wouldn't quite understand it. But if I live to be 125 and someone remarks about my looks, I will free to reply that they will probably not look like me when they reach 125.

If I tell them, "Hey, youngster. Your time to find and accomplish your purpose is running out," it's easier to dismiss than if I make them actually realize it themselves.

Servers envorement, duh.

[KZigurs]

It's simple. Everyone knows that you can scale better in heavly multithreaded envorement by adding more CPU's instead of increasing speed of single CPU - it consumes less time on each CPU spent for managing tasks and context switches than on single, since each cpu has to work with smaller amount of threads.

And since currently even a desktop computer starts to approach point where there are hundreds of threads running (check in task manager or top) - this makes quite a lot of sense.

Also, a lot of people mistakenly believes that Hyperthreading in their intel CPU's brings similar benefits as SMP. No, it is just a nifty trick to keep the long p4 pipeline filled with as much data as it can. But multicore chips are in fact two CPU's in the same casing. (think - real hyperthreading :P).

Also there is the issue of simplified motherboard design (less traces for the same amount of cores), reduced packaging costs and higher computing density. All three being quite considerable points. As they say - computing today is all about integration. And multicore CPU's are one of the answers to allow simpler integration and allow greater flexibility (same costs to produce MB that supports single or multicore cpu. But the performance benefits - quite significant).

Re:Servers envorement, duh.

[ckaminski]

Really? Practically every benchmark I've seen of P4 and it's hyperthreading goodness gives a modest 20-30% performance benefit in certain use cases. That's FAR from 80% of "true SMP".

SMT has it's place. It's not a panacea, and it's not SMP.

Is your compiler single-threaded?

[mosel-saar-ruwer]

Watch a DVD while waiting for a project to finish compiling or whatnot.

Just out of curiosity: Is your compiler single-threaded?

Re:Awesome! 939 Huzzah!

[Anonymous Coward]

No-one's forcing you to upgrade at all. Stop believing the hype!

Hrmmmmm

[REDSECTOR1]

MAKEOPTS="-j3"

Horray

Re:Hrmmmmm

[Hal_Porter]

--email-c-code-to-china-to-be-hand-translated-into -assembler is 0.05% quicker.

Complexity

[Efialtis]

They have been expirimenting with multi-layered parallel processing for a long time, and I think this is the "realized results" of those expiriments.
We will see newer dual and multi-core processors come out in the future, and tha ability to parallel process with multiple chips on one board...
Should be exciting...

Re:Complexity

[Mistah Blue]

Hmmmm... why didn't you just say can we have a Beowulf of these? :-)

Re: Progress

[UCFFool]

This is what we refer to as 'progress'. Another way to refer to this is as a tool for profit. As has been discussed in everything from games, to movies, to personal electronics, Profit ceases without invention. 'Newest & Hottest' = Profit, and thus a sure line toward more New and Hot stuff...

Speaking of Hot stuff, I'm pretty much ashamed that I'm using slang from an idiot millionare's vocabulary. Just thought I'd share.

Could we make one core a security processor?

[gelfling]

Could we just kill the major security issues once and for all by dedicating one core 100% to security and encryption chores? I'd love to see hardware implementations for virus scanning, spyware, firewalling, encryption, VPN tunneling, authentication and smart patch managment that have free use of their own high end CPU

Re:If this is anything like

[Kiriwas]

I'm sure you meant to type: "Please leave, you're making us feel inadequate" :)

Sweet...

[sapgau]

Intel wake up!! See how easy it is to upgrade, no new socket layouts, no new motherboards.

Besides trying to determine what model is the Pentium dual core gives me headaches.

/owns AMD, trying very hard to repress fanboy attitudes.

Re:Sweet...

[D4rk Fx]

But Intel has plans to go up to 32 cores on a single CPU. I don't understand what you're expecting Intel to wake up about. I believe they plan to keep the current socket for the dual cores.

Re:Sweet...

[drw]

The reason AMD processors won't need a new socket or motherboard is because dual core was designed into the processors from pretty much the beginning.

Intel, on the other hand is trying to retrofit a single core design into something with 2 cores, and trying to do it as quickly as possible before loosing more ground to AMD.

Dual Processor, Dual Core?

[cvdwl]

So, what happens if you put two of these on a dual processor motherboard?

"I need more power, Scotty!"

And yes, I ran a dual Athlon MP1400 for several years, and loved it. It still kicks ass over my 3.4GHz P4 POS Dell my boss suggested we buy.

Re:Dual Processor, Dual Core?

[speculatrix]

there's nothing in the article to suggest that a dual-core amd64 can be used in a dual-cpu motherboard?

I would think that AMD will have to make multi-cpu versions of the dual-core chips, to support the inter-cpu comms, just as there are 1xx, 2xx, 4xx and 8xx versions of the regular opterons. But it'd be cool to cheaply upgrade a dual mobo to quad!

We have some dual-cpu opterons tyan motherboard machines here, and they are awesomely quick - we use FreeBSD, amd64 where possible but too often i386 mode because java and other things aren't supported in 64 bit mode at all yet.

rc5-72 rating is about 9M keys per second, twice that of a P4 2.8GHz machine; however, they are slaughtered by a dual Apple G5 system which achieves nearly FOUR TIMES the performance.

Re:Dual Processor, Dual Core?

[mako1138]

They've been talking about the dual-cores in existing n-way systems thing ever since dual-cores were announced. It would be highly disappointing if this didn't turn out to be supported. There's no real reason why it shouldn't be. The processors just need, as you say, sufficient HyperTransport linkages, because the memory controllers are on-die.

Re:Dual Processor, Dual Core?

[NerveGas]

It'll work just fine. In fact, NewIsys' Horus architecture ties multiple 4-way Opterons together into a single, coherant system with up to 32 processers - and their docs explicitly state that with dual-core chips, the 8 machines would effectively have 64 CPUs. Not bad for commodity-level stuff.

steve

Reason for Dual Core

[Anonymous Coward]

The speed of light in Silicon is a limiting factor in CPU construction. It takes electricity a certain amount of time to go from one side of the CPU to the other. As proc speeds keep going up, we get closer and closer to that limit. To combat this, CPU designers shrink the die. That way, the maximum distance needed to be traveled in one clock cycle decreases.

The cheap way of decreasing this distance is to split up the CPU into 2 distinct cores. Each core can be smaller, so it takes less time for electrons to get from one end of the core to the other.

This is why CPU designs are moving to multi-cores. They are hitting the limits of shrinking the components. As the components are shrunk, it gets harder to keep electrons flowing down the correct wire and not jumping across to other wires. Additionally, it just becomes excessively difficult to produce smaller and smaller wires inside a CPU.

So buy a cheaper processor

[mapmaker]

With current fx-55's costing around a grand, what can we expect these to cost? If AMD wants to remain competitive with Intel they are going to have to work on that.

If you need a car but you're poor, you buy the Chevy Cavalier, not the Chevy Corvette.

If you need a processor but you're poor, you buy an AMD Sempron, not the AMD FX-55.

Complaining that AMD needs to lower the price on their top processors is like complaining that Chevrolet needs to lower the price of Corvettes.

Re:So buy a cheaper processor

[shamowfski]

I'm really not poor. I was just making a point.

Re:So buy a cheaper processor

[mapmaker]

I'm really not poor. I was just making a point.

I know, and so was I. That point being that it is silly to demand that top-shelf products be priced like low-end ones. AMD has a complete strata of processors at all price points. You want the top of that line, you pay the top price.

Re:So buy a cheaper processor

[shamowfski]

I know you were making, and I know you know I was making a point. But there has to be a ceiling. My roomate worked an internship at AMD and the only reason they charge that much for them is because we will pay it. If we don't pay it, then they'll lower the price. I would just rather not go through the hassle of not buying one, eh?

Re:So buy a cheaper processor

[PitaBred]

So how do you propose they recoup their R&D investment? There's a lot more than just production costs going into the processors. They're really freaking reasonably priced, all things considered.

Re:So buy a cheaper processor

[Primal_theory]

But at least its not a viper, those are like 20k more, and the new vette (C6) seems like it will eat a viper...at least on the track, idk bout q-mile performance, but on a track the vette is king (stock)....

So what company will be the one with value? Will amd be like chevy, and have a nice dual core proc, for say $200 less then intel's "viper" which offers almost the exact same thing...but for $200 more?

I also find this a good comparison because it would be like the 2 cars in the fact that intel will probobly have more mhz (horsepower), but amd would have better multi-threading or such (track performance)...

Re:So buy a cheaper processor

[pipingguy]

If you need a car but you're poor, you buy the Chevy Cavalier, not the Chevy Corvette.

Or just buy a 5 year-old, used Corvette (95% as "good" for performance issues) for about the same price as a new Cavalier and avoid the drive-off-the-lot instant depreciation.

This of course will have an impact on bragging rights and might make you make extra trips when dragging offspring about, but, hey, driving a fast, sexy-looking auto is all about status and metaphoric dick-lengthening, no?

I confess that I do like fast cars for the experience of control over a machine but my perception is that many people buy them as lifestyle accessories and status symbols.

Cost per CPU for DB

[Erik_]

If you use commercial database, you will also see an cost increase due to the license-per-cpu. Lucky open source databases will take up on that slack.

Re:Cost per CPU for DB

[LurkerXXX]

FYI, not all commercial databases will be charging dual-cores as 2 CPUs.

Re:Awesome! 939 Huzzah!

[SIGPUNKT]

Well, up to a point. I'm getting pretty tired of my P-II/450 running Windows. OTOH, I still enjoy working on my 200MHz PPro running NetBSD. Soon I hope to enjoy the best of all worlds and pick up a Mac Mini. Which will probably be my main machine for another 4-5 years...

Re:If this is anything like

[masterofsw]

Actually, HT is not "simultaneous multithreading". Intel added extra hardware registers, etc, to make switching threads faster. HT still is a single threaded processor.

Re:If this is anything like

[SIGPUNKT]

programmers could by default write their programs to automatically run in....

HA HA HA HA HA HA!!! Obviously someone who has NO IDEA how to write software, or how much more difficult it is to coordinate parallel processes. This has been a hot research topic since I first used Concurrent Euclid back in the early 80s, and while the pthreads library and the Thread class in Java make it less difficult, it's still hard to do it well (and easy to do it so poorly you actually get worse performance).

Automatically? Oh yeah, I'll just drop in a #define SMP_MODE=TRUE and it'll just happen....

Re:Awesome! 939 Huzzah!

[Nimey]

That all depends on whether your mobo's manufacturer will release a BIOS upgrade to work with dual cores. I guarantee you that some makers won't.

AMD will come out on top if they play this right

[shr1n1]

This is another chance that AMD may be able to beat Intel. I think the main advantage woule be laptops. They will have bragging rights for the first SMP consumer laptop.

Sure just a "BIOS upgrade"

[SID*C64]

Like hardware vendors are going to pass up the chance to make an extra buck on you buying an upgraded "dual core" motherboard.

Gotta give it to them

[Reapman]

First off let me say I'm an Intel user... I'm glad AMD exists, Intel needed a good kick in the butt, but for reasons I won't get into I'm still an Intel person.

With that said... I love the fact that AMD allows something like a dual core to be potentially used in existing hardware.... every cpu intel releases seemingly needs a new motherboard and possibly RAM, I really wish this is one thing they'd start doing. Kudo's AMD, keep up the good work.

Not only multithreading

[ChrisMaple]

Pipelined applications also benefit from SMP. Each stage of the pipe can be assigned to a different processor.

Lockless techniques to the rescue

[bsmoor01]

I think this will be overcome with more atomic instructions from cpu vendors. Lockless techniques generally give much higher performance, and can often acheive the same goals as the old 'lock to enforce synchronization' paradigm you mention.

I use lockless counters heavily in the code I work on in order to reference count objects. Very handy, and much faster than lock-based counters.

The pain with lockless coding is that there aren't many portable primitives. So I have to maintain my own abstractions for every platform I work on, which is a pain since I'm in embedded systems. It would be awesome to have a standard (AND portable) lockless utility library. One day perhaps...

Re:Lockless techniques to the rescue

[battjt]

Your email doesn't work.

Can you give some examples or pointers to "lockless techniques"?

Re:Lockless techniques to the rescue

[bsmoor01]

Hrm... that email should work. I'll go harass my friend who runs the server and see if something's up.

Compare and swap is the only technique I've personally seen put to use. A quick google shows lots of hits: google it

There was also an article recently in C/C++ Users Journal involving lockless maps.

Basically, compare and swap will compare a given memory address with a value. If the compare succeeds, the address is loaded with a new, second value. Then, you can look at the result of the compare and swap operation to see if your new value was swapped in or not. If the value was not swapped, you just try again. No locking needed. If two threads have contention for a memory address, one of them wins, and the other one has to try again. The awesome part is that the threads will generally not have to try again. So the compare and swap is run only once, and avoids the unnecessary overhead of locking and unlocking some system resource (mutex, semaphore, etc).

I use compare and swap to implement an AtomicCounter class that I use for reference counting. Here's a quick snippet of the C++ code that increments my counter (sorry, but I couldn't get it indented correctly...):

long int counter = 0;
void incrementCounter()
{
long int temp;
do {
// get a copy of counter's current value to compare against.
temp = counter;
} while(!compareAndSwap(&counter, temp, temp+1));
}

The above code basically says "Hey, as long as counter == temp, set counter = temp+1. If counter != temp, then try again."

Now, imagine using compare and swap with pointers, and all kinds of possibilities open up.

Did the Uber-Parent Post RTFA?

[MSTCrow5429]

Um, yeah, "The 90nm technology chip will use the same 939-pin infrastructure and cooling solutions as the current Athlon 64 chips, meaning that upgrading to a dual core chip from your current AMD64 will require little more than a BIOS update. Available in the second half of this year, the chip will be added to AMD's current line (Athlon64, Athlon FX, Sempron)." seems to have been completely made up guesses, as this is no where in the source article. Oh, and also, cooling solution, not mentioned. Only that that the power dissipation was closer to single-core chip, not that the cooling solution would be identical to single core chips.

Re:Did the Uber-Parent Post RTFA?

[mako1138]

Just that TFA is not very informed, and rather sketchy. AMD has already said before that their dual-cores will use the same cooling solutions as existing processors.

See this article at the Inquirer.

Re:Awesome! 939 Huzzah!

[timeOday]

There could also be power issues since the dual core chips will need more juice. I know the current 939 socket and motherboards designs are "supposed" to be sufficient for dual core, but I'll believe it when I see it.

My experience with CPU upgradability is that it's nice when it happens but it's not worth paying more for a motherboard up front if there's no immediate benefit. It's not just the CPU socket, everything on motherboards (ram format, chipsets, agp, pci, sata) is churning all the time.

[insert experiences to the contrary here].

Re:Dual What?

[Hoser McMoose]

Intel CLAIMS that they will have their dual-core, 64-bit capable desktop chips out in 2Q of this year, perhaps a few months ahead of AMD... Of course, I'll believe it when I can actually go into the store and buy one (err.. assuming I wasn't broke! :> ).

Ohh, and in case you missed it, Intel just released the 64-bit capable P4 600-series this past weekend. They might have been a year and a half late to the party, but they have arrived now. These chips are actually available and in stock now, which is quite a change from Intel's recent CPU releases!

Re:If this is anything like

[Kiriwas]

Actually I beg to differ. http://www.intel.com/technology/hyperthread/ says explicitly that HT is a form of SMT (simultaneous multithreading). The processor contains multiple PC registers which allow it to actually follow multiple threads simultaneusly -- which means grabbing instructions from multiple threads simultaneously. Again, this is really just an extension of superscalar, which could only grab instructions from a single thread.

Re:Awesome! 939 Huzzah!

[p3d0]

I understand that sometimes it's necessary to upgrade motherboards instead of just chips (FSB adn so forth)

That's the nice part about AMD64: no FSB. It has the memory controller built-in.

Re:You have no idea what you are talking about.

[pivo]

rendering ..[is].. a purely sequential problem

Huh, try telling that to Pixar. Rendering certainly is something that can be split up among multiple CPUs. I wrote a distributed rendering engine several years ago, so I have some experience with it. Any decent rendering engine will support viewports (or something simialar) that let you specify the portion of the scene to render. That makes it easy to divide the scene by the number of CPUs and give each one a part. You create a semaphore to wait for all parts to be done and the display the frame.

Closer to reality?

[cmclean]

Dual-core chips are already a reality, Sun's UltraSPARC IV uses 2 UltraSPARC-III pipelines.

Perhaps the author means "x86 dual-core chips"?

Re:Closer to reality?

[Uncle Kadigan]

Don't forget, IBM's POWER4, POWER4+, and POWER5 are also multicore.

Re:You have no idea what you are talking about.

[Tim C]

Second, games are already multithreaded, but simply moving the sound and input to seperate threads doesn't do anything.

So move the physics and/or AI engines.

Re:If this is anything like

[fitten]

Assuming they release the part at 2.4GHz from the start. The part demo'd was an engineering sample, not the final released part.

The first engineering samples of the original Pentium that I ever saw were 75MHz parts. The first released Pentiums that I ever saw were 60/66MHz parts.

Re:If this is anything like

[brsmith4]

Some algorithms do not parallelize very nicely... especially, in general, the abstract phrase video editing. On a shared memory system (big sun SMPs or in this case, dual cores), it would be pretty easy to parallelize such things as filters and the transforms needed for encoding, but as for editing, I don't see that happening. I'm a beowulf guy so I have grown accustomed to course-grained, embarassingly parallel, to moderately parallel applications. These chips will fit in quite nicely with our MPICH/OpenMP environments but these dual core chips will likely do nothing for consumers rather than allow them to run more things at once. I can't say I seriously see software companies writing parallel versions of video renderers, encoders, or any other cpu intesive stuff simply for consumers to use.

Re:Awesome! 939 Huzzah!

[toddestan]

It wouldn't surprise me in the least if AMD changes their mind about this compatible with existing boards thing. I learned a while ago, just treat the motherboard and processor as a pair - buy them at the same time, and assume that if you upgrade one you'll have to upgrade the other.

However, I'm somewhat surprised that I could still use my 3 year old DDR333 memory in a new system (with a speed hit, of course).

Re:You have no idea what you are talking about.

[mogwai7]

Isn't this exactly what SLI, and the old dual 3DFx cards did? I believe each GPU renders a portion of the scene in these setups.

Re:You have no idea what you are talking about.

[Short Circuit]

Different kind of rendering, I suspect.

With raytracing and similar per-pixel rendering systems, for example, you can easily divide your rendering job between as many threads as you care to. With poly-based rendering, the CPU's work doesn't divide the same way.

That's not to say it's not divisible, though. Rather, concepts and techniques behind dividing it haven't seen enough attention to have developed into mature technologies.

Hope they make cheap ones

[springbox]

I hope that the processor manufactures make some economy version of the dual core CPUs like Intel has done with the Celerons, since anything new coming out from that industry seems to be generally pretty expensive.

Mmm dual core Celeron..

Re:You have no idea what you are talking about.

[Short Circuit]

What's that got to do with video games? Last I knew, 3DSMax was used for producing models and videos, not dynamic scene composition ala HL2 or Doom III.

not sure about RAID-0 but RAID-1

[MarcoAtWork]

is very noticeably faster on my system, for a bit I was running a single HD (due to other issues) and as soon as I added the 2nd and mirrored it my XP boot time has gone down by around 40% as well as Half Life 2 load times.

I know I am paying a penalty in write speed but the doubled read speed more than makes up for it. With HDs as cheap as they are now (I have 2x200G Seagate SATA) and RAID controllers integrated in most mobos (I have a Silicon Image in my a8n-sli board, which I prefer to the nvidia chipset raid) I think it's stupid *not* to go the RAID route, as with a very modest cash outlay (for a 2nd HD, or for a 3rd and 4th if you plan to run RAID0+1) you'll see a noticeable speedup, not to mention that if one of your HDs packs it in you won't be SOL.

For many businesses it makes sense

[Acer500]

I'm sure programmers and small businesses will take advantage of these.

You do know that many businesses do use non-server PCs for their databases, etc? At least over here we do (I have a normal PC being taxed by Sharepoint 2003, another with SQL, etc.)

I'm not sure how they will market them, but maybe they don't make sense for the home user (yet).

actually there is a differnce....

[kidoman]

i beg to differ,

however, anyone who has compiled [anything] in a SMP (2 way or 4 way) knows that SMP does make a big difference.

therefore dual cores will eliminate that mental bottleneck of having to buy 2 processors and the backwards compatibility is just an awesome gift!!!!

Re:actually there is a differnce....

[NerveGas]

Well... it remains to be seen whether a dual-core processer will cost any less than two single-core chips.

However, even if it does, you're still missing on one of the huge, principle benefits of the Opteron architecture: The multiple memory controllers. Two single-core chips will have two memory controllers, whereas a single dual-core chip will only have one.

Now, for compiling code, that's not going to make much difference, but for more memory-intensive stuff, it certainly can.

steve

Remember this post...

[dfghjk]

Remember this post to see how much of it is wrong. Sure the P4 is crap and always has been, but the cell is not a general purpose processor and clock speeds haven't finished going up. It's all about best computing bang for the buck when everything---die size, power, complexity---is taken into account. Oh, and game programmers already aren't the first "over the wall". All in all, they don't even really matter. We're getting multicore because it's the best way to spend the extra transistors going forward. Multiprocessing is old news.

what if that process has two threads?

[HornWumpus]

I am only a normally incompetent programmer.

I typically only generate one runaway thread at a time. Thus I am able to summon my debugger.

Re:If this is anything like

[dfghjk]

Whether L2 cache is shared or not makes no difference. You can build an SMP system without any L2 cache at all if you want to. What about multicore isn't "Symmetrical Multi Procesing" if it's such a falacy?

Re:If this is anything like

[dfghjk]

And yet we've had SMP PC's for quite a while but video editing packages do not consistently benefit from their performance. Same is true for image editing. As it currently stands, video editing on PC's is not a strong argument for SMP. Hopefully multicore will drive SMP penetration to the point where packages like that will actually use it. Premiere's SMP utilization is embarrassing and they have the big market share.

Serious amateurs and small professional shops are heavily constrained by their render farms

Yeah since they don't have them. Video editors don't use "render farms".

Re:You have no idea what you are talking about.

[Dahamma]

Second, games are already multithreaded, but simply moving the sound and input to seperate threads doesn't do anything. You will get less than 1% CPU usage on the second CPU

Try using Intel's new "HD Audio" (or VIA's Envy24T, etc) and see if it uses less than 1% for audio processing. Sure, if you buy a $200 Audigy card you'll probably get down to 5% CPU for audio, but I imagine the dual core will cost less than that, will have more CPU to spare and be more useful for general purpose applications, too.

Re:Why not forge the clock altogether?

[Vitriol Angst]

I remember reading an interesting article a few years back about research into not using a "clock". There is a lot of waste in modern processors it parts waiting for other parts. Out of order execution helps some, but you can just look at the techniques like the IBM G5 processor that uses NOps (Non Operations) to stuff the processor to handle non-optimized instructions or waiting on threads.

Some laptop processors achieve efficiency by allowing parts to cycle slower or do little "naps".

But getting rid of the dependence on clocked chips has enoromous advantages to heat, engergy and calculation efficiency. Current processors run every bit as fast as they can and then function at the lowest common denominator so that everything arrives "just in time", otherwise it will lose a cycle.

It would be interesting in this discussion if someone who knew more on the subject of CPU designs without clocks could add something. The term is "Asynchronous Logic". I found a few articles on the topic;
http://www.embedded.com/story/OEG20020824S 0001
http://www.win.tue.nl/async-bib/
And even Dvorak posits some ideas (which doesn't bode well);
http://www.pcmag.com/article2/0,4149,12122 87,00.as p

Personally, before we see clockless CPUs in the wild, we will probably see many more custom ASICs and architectures that bypass the CPU (many new bus designs allow this). But it makes me look back at how AMIGA got it right so many years ago. They had a chip to control sound, the mouse... many functions were separated and the sum of the parts could do impressive things.

Apple's new CoreGraphics and CoreVIdeo will finally move many imaging processes to the graphics card which can do them much faster than the CPU.

The new CELL chip might be party towards this.

A lot could be improved with a custom text processor or a few TI chips for DSP (like the early 640AV from Apple-again the innovator).

oooh... wow....

[Cryptnotic]

Three digit userid.... oooh...

Re:Awesome! 939 Huzzah!

[qtothemax]

It woulden't really suprise me if they changed their mind, but it would probably really work out to benefit them if they do go forward with the compatability. I am another who purchased a 939 motherboard and 3200+ with the intention of upgrading to dual core in 2 or 3 years. Unlike Intel, AMD does not make chipsets, so there is no loss for them there if people don't buy new mobos, but a lot of people will be more inclined to upgrade if they can keep their mobo, which is a big win for them. It makes a lot of buisness sense to keep the current socket.

Did you factor in....

[Kjella]

...the SCSI over IDE premium? PATA/SATA RAID is nice because you can use nice, consumer-priced disks. If you're one person, it's about 99,9% sure that what you need is lots of space (check), sequential read/write speed (check) and not massive random access (not check).

I bought a SCSI controller card once (and it was rather nice, because it was attached to a CD burner in the old days while buffer underruns was a problem), but I never found the price acceptable to buy a single SCSI disk. If I'm building a RAID server today, I'd use 2-300GB SATA disks and a SATA RAID card.

Kjella

Re:Did you factor in....

[Nutria]

If I'm building a RAID server today, I'd use 2-300GB SATA disks and a SATA RAID card.

RAID with only 2 drives?

Unless you want mirroring (RAID 1), that's totally stupid.

Re:You have no idea what you are talking about.

[Bert64]

But currently these games run on windows, which does context switches the i386 way and doesn't utilise the AMD64 architecture properly atall.

Re:If this is anything like

[masterofsw]

Yes, true, but continue reading the details:

"This is achieved by duplicating the architectural state on each processor, while sharing one set of processor execution resources."

From this page: http://www.intel.com/technology/itj/2002/volume06i ssue01/art01_hyper/p03_htt_architecture.htm

"Hyper-Threading Technology makes a single physical processor appear as multiple logical processors [11, 12]. To do this, there is one copy of the architecture state for each logical processor, and the logical processors share a single set of physical execution resources."

And:

"Logical processors share nearly all other resources on the physical processor, such as caches, execution units, branch predictors, control logic, and buses."

The HT processors take advantage of the idle time in one thread by allowing the other thread (logical processor) to use the core. There is only one core and hince NOT true SMT. You can only have true SMT by having multiple cores.

Re:If this is anything like

[Junta]

As mentioned before, demo tech frequently runs above the spec that the released product can achieve in stable fashion.

Also, keep in mind, that single core AMD chips are running at higher speeds than 2.4 GHz in their near-production future, so a 2.4 GHz *demo* is *reall* an indication of lower clock speed. By the time it releases, it will probably be lower than demo, *and* the single cores will have moved beyond 2.4 GHz...

Re:Of course! Its so simple!

[Tim C]

You can[sic] start rendering until you've done your AI and physics and everything else, so you know what you need to render.

What's to stop you preparing everything for rendering frame n+1 on one proc while you're rendering frame n on the other? What effect on the inputs to the AI and physics engine does the output of the rendering engine have?

C-64 power

[Muchsake]

Twenty years ago I knew a programmer who worked on colossus. The things he managed to get a C-64 to do were awesome. I couldn't match them till I got my first 486 (DX2 66 Mhz).

Re:Laptops

[bprime]

The Acer Ferrari 3400 uses an AMD Athlon 64 Mobile, which sits in a 754-pin socket (source here). The dual-core sockets have 939 pins. Also, your Ferrari would overheat with a non-mobile S754 - nevermind a dual-core S939.

Re:If this is anything like

[Kiriwas]

No, SMT IS by definition the sharing of resources, in many cases because there isnt a thread level partitioning of resources this can be FASTER than multiple cores, which for your information is called CMP, chip multi-processing. Just because SMT shares resources doesn't mean that it doesnt have twice the resources as a normal non-SMT chip.