AMD Launches Counterstrike Against Core 2 Duo

DigitalDame2 writes to mention a PC Magazine article about the AMD 4x4 enthusiast platform, which is meant to counter Core 2 Duo. The article observes that AMD is now facing many of the same business practices it used in its war against Intel. From the article: "While imitation may be the sincerest form of flattery, improvement can often be a slap in the face. Intel's C2D was designed with both low power and performance per watt in mind, two key design metrics that helped AMD cut into Intel's market share with the Athlon 64 and Athlon 64 X2. And, as preliminary numbers have indicated and final performance reviews now show, the C2D has learned its lesson well: its performance now tops AMD's Athlon 64 architecture by a substantial margin."

Comments

4X4 is more a marketing ploy than anything else

[Harry Balls]

4X4 sounds more like a marketing ploy to me than like a feasible solution for Joe Average or even Joe Gamer.

Why?

Consider the cost of Athlon X2 processors:
http://www.pricewatch.com/cpu/442067-1.htm
The least expensive Athlon X2 costs a cool 300 bucks, while the mid-range Core 2 Duo (Conroe) E6600 costs $315 (projected wholesale price).

Now factor in a more expensive (because of 2 processor sockets) 4X4 motherboard, two Athlon X2 chips at $300, and you wind up with a $350 to $400 surcharge for being an AMD fanboy.

The situation gets worse if you want a high-end system:
Two FX-62 will set you back $1045 + $1045 = $2090
http://www.pricewatch.com/cpu/992212-1.htm
and while this combination is expected to outperform a single Core 2 Duo at $1057
http://froogle.google.com/froogle?q=E6800&btnG=Sea rch+Froogle&lmode=online&scoring=p
factoring in the more expensive two-socket motherboard expect to pay a cool $1100 more than for the E6800 system.

Personally, I'll probably buy an E6600 ($315) or an E6400 ($240) as soon as they become available.

Re:4X4 is more a marketing ploy than anything else

[MarcoAtWork]

remember that AMD is slashing the prices of several X2 processors by about 50%, hence the price differential is mostly only the mobo differential, which I don't think will be that much...

Re:4X4 is more a marketing ploy than anything else

[Anonymous Coward]

Like another poster said, you are forgetting that there is a price cut coming, and if it is as much as they claim it to be, AMDs 4x4 will work rather well until the 65nm cores are out and then they can afford to ramp speeds. It is also important to note that this should be a PR move more then anything. Intel has been losing this fight for years, and they still have the majority of market share because of the crap people believe when they see commercials for Intel and/or Dell (Intel's bestest friend).

Benchmarks on systems have shown the differences to not be as massive as thought in high end games run at real resolutions, who are the people who matter most when it comes to FX and Xtreme processors. Actually, the biggest limit right now on modern games appears to still be the GPU. Honestly, we got two AMD v. Intel posts in one day...I would wait to buy your new E6600 or E6400 until a few more real reviews are done, and not just the Intel phantom system reviews...I hate those stacked previews companies do...

Re:4X4 is more a marketing ploy than anything else

[GoatMonkey2112]

The situation IS worse, because you HAVE to use FX processors in a 4x4 system.

Re:4X4 is more a marketing ploy than anything else

[Hoi Polloi]

They even talked about 8x8 (2x 4 core CPUS). Just 4x4 strikes me as wasted power that the vast majority of enthusiasts would never touch. Most of the time the 2nd core is barely used even now.

Gillette has already created a version of this overkill in shaving:

Mach 5
Platinum Mach 14
5 Blades!

Re:4X4 is more a marketing ploy than anything else

[Azarael]

[Blockquote]Mounted on each socket will be an AMD dual-core processor (not necessarily an AMD Athlon64 X2, according to an AMD representative)(Emphasis Mine), for a total of four cores.[/blockquote]

The article says otherwise.

Re:4X4 is more a marketing ploy than anything else

[(H)elix1]

Bingo. The price is what is killing AMD.

Not sure if Intel is trying to salt the fields here, but AMD did not drop prices at all until they were forced to over the last couple years. Benchmarks can be somewhat unreliable, but with enough reading you can find how the midrange CPU's compare to each other. Since AMD also dropped the 939 socket, I'm going to look real hard at Intel as I have to update RAM and mainboard the next time I do a major update. Were I buying today, it would be Intel - that has not been the case for me since I replaced a 450 mhz slot 2 xeon with a 700mhz slot A thunderbird. I'm not the type of guy to buy an FX or Extreme! Edition of anything, but when I stack up what kind of bang for the buck I can get between $200-500, AMD has a real problem on their hands. Both the X2 and Core 2 Duo are solid technology, but I will not pay for 'brand'. The AMD kit is going to have to drop a fair bit more to be competitive in the landscape I buy in.

Re:4X4 is more a marketing ploy than anything else

[Surt]

The apps that are going multithreaded now, though, are typically going to 4+ threads to be future ready for the next couple of years. For people who use any of those applications, they'll see an immediate boost per core.

Personally, I'm dying to get my hands on an 8+ core workstation.

Re:4X4 is more a marketing ploy than anything else

[darthdrinker]

Personally, I'm dying to get my hands on an 8+ core workstation.

No flame intended but what are you going to do with those 8 cores? As stated before, there is virtually no consumer software wich takes advantage of 2(or more) cores. As a softwareengineer I can also tell you writing good parralel software is one of themost difficult things. Furthermore it is hard to think of software wich you, the consumer, can use wich can take advantage of parralel programming.

I know what you'r thinking: one core should be enough for everybody. This time.. aah well what's the use....

Re:4X4 is more a marketing ploy than anything else

[Cyno]

Yes, but I can overclock a $300 Intel chip to 3.0 GHz..

I could have 4 AMD cores, sure, but they'll still be 2.0 GHz no matter what I do, and cost me twice the price + extra expensive and buggy SMP motherboard. The Intel will be faster per clock and have a billion extra clock cycles per process/second.

AMD lost me here.. They need to convince me they will continue to lower prices and maintain a competitive product. At this point I'm not convince they can catch up to Intel again.. Intel has greater market share, lower costs and wider margins. AMD had opportunity, but I think they won't have it for long..

Re:4X4 is more a marketing ploy than anything else

[Anonymous Coward]

buggy? you had a good point, you don't have to lie

Re:4X4 is more a marketing ploy than anything else

[1729]

No flame intended but what are you going to do with those 8 cores? As stated before, there is virtually no consumer software wich takes advantage of 2(or more) cores. As a softwareengineer I can also tell you writing good parralel software is one of themost difficult things. Furthermore it is hard to think of software wich you, the consumer, can use wich can take advantage of parralel programming.

There's plenty of multithreaded software out there that scales naturally to 2 (or 8) processors. Also, you don't need multi-threaded applications to take advantage of multiple processors. Personally, I use an 8 core Opteron box for compiling, and instead of waiting an hour or two for a build to finish on a dual-Xeon box (or even longer on any of my single CPU boxes), I have a complete build in 5-10 minutes.

Re:4X4 is more a marketing ploy than anything else

[TTK Ciar]

I am a software engineer working at The Internet Archive, and I write parallel software every day (sometimes with PVM for "real" applications, but more often as throwaway perl slammed out on the command line, using open3() to open several simultaneous subprocesses, sometimes fed data by the parent but more often each reading from a different data file). Much of what I do is "trivially parallelizable", meaning it's pretty easy to make scale across multiple processors or machines. It is my impression that most real-life problems seen by most businesses are trivially parallelizable, with the rare exceptions hogging all the attention by dint of being more interesting.

My workstation is a single-processor machine, but I have at my exclusive disposal a dual-xeon machine and two AMD dual-core machines. I'm always scp'ing my work up to them from my workstation so I can take advantage of their multi-process goodness. (Developing while ssh'd into those machines is usually not a good idea, since the network likes to go down or slow down a lot between Archive HQ and our datacenters, and our HQ firewall blocks PVM so I can't just make my workstation the PVM master node with the other three machines slaves.)

When I read this article, my initial reaction was "Enthusiasts, hell! I want as many of these as I can get for servers!" (assuming this 4x4 product is significantly cheaper than current dual-opteron products -- we're a non-profit, without a lot to spend on hardware, and we're always running on the edge of starvation. But maybe that's a bad assumption and these will be prohibitively pricey).

If someone offered me a 4x4 or 8x8 for my desktop, though, I'd accept it gladly, and make good use of it, parsing/analyzing Archive metadata, processing multiple simultaneous http streams (we use a lot of http-rpc here, and xml data representation which means each http-rpc stream can suck down a lot of processing power), md5'ing multiple files in parallel, and the like. I'd probably also make more extensive use of bzip2 than I do currently :-)

My datasets commonly consist of hundreds or thousands of files, each of which can be processed in parallel, so I can keep throwing cores at the problem with near-linear scalability until I grind against disk or bus bandwidth limits (at which point the data needs to start out distributed in order to keep scaling).

Just my $0.02

-- TTK

Re:4X4 is more a marketing ploy than anything else

[Cyno]

that has not been the case for me since I replaced a 450 mhz slot 2 xeon with a 700mhz slot A thunderbird.

Same here. Oh, well, what's one less AMD fanatic? They'll never miss us.

Re:4X4 is more a marketing ploy than anything else

[Xichekolas]

I don't know about you, but I can't remember the last time I was only running one peice of software at a time.

I just got a new Turion X2 laptop, and I can say that it handles running Eclipse, MySQL, Apache, GAIM, my 40 Firefox tabs, mp3 player, and a bunch of other random shit much better than my other (single core) machine. How much faster is it? Heck if I know. I'm not really a benchmark fanboy... but dual/multi-core systems don't require awesomely parallel software. They just require some degree of multitasking and a somewhat recent OS.

Re:4X4 is more a marketing ploy than anything else

[Jeremi]

I suspect that many people (including myself!) like multicore computers for the same reason other people like SUVs... having all that extra power to play with is just cool, even if you won't actually use it much. Besides, I might someday write a parallel program that uses all 8 cores (just like the SUV owner might someday take his toy offroad ;^))

Re:4X4 is more a marketing ploy than anything else

[Surt]

I'm going to run video editing, 3d rendering, and various custom software all of which scales to at least 4 cores, often at the same time (even with a quad processor box with all 4 procs maxed, i'm too often waiting on hour long renders). I have a lot of software that will likely scale just fine to 32+ procs (I'd love to find out .. the current biggest box i've been able to run on is only 4x).

Writing parallel software is not that hard. By the time you've written a couple of enterprise applications, you know the basics, because there your software has to sync across multiple boxes. Syncing on one box is all the easier. Parallel software is really close to trivial, you need only know how to a) synchronize and b) partition workloads. A is very easy, and B is only hard some of the time (for many media tasks, B is utterly trivial).

Re:4X4 is more a marketing ploy than anything else

[FuturePastNow]

You've touched on exactly what makes this so hard to understand. AMD used to be the ultimate value buy. Two of my college roommates and I had, between us, an Athlon XP 1800+, 2100+, and 2500+. I've also got a Sempron 2200+ machine. All four of those computers were relatively cheap and offered a lot of bang for the buck. But the Athlon 64 and X2, while fast, are way too expensive for any of us to consider, and this coming price cut by AMD is too little, too late.

I have an Intel machine now, and both of my old roommates are switching.

Re:4X4 is more a marketing ploy than anything else

[TikiTDO]

If you have the money for an 8 core Opteron box then surely you can just buy 10 cheap dual core boxes, make a cluster and throw on distcc for even faster compilation.

Re:4X4 is more a marketing ploy than anything else

[MarcoAtWork]

definitely, you'd think that at least they'd have

GFX thread (self explanatory)
Game logic thread (might not benefit hugely)
Physics thread (would be amazing to be able to offline this in terms of the effects you could create)
Sound thread (imagine the soundscapes you could create with a dedicated CPU)
Networking thread (the more processing power -> the better curve fitting you can use when doing predictions)

I do think with 4 cores and a game designed to take advantage of them you'll see vast improvements in the user experience, without having to bother with silliness like specialized "physics coprocessors" etc. etc.

It's great to see the industry moving towards multiple CPUs, hopefully this will not be just a fad and 4-way dual-socket systems will become standard within a year or two, relegating single-sockets to today's celerons, and having quad-socket 8-way systems for the high end market.

Re:4X4 is more a marketing ploy than anything else

[darthdrinker]

My point is this, john doe will not be able to use the power of multiple cores/proc's. Nor is there software wich Jon Doe uses wich can take advantage of this. Optimizing John Doe software to take advantage of more cores is nearly impossible. Try optimizing word or powerpoint or IE... While you might use it everyday, John Doe does not.

Re:4X4 is more a marketing ploy than anything else

[networkBoy]

Nevermind that to me 4x4 implies four chips each with four cores.
likewise 8x8 is 8 chips of 8 cores. I don't know WTF the marketing dept was thinking. They are really selling a 2x2 and a 2x4.
-nB

Re:4X4 is more a marketing ploy than anything else

[1729]

If you have the money for an 8 core Opteron box then surely you can just buy 10 cheap dual core boxes, make a cluster and throw on distcc for even faster compilation.

Well, that doesn't scale very well. The Opteron box I use is actually just one node of a cluster.

Re:4X4 is more a marketing ploy than anything else

[1729]

Optimizing John Doe software to take advantage of more cores is nearly impossible.

That's absolutely absurd. Many consumer apps are already multi-threaded, and the relevant OS kernels are definitely multi-threaded. Many of the problems these consumer apps handle are embarassingly parallel. In addition, there are libraries (OpenMP, Java threads, etc.) that make it very easy to parallelize these types of code.

Re:4X4 is more a marketing ploy than anything else

[Anonymous Coward]

Because buying a 2x4 is something that you do at the lumber yard.

Re:4X4 is more a marketing ploy than anything else

[Cyno]

I have an AMD MP tyan system, unstable as hell.

I don't just "believe" new technology is bug free until I verify it. Unfortunately Opterons have been out of my price range. I suspect there will be bugs in their cheap multi-cpu consumer class boards. Want to bet on it?

Re:4X4 is more a marketing ploy than anything else

[TTK Ciar]

Optimizing John Doe software to take advantage of more cores is nearly impossible.

Well, putting aside for the moment that AMD is marketing this 4x4 towards "enthusiasts" and not "John Doe", I disagree with your statement. Does John Doe use Google Video or similar? Multiple cores may be used to decode that video faster, for smoother playback. Does John Doe play games? Many games can be made to take advantage of multiple cores. Does John Doe browse sites which deploy Macromedia Flash based services? Each shockwave application (many of which will eat CPU whether their window or tab is open or not!) can run in its own thread, at the very least. Does John Doe search his email folders? Searches on multiple distinct datasets are really trivially parallelizable. Does John Doe browse large webpages? Many webservers are configured to transmit content in compressed form (the page you are reading right now was transmitted in gzipped format by Slashdot's server) to ease their network load, and the browser could pretty easily stick the decompressor in a different thread.

I could go on for a while. In theory, all of these (and more) can be re-implemented to take good advantage of multi-core technologies. It "just" takes more code-monkeys mashing their fingers down on keyboards to write the necessary code. Unfortunately, in practice, getting code-monkey resources allocated to such tasks doesn't happen often enough. Perhaps in that respect getting John Doe software into a multithreaded state is "nearly impossible", but that could change as soon as John Doe starts buying multicore hardware (which he might, if at least one of his favorite programs can take advantage of it).

-- TTK

Re:4X4 is more a marketing ploy than anything else

[jez9999]

AMD lost me here.. They need to convince me they will continue to lower prices and maintain a competitive product. At this point I'm not convince they can catch up to Intel again.. Intel has greater market share, lower costs and wider margins. AMD had opportunity, but I think they won't have it for long..

So you'll soon have the problem of there being one, uno, major PC CPU manufacturer in the market, with a barrier to entry so high it makes the Great Wall of China look like a garden fence. Much better.

Re:4X4 is more a marketing ploy than anything else

[Cyno]

Yep.

Its called capitalism. This is what happens naturally if we do nothing to slow it down or stop it. But it will probably be your kids problem to deal with, not mine.. ;)

Re:4X4 is more a marketing ploy than anything else

[Bill_the_Engineer]

You forgot about priority inversions, and deadlocks.

Re:4X4 is more a marketing ploy than anything else

[GooberToo]

Those types of issues are normally motherboard specific. In other words, you bought a crappy board...that has nothing to do with AMD. And for the record, based on historic record, you won't have trouble finding crappy SMP boards for Intel chips either. Lastly, depending on the OS and/or the kernel you run, you might have a finger to point there too. Long story short, saying AMD = buggy SMP is a total fabrication, misrepresentation of the facts, or simple ignorance of the facts.

I'm not saying you have SMP issues with your setup. I am saying, it's very, very, very unlikely that it has anything to do with AMD.

Re:4X4 is more a marketing ploy than anything else

[sbrown123]

You are taking a lot of assumptions. First, you are thinking 4x4 will only contain the most current AMD chip technology as of today (FX-62). That's possibly a bad assumption since the 4x4 (1) hasn't been released and (2) we have no idea what chips will be in it and (3) we have no idea about what AMD will charge for this in the future. I could make assumptions that the 4x4 will contain chips that will be 4 times faster than the fastest Conroe. Think I'm wrong? Prove I'm wrong. You can't, just as much as I can't prove I'm right since it's all assumptions and wild guesses. Thats the future. Live with today. Conroe chips are faster and cheaper. If your getting a new computer you might want to get a Conroe instead of an AMD-FX62. But when the 4x4 comes out you might want to check again and see who the price and performance leader still is.

Re:4X4 is more a marketing ploy than anything else

[sbrown123]

Ofcourse there may be bugs in the 4x4. But who is to say there are no bugs in the Conroes? I have an old Athlon XP on my puter right now, and in its day it was a leader in price for performance. Conroe looks like the leader of the moment, so if I were to buy a new computer I would look at it. But who knows what the future holds, or what the 4x4 will be like?

Re:4X4 is more a marketing ploy than anything else

[Surt]

Deadlocks are a symptom of not understanding sync. Priority inversion isn't one i've ever hit in actual practice, possibly because the multithreading i'm interested in doesn't have priority level issues.

Re:4X4 is more a marketing ploy than anything else

[DrMrLordX]

Actually, quite a bit is known about the 4x4 intitiative. It's sad that so many people posting about it here seem to know very little. The gist is:

1). It will only be able to use FX processors.
2). AMD will have, at best, the FX-64 out by the end of '06

The FX-64 is just another K8 with an unlocked multiplier and 1 meg of l2 cache clocked at 3 ghz. It'll cost over $1k per processor.

Long story short, 4x4 will be expensive and not perform terribly well. You'd be better off getting a s940 2p Opteron 285 system or something like that, especially after Socket F launches, unless the lack of "enthusiast" options on most s940 boards really gets on your nerves.

Re:4X4 is more a marketing ploy than anything else

[RightSaidFred99]

Oh Gosh, this sounds so plausible! I mean, you write parallel software every day and you work for something called the "internet archive". Seriously, does it get more plausible than that?!

Okay, that attempt at pleasentry done - do you really think anybody buys that? I mean I'm sure your average sickeningly contrarian SlashDot Nerd ate that shit up, but personally I find your trenchent analysis to be nothing less than PEDANTIC.

Give me reproducible benchmarks and I may believe this jibber jabber, but until then you sound like you're just making shit up. Strike that. You ARE making shit up.

Re:4X4 is more a marketing ploy than anything else

[Anonymous Coward]

You're a dick.

Re:4X4 is more a marketing ploy than anything else

[toddestan]

I don't see much change in the lower end of the CPU line ups. The last time I checked, AMD's Sempron line offered great bang for the buck on the lower end, especially compared to Intel's outdated P4/Celeron chips in the same price range. AMD is still the ultimate value buy.

What has changed is that for a bit more money, you can get a dual core chip. A lot of people who are now buying new systems or upgrading at the moment feel like they have to have a dual core system (I can't blame them for this). So they only look at the dual core chips, and ignoring chips like the Sempron 64. This is where Intel is beating out AMD. I'm seeing a lot of people who used to always buy in the "best value" category of chips are now moving up the product lines to get a dual core system. I predict that given a few months or a year or so, dual core processors will have trickled down into the value chip range, and some people will shift back over to AMD for their next system.

Re:4X4 is more a marketing ploy than anything else

[Jagen]

http://www.archive.org/ School yourself, n00b.

Re:4X4 is more a marketing ploy than anything else

[ExProg]

Near future price reductions by AMD have to be taken into account. X2 CPUs will get a (up to)50% price cut around the release date of Intel's new processors (24-JUL-20060. Things will be more interesting then...

Re:4X4 is more a marketing ploy than anything else

[Cyno]

I imagine the 4x4 will be about 50% faster than the faster Intel desktop system. But its still locked down. CPU clock multiplier lock and CPU/cHT lock to prevent more than 2 CPUs. Its stupid to do this intentionally, when we're capable of so much more.

What incentive does it give me to buy a system that's twice as fast for twice the price?

Re:4X4 is more a marketing ploy than anything else

[Anonymous Coward]

Well you probably won't see this since it is kind of a late reply, but just in case you could still make your machine a PVM master, just use ssh port forwarding to bypass the port block.

Re:4X4 is more a marketing ploy than anything else

[TTK Ciar]

Well you probably won't see this since it is kind of a late reply,

Most /. articles aren't very interesting, so I go back to older ones and pick up late replies. :-)

but just in case you could still make your machine a PVM master, just use ssh port forwarding to bypass the port block.

(embarrassed silence) . . . you're absolutely right. Why the hell didn't I think of that? Thank you! I think I will.

-- TTK

Re:4X4 is more a marketing ploy than anything else

[TTK Ciar]

Normally I don't feed the trolls, but taken at face value your request is reasonable, so ..

http://www.petabox.org/ (note contributors list on left)

http://www.ciar.org/ttk/images/petabox/

http://workstation20.archive.org/ttk/tools/manifes ting

http://workstation20.archive.org/ttk/tools/researc h

http://workstation20.archive.org/ttk/wp/

http://workstation20.archive.org/ttk/wp/software/2 006-04-11/

My Resume

-- TTK

Re:4X4 is more a marketing ploy than anything else

[rapier1]

Also, if your servers are more than a 64k BDP away then you probbaly want to install the HPN-SSH patch from PSC. This path overcomes one of the perfromance bottlenecks in SSH (limited window size). It also give syou the option of use the NONE cipher - the neat thing is that it still uses encryption for authentication. Unfortunately, it only allows the use of the NONE cipher switching for transfering files and not shell sessions. http://www.psc.edu/networking/projects/hpn-ssh

Re:4X4 is more a marketing ploy than anything else

[manthrax3]

You are a moron. It's capitalism that gave us the great FX platform and it's capitalism that's giving us the Core Duo. It will be capitalism that gives us the next round of good processors.

To review: you're a moron.

Part of the vicious cycle in Tech

[Anonymous Coward]

Part of the vicious cycle in Tech, darwin^2.

Re:Part of the vicious cycle in Tech

[andrewman327]

I don't see why we're talking about flattery and being slapped in the face. It seems that AMD and Intel are competing more directly than in the past, which could ultimately be good news for consumers. By reducing power (/. reported on congress' urge to reduce power consumption earlier) these chips save money and run very quickly. Now that both parties are fighting for efficiency and other similar things, they will have to pull out some amazing science to directly compete instead of simply bosting that their paradigm is superior.

Re:Part of the vicious cycle in Tech

[Uryene]

...A cycle I learned my lesson about many, many moons ago.

At home, I keep a $640 check I wrote back in 1990 for a 486 CPU.
It's framed and visible on top of a bookcase to serve as a reminder.

At the time, I thought it was a great deal; screaming processors were
never going to get much cheaper than that!

These days, last years tech (or even two years ago tech) is usually
MORE than sufficient. Except for games, which always seem to
need NEXT years processor in order to be playable... ;-)

Re:Part of the vicious cycle in Tech

[Punko]

Pop paid almost C$3,000 for a wide format (17") dot matrix printer to hook up to our TRS-80 Model III in 1981. I spent a bunch of time making really freakin' huge banners on that continuous feed computer paper. Adjusted for inflation, that's C$6,755 for a dot matrix printer! I'm fairly certain I could get a decent printer now for that much . . .

Re:Part of the vicious cycle in Tech

[kisrael]

Heheh, yeah. I got thwomped by freshman in their shiny Pentium's vs my trust 486/16 on Duke Nukem's Water level (over the brand new dorm ethernet!) when the FPS became 1 , and I've been a console gamer ever since.

I just bought one of those amazing cute HP slimlines for like $550. It's probably the first PC I owned that does well with "Liberty City", and does everything else amazing well. If you don't need a PC for games, you can get by with some very old hardware indeed. (And yes, I do recognize how much better PCs do in certain genres, and how much I suck for not like WASD/mouse play)

Re:Part of the vicious cycle in Tech

[shelterpaw]

Try running a slew of virtual software synthesizers and effects and you'll be happy to see faster processors. I know I couldn't be happier and I can't wait to see which one Apples puts in it's next machine.

Re:Part of the vicious cycle in Tech

[TheSalzar]

who cares what apple chooses, think for your self and buy one of the many choices on the x86 market, and choose a good mobo. Maybe if apple was selling some non comman cpu arch (for example PowerPC), i could justify buying one, but now its the same insides. Used a mac since I was a child until they decided to kill off the clones.

Re:Part of the vicious cycle in Tech

[RightSaidFred99]

Gee.. This post looks _awfully_ familiar. Methinks it is either a dupe or a copy. If you posted it the first time, then no foul - otherwise...

Re:Part of the vicious cycle in Tech

[damacus]

Frontloading the cost of technology isn't always a bad thing. A 486SX/25 powered the first computer owned in my family: a $1500 Gateway2000 box with 4MB RAM, 170MB HDD and 1MB ISA video card. It was purchased in December 93 for Christmas.

I imagine if you bought a 486 in 1990, you got years of service out of it, and if nothing else, you got to have a killer machine for years before other people started catching up, albeit for less.

For some people, that more than warrants the extra cost. It's capitalism at work.

Heh...

[Anonymous Coward]

If Intel had tried this a year ago with their inferior P4 processors, AMD fanboys would have complained that they were being stupid and basically saying they needed two chips to beat AMD's single chip.

Now, I'm sure that AMD is "innovative" by introducing this platform. Genius, I tell you!

Re:Heh...

[Anonymous Coward]

Come on, it's FANBOI - not fanboy...

Re:Heh...

[P3NIS_CLEAVER]

You could tell that Intel was nearing the end of the processors cycle when they put ridiculous amounts of cache on the processor. I guess doubling up the cpu will be AMDs.

Re:Heh...

[deficite]

It's funny how many sensible arguments get modded as flamebait simply because people don't want to face the truth. I think the parent should've actually been modded up instead of down

umm? comparison to Intel please...

[sofar]

but does it perform better than core 2 duo? I fail to see any performance test between them, and it's also AMD having the bigger market share right now, not intel. Seems like a lot of AMD FUD nowadays... AMD is no longer the underdog here.

Re:umm? comparison to Intel please...

[treeves]

Not so much, but AMD still doesn't have larger market share than Intel - not nearly. It's just that the gap is shrinking.

Re:umm? comparison to Intel please...

[qbwiz]

I think the grandparent was thinking about retail market share, where AMD has indeed passed Intel. Overall market share is a different story....

Re:umm? comparison to Intel please...

[treeves]

Even that's not quite true. . .Intel takes back lead in US retail | Tech News on ZDNet

Re:umm? comparison to Intel please...

[qbwiz]

Well, I did say passed. I just never mentioned whether they were still ahead.

Re:umm? comparison to Intel please...

[Anonymous Coward]

You said "has passed" implying that they still are past. You could have said passed or had passed and not implied that.

Re:umm? comparison to Intel please...

[Mulielo]

So what does that mean, everyone's gotta jump ship and start rooting for Intel? Haha, but I agree, FUD is FUD no matter who spits it out, and you can't compare two processers without performance results.

Re:umm? comparison to Intel please...

[Anonymous Coward]

See? English isn't complicated. It totally makes sense! Why can't immigrants just learn it?

Counterstrike

[m_chan]

I tried to read this article and all I could think of was that AMD is mad that strafe jumping got patched and that Intel learned how to bunny hop. I'm hung over. Need sleep.

Re:Counterstrike

[Trouvist]

Strafe jumping is bunny hopping...

Re:Counterstrike

[Anonymous Coward]

Close, but not quite.

Re:Counterstrike

[alexhs]

Yeah, we soon will get two incompatible MS-Windows systems :

1. Intel Pentium / nVidia Geforce / iD Quake
   
2. AMD Athlon / ATI Radeon / Valve Half-Life
   

I guess I need to sleep too ;)

Performance improvement?

[Atroxodisse]

Not if you ask hard OCP http://enthusiast.hardocp.com/article.html?art=MTE wOCwxLCxobmV3cw==/

Re:Performance improvement?

[Iamthefallen]

How big an issue was the GPU in that test? None, or some? When they dropped a quality level down, the gap increased.

Another OCP comparison, without GPU limitations.

Re:Performance improvement?

[InsaneGeek]

Only problem is that article basically shows that the CPU isn't the bottle neck in what they were doing. It's like saying there isn't a performance improvement from an original K6 to an opteron and to prove it I'm going to time how long it takes to transfer a 100mb file over a modem with the different CPU's and prove to you how the performance difference is minimal.

Re:Performance improvement?

[Surt]

But then, if you trusted HardOCP, you'd have to be stupid, so why should we care if you're misinformed?

Re:Performance improvement?

[Anonymous Coward]

Bah... Point us to a better quality review site. Tom's Hardware just doesn't cut it for most enthusiasts.

Stop acting like a clownshoe. Its not like we're talking about the Inquirer or something.

Re:Performance improvement?

[Surt]

I usually see good quality out of anandtech. They seem to have the least average daily bias, and their selection of benchmarks is usually sensical.

Re:Performance improvement?

[Anonymous Coward]

That is kindof their point. Who plays games at 800x600? If you already have a very good graphics card and decent processor, don't bother upgrading. This is the message I took from the article.

Re:Performance improvement?

[tls2000]

But then, if you trusted HardOCP, you'd have to be stupid, so why should we care if you're misinformed?

Can you qualify that? HardOCP does tests based on real world gaming, and they don't like using artificial benchmarks. Yes, Kyle Bennet can be opinionated, but regardless of that, his site does write decent articles and reviews on gaming performance.

Re:Performance improvement?

[courtarro]

Wow, I've never read a more biased article. I have a feeling [H] has a deep love for AMD and they're just convincing themselves that the AMD procs they just bought are still awesome. When it comes to graphics-intensive games, we all know that processor speed has comparatively little to do with performance. However, other tests of raw processing speed have shown significant improvements on the Intels, and they make no conciliatory mention of that in the conclusion. All they did was say, in 11 pages, that the CPU doesn't affect a certain collection of games.

Re:Performance improvement?

[RightSaidFred99]

It's just hilarious that you mention that article. HILAROUS. Seriously. "Next, on HardOCP: Buying one of those wasteful $400 video cards doesn't help you in video encoding to DivX. News at 11!" I love you man! It's hilarious that you would reference that ridiculous "review".

Performance number?

[Soybean47]

FTFA:

AMD also plans to push a sort of "performance number" into the market to redefine how consumers should think about power, they said.

Doesn't AMD already label their processors with a relatively meaningless number designed to... say... redefine how consumers think about processor speed?

Was that a highly effective marketing technique? I mean, I guess it did get people to think about speed, and it helped convince many people that GHz isn't the be-all and end-all of processor comparison. But at some point won't people just be annoyed by the mess of pretend numbers AMD is throwing around to "make us think?"

Re:Performance number?

[TheSunborn]

I think they mean power, as in Watt usage. Currently they just rate the maximum usage which is not really that usefull.

Re:Performance number?

[Soybean47]

I got that. It just seems kind of odd to me that they're trying to convince people to measure speed in magic AMD numbers instead of GHz, and now apparently power consumption in some different magic AMD numbers instead of Watts. It's good to get people thinking, but after a while it just seems like you're trying to fudge things to make yourself look better.

Re:Performance number?

[Siward]

Frankly, I tend to think of GHz as being meaningless. If your architecture is powerful enough that you don't need to push big numbers at people, why bother? Intel's high-range of GHz values makes me think of car manufacturers who advertise huge horsepower figures, when the true measure of a car's ability to accelerate is a lot more subtle than that.

AMD has been pushing processor architecture for a while, which is something that a handful of people understand. Do you really think the average computer expert is going to understand architecture improvements? Not to speak of mom and pop buying a new computer -- they're going to look at rating.

The bottom line really is that you shouldn't be paying attention to numbers at all. Benchmarks (although some can be skewed) are the true test of a processor's power. Do you think they want to tempt fate and test the hardware review community? Beyond that, do you think AMD really wants to overinflate their processor value so they can piss off the enthusiast crowd they pays attention to these things? Aside from poor business practice, it would be dumb for them to just throw big numbers around if those numbers didn't correspond to some well-understood performance value. (I realize I just said that GHz is meaningless, but it is the historical metric.)

Re:Performance number?

[Martin Blank]

I find normal consumers have a great deal of difficulty figuring out what the performance level of a given Intel chip is compared to others given how their model numbers jump around. AMD's numbers are more straightforward, though they still don't always tell the right story.

Re:Performance number?

[Afrosheen]

It was a brilliant move by AMD marketing to get people to think outside the GHz box because it's how much work a processor can do per cycle that counts, not how fast its internal clock rate is. They proved, with the Athlon64 chips, that a lower-clocked chip can perform as well as, or outperform, a chip with a higher clock rate.

  The old timers, such as yourself, that think GHz ratings mean anything anymore are just wandering around blind. Intel proved that themselves with the switch from Pentium 4 to Pentium M for their mobile chipsets. All of a sudden you have a 1.6ghz chip that outperforms the older P4 2.xghz chip. So yeah, it seems like magic to the unitiated, but then again, everything technical does.

Re:Performance number?

[Anonymous Coward]

it's how much work a processor can do per cycle that counts

No, it's how much work a processor can do, period, that counts. A fast, narrow, deep architecture is not immoral or inferior compared to a slow, wide, shallow one. They're just different design choices representing different tradeoffs. At the end of the day, all that matters is how much the processor can do for you, and what you have to pay to get it to do that.

It doesn't matter whether a processor can do 2 units of work per cycle, but can only be clocked to 1 MHz because the design is complex, or whether it does 1 unit of work per cycle, but can be clocked to 2 MHz. Performance here is the same, and one strategy "counts" just as much as the other.

It would be really easy to design a processor that maxes out on the "work per cycle" metric, yet has dismal overall performance. Say, for example, that you gang up 16,384 12 MHz 8051-based 8-bit microcontrollers (about the same gate count as a Pentium-D). Lots of work every clock, but hard to get much useful throughput. Work-per-cycle isn't important; total work is.

"What you have to pay" can refer to dollars. But in some applications power (watts) are important. In others, size and heat matter. So there's no even a single metric for "good", much less a single proper design solution for all cases.

Performance per watt has become a more important metric thanks largely to cell phones and PDAs, plus a shift toward laptops and away from desktops. In all these applications, limited battery life is a major problem for the device, and absolute top end performance is not needed. Hence, designs oriented toward performance per watt rather than performance per dollar or just plain performance. This isn't due to inherent moral or engineering superiority of such designs, just marketplace demand.

Fanboyism...

[TripHammer]

is wasteful. I'm glad to see Intel back in the mix with some good offerings. I think those of us whom are fickle come out on top: you buy what makes sense. Fanboys step back!

Re:Fanboyism...

[Anonymous Coward]

"I'm glad to see Intel back in the mix with some good offerings."

What about Intel's love for DRM? I haven't been following CPUs much ever since I stopped playing games for some time now. But does anybody know if AMD is going to/is doing support for DRM in their stuff?

Re:Fanboyism...

[Anonymous Coward]

Well my last proc was an AMD. Why? It was the old AMD socket 657 (not the new 989) or whatever. Im getting mixed up.. so I got it at a price that simply was better than anything I could get from Intel at that time. I always said that Intels Centrino procs were the best tech around and now they've advanced those into a new generation designed for the desktop in the form of core duo.

Right now I'd by an Intel proc but its not that big a deal. We're still pretty much at the end of the semiconductor proc envlope thats why theres all this multicore nonsense. AMD can still compete with Intel far into the foreseeable future (however far that is ..)!

It's about Goddamn time!

[Anonymous Coward]

AMD Launches Counterstrike Against Core 2 Duo

We need corporate wars to thin things out. Fuck'n A! The Governments of the World are just too incompetent! It's obvious that the MBAs of the World need to unite and show these Bozoes how to fuck'n do it!

Yes siree, profit above all else! Fuck these Goddman bald monkeys!

Hey, I'm not done yet! Put those fucking jackets away!!! Hey!!! Mmmmmmmm!mM!M!M!M

Put in straight jacket and sent to a Ph.D business program.

Re:It's about Goddamn time!

[ak3ldama]

AMD Launches Counterstrike Against Core 2 Duo We need corporate wars to thin things out. Fuck'n A! The Governments of the World are just too incompetent! It's obvious that the MBAs of the World need to unite and show these Bozoes how to fuck'n do it! Yes siree, profit above all else! Fuck these Goddman bald monkeys! Hey, I'm not done yet! Put those fucking jackets away!!! Hey!!! Mmmmmmmm!mM!M!M!M Put in straight jacket and sent to a Ph.D business program.

That is funny, not flamebait. Mod Parent up. I would but I have no points! Is there no justice in the world?

In other news...

[Anonymous Coward]

The latest technology from XYZ corp beats 3 year old technology from ZYX corp.

-AC

Re:In other news...

[Anonymous Coward]

No it doesn't. ZYX's awesome Widget X4 will kick the pants off of XYZ's 4X Thingamajig. You just wait and see.

"well.. my dad can beat up your dad!"

[Churla]

This smacks to me of AMD realizing Intel had actually landed a well placed shot into thier gut and needing a fast "get positive attention back on up" spin.

So we'll have to buy TWO processors to compete with what Intel is doing with one? If they're aiming for the Enthusiast market they have to remember that "enthusiasts" have price constraints (usually referred to as "wife")

I could be wrong. But I really don't think I am.

Re:"well.. my dad can beat up your dad!"

[Anne_Nonymous]

My processor goes to eleven.

Re:"well.. my dad can beat up your dad!"

[bigmike_f]

You're mistaken... AMD solution is 4x4... which means with two chips you get four processors.

Re:"well.. my dad can beat up your dad!"

[uop]

Well, having the strongest offering usually does count for something (especially if you ask the marketeers).
The 4x4 initiative basically looks like DP for the desktop, which Intel offers as well (although Xeon only).

imho, the really interesting thing about 4x4 is the possibility of plugging in a coprocessor in the future.
For example, you may settle for a single Athlon64 X2 in a 4x4 board for now, and add a physics/video/dsp/whatever coprocessor in the future.
That's wild speculation, of course, but it does make the 4x4 setup intriguing as a future-proof product.

--

Re:"well.. my dad can beat up your dad!"

[Zoop]

price constraints (usually referred to as "wife")

Oh, is that what we're calling the dot com crash now? Clever. Explains the lower salaries since the dot com crash and gets your mom off your back.

Now, do we call the computers "grandkids"?

Re:"well.. my dad can beat up your dad!"

[Moofie]

Doesn't that make it a 2x2? I'm sorry, I don't do marketing-flack math.

Re:"well.. my dad can beat up your dad!"

[LurkerXXX]

My guess is it's far more likely just to be a re-branding of the 2x Opteron series... Marketed to enthusiests.

Re:"well.. my dad can beat up your dad!"

[Durandal64]

4x4 implies a matrix of 16 elements, not 4. Christ, can't marketing guys multiply?

Re:"well.. my dad can beat up your dad!"

[Anonymous Coward]

The Intel chip also has 2 cores so the argument is still valid. AMD needs twice as many to beat Intel.

Re:"well.. my dad can beat up your dad!"

[LordKazan]

If they're aiming for the Enthusiast market they have to remember that "enthusiasts" have price constraints (usually referred to as "wife")

excuse me? WTF are you smoking. The people I think of buying the hottest newest CPUs with multicores and multiple CPUs in the enthusiast (read: gamers) market is people who buy more hardware for ePenis only.

These people don't have wives! :D

Re:"well.. my dad can beat up your dad!"

[Cyno]

I agree, I don't see anyone lining up to sell cHT chips..

physx chips are already on graphics and PCI cards..

AMD NEEDS the extra CPU to keep up with Intel, AND the extra money from enthusiasts.. but somehow I bet this will backfire.

Re:"well.. my dad can beat up your dad!"

[Thundersnatch]

If they're aiming for the Enthusiast market they have to remember that "enthusiasts" have price constraints (usually referred to as "wife")

Judging from most of the posts I read here, I think "mommy" is more likely.

Re:"well.. my dad can beat up your dad!"

[Anonymous Coward]

This smacks to me of AMD realizing Intel had actually landed a well placed shot into thier gut and needing a fast "get positive attention back on up" spin. So we'll have to buy TWO processors to compete with what Intel is doing with one? If they're aiming for the Enthusiast market they have to remember that "enthusiasts" have price constraints (usually referred to as "wife") I could be wrong. But I really don't think I am.

We will find out if you are wrong by asking your constraint!

Re:"well.. my dad can beat up your dad!"

[twitchingbug]

I don't think you can ever future proof computer products.

Anyhow, I doubt any company would make such a co-processor for an AM2 socket. There just wouldn't be a reason to do it - not a large enough market. Just use a PCI Express slot. There are ton more computers out there with those than an extra AM2 socket.

-Don

Re:"well.. my dad can beat up your dad!"

[Anonymous Coward]

My processor goes to eleven.

Oh yeah? Well we've got Armadillos in our trousers. It's really quite frightening.

I for one...

[rowama]

am happy that I finally know what will be in my computer 5 years from now, when I swap out my pentium III based system. Us poor folk at least get to enjoy the anticipation longer.

Re:I for one...

[doti]

Well, a pIII can play movies and run quake3.
That's the two most demanding uses for a computer.
The rest is futile.

Re:I for one...

[Donniedarkness]

You forgot the other one:

It can, in fact, run Linux.

Re:I for one...

[fotbr]

Not poor, but still happily running my dual P-III 1Ghz setup I built 5 years ago. I've upgraded from the original GeForce Pro to an ATI Radeon 9800Pro last year, swapped out the CD burner for a DVD Burner the year before that, and added two 400GB drives 6 months ago to complement the two 80GB drives that I originally had. The thing that'd really help me would be more RAM, but from day one it was maxed out with 2GB.

Until I meet anything it can't do that I really want to do, I don't see the need to replace the machine. Unfortunetly, it is about at the end of its upgradeability -- new graphics cards will require PCIe, which means new MB, and therefore new CPU(s). More RAM would also require new MB, etc.

Maybe I'm frugal, maybe I just don't see the need to always have "the latest and greatest" but I'll stick with my strategy of building a beast of a machine every few years, but not throwing much money at it after its built.

Re:I for one...

[sootman]

I hope you get some 'insightful' points for that in addition to the one 'funny' point you've already got. I'd like to see more computers moving towards small/efficient/quiet, like the Mac Mini. I just got a Dell Dimension 3100 and it's huge and heavy for what it does. It might be 2x or 4x faster than my G4 Mini, but it's also got to be at least 30x larger. Hell, I could put the Mini into one of the Dell's drive bays and tuck the power supply inside as well. And maybe add a tiny toaster-oven door and shelf to keep my meatball sandwich warm. Wait, what was my point again?

Re:I for one...

[idonthack]

Why would you need a sandwich-warmer? They're warm right after you make them.

Re:I for one...

[Tim Doran]

The best part about a meatball sandwich in your drive bay is when the grease makes your case transparent.

Re:I for one...

[Silverlancer]

A P3 cannot play DVD-resolution H.264 video, which means that 70% of my 250GB D drive would be completely useless.

Re:I for one...

[WuphonsReach]

Maybe I'm frugal, maybe I just don't see the need to always have "the latest and greatest" but I'll stick with my strategy of building a beast of a machine every few years, but not throwing much money at it after its built.

System improvements (clock speeds, FSB speeds) have pretty much stagnated over the past 3-4 years. Dual-core is pretty much the only major performance bump to arrive in the past few years. So you shouldn't be surprised that your 2GB of RAM, 1GHz machine still feels adequate for most tasks (except CPU-bottlenecked ones). For a general-purpose machine, that (due to the 2GB of RAM) that could probably be used until 2010-2012.

That extra RAM makes up for a lot of slower CPU speed. The fact that it's a dual-CPU system is just icing on the cake and another big reason why the system still feels responsive. Dual-CPU might even make it feel responsive enough for desktop use until 2015.

A moderately spec'd modern PC can easily last 6-8 years now. Probably 2-4 years longer if it's a dual-core or dual-CPU system. Our power users are getting 2GB dual-core and/or dual-CPU machines and we don't plan on handing them off to other users until 3 years in.

Re:I for one...

[fotbr]

Hmm....No one bothered to tell me that I can't do that with my dual 1ghz P3 setup. Granted, I wouldn't try it on a single cpu 500mhz P3, but to say it cannot when the P3 line spanned a HUGE performance difference between low and high, is sort of rediculous.

The name certainly is

[Lonewolf666]

Because dual-core Opterons have been around for a while. The 2xx series will run in systems with two processor sockets, the 8xx in systems up to 8 sockets. Giving you a maximum of 4 or 16 cores.
I think the new "4x4" processors will essentially be rebranded Opterons from the 2xx series. So if you really want it and are willing to pay up, you can have a "4x4" AMD system now.

Re:The name certainly is

[Cyno]

That's funny.. but true. They're effectively lowering the price of their Opterons..

But since they've just kept us consumers locked out of the SMP Opterons with their socket 939 I feel like they intentionally tried to slow progress towards multi-cpu desktops.. Now they're all for it? Hypocrites!

Forget the small details...

[pieterh]

It does not matter how much these processors cost today, nor whether AMD's 4x4 is real or a maketing ploy.

What matters is that AMD has captured sufficient marketshare over the last years to become a real competitor to Intel. Opterons have become the CPU of choice for large servers, the niche that Itanium was meant to capture.

Now Intel's comeback means we're seeing the start of a new growth of CPU power, this time into multi-core land, a nice solid metric on which to compete. You can fudge the Ghz but you can't really fudge the number of cores. This means we have the perfect conditions for an explosion of growth, until the numbers get into meaningless territory. Within 3-4 years, common desktops will have 8 to 16 cores, and high-end workstations will have 128 or more.

I'm just very glad my company made the move to writing multithreaded code so we can get the best from this new landscape.

Re:Forget the small details...

[joto]

Now Intel's comeback means we're seeing the start of a new growth of CPU power, this time into multi-core land, a nice solid metric on which to compete.

Look, we have hyper-threading/SMT/whatever, so our 2 cores are as good as 4 of your cores. Besides each of our cores are faster than your cores. And nobody needs more than 4 cores anyway. With our supercallifragilisticexpiallidotious memory bandwidth, even our dual core processors will beat your 16-core processor, because memory bandwidth is what really matters. So in order to make consumers think about performance again, and not just the number of cores, we will name our processors with a "virtual core number" that reflect real-world performance better. Thank You!

Re:Forget the small details...

[Splab]

Nice prediction, but you are wrong. While MS sits firmly on desktop market it's going to be hard going beyond 2 cores unless you do your stuff on linux. Granted some of us are going to buy the multicore setups, but thats because we need the power and feel at home with linux.

On top of that, last time I checked the cache wasn't shared, which means we won't get anything really usefull with more cores unless they start doing crossbar switching in the cache (expensive, but damn its nice).

Re:Forget the small details...

[Moofie]

Who's "we"? Got a mouse in your pocket?

"And nobody needs more than 4 cores anyway."

'Cause nobody's ever gotten egg on their face making statements like that.

Re:Forget the small details...

[KingMotley]

Uh, yeah. Tell that to the 32 processor machine sitting next to me that it needs to run linux. It seems to be running windows just fine ;-) Windows was running on 32-way servers before linux was even a thought.

Re:Forget the small details...

[Cyno]

I don't know where you studied comp sci, but NT can easily handle as many cores as you can afford.

And I'm a freakin Linux Zealot! Maybe RTFM next time.

last time I checked the cache wasn't shared

I wonder, when was the last time you checked?

Re:Forget the small details...

[Anonymous Coward]

Huh? Windows NT could only support up to four processors without some serious modification. It took until 2000 for real SMP to be made available. Meanwhile, Linux has supported dozen-way SMP for a decade (and other UNIX-es before that).

Re:Forget the small details...

[Anonymous Coward]

sarchasm

Re:Forget the small details...

[Anonymous Coward]

True, but how many consumers are going to be willing to pay $X,000 for a Windows Server DataCentre to run a 32 CPU box? We're not talking technology, we're talking market differentiation. How long will it take Microsoft (and most other software providers) to adjust their licencing to the new realities of 8+core consumer machines so that you don't have to pay an arm and a leg? Not a problem with Linux, now that it has NUMA support.

Re:Forget the small details...

[adrianmonk]

Now Intel's comeback means we're seeing the start of a new growth of CPU power, this time into multi-core land, a nice solid metric on which to compete. You can fudge the Ghz but you can't really fudge the number of cores.

Sure you can. And as a matter of fact, the trail to do it has already been blazed.

Namely, Sun's Niagara processor (i.e. UltraSPARC T1) has 8 cores on a die and 4-way hyperthreading (to use Intel's terminology) per core. The goal is maximum throughput, and I'm a fan of the processor. But make no mistake: the 8-core chip isn't 8 times as fast as other SPARC single-core chips. Instead, the cores are stripped down. The idea is to make a core that's fast, but not state of the art and super sophisticated, on the theory that the smaller you make each core, the more cores you can fit on a die and that more cores == better for certain workloads.

So each core on a Niagara chip is fairly unsophisticated. I don't know the exact technical details, but we can start with one thing that most cores have that a Niagara core doesn't have: an FPU. Instead, there is an FPU that is shared among all 8 cores. This makes perfect sense considering the goal was to build a chip that can do stuff like web servers and databases with the best throughput possible, and both of those rarely if ever use floating point.

But, I'm sure there are other ways they have economized in the cores as well: modern superscalar processors have all kinds of clever tricks like dispatching operations to more than one ALU at once, branch prediction, deep pipelines, speculative execution, SIMD operations (MMX, SSE, AltiVec), and when making a multi-core system, you could easily leave out some of the more expensive (in terms of die real estate) items and make a simpler core. Maybe you put 2 ALUs instead of 3, or maybe you just reduce the amount of L1 cache a little bit.

Whether any of this would actually happen in competition between AMD and Intel is another question. Somebody will do benchmarks, and the truth will out if they skimp on features per core. But if the competition reaches a point where there is pressure to get 8 or even 16 cores on a chip, I guess it could happen that they might start cutting corners on individual cores to reach the overriding goal of more cores. On the other hand, designing a whole new core is quite an undertaking, so the pressure to do this would have to exist for a while before they'd bother, probably.

Re:Forget the small details...

[Planesdragon]

So in order to make consumers think about performance again, and not just the number of cores, we will name our processors with a "virtual core number" that reflect real-world performance better. Thank You!

FWIW, Vista is going to rate your PC for you--so you finally have a (relatively) nonbiased way to compare actual performance with one number you can read off a box.

Re:Forget the small details...

[Cyno]

I wouldn't mind paying an arm and a leg to get a 8+ CPU 16-way NUMA server for under $10k.

I just don't want $8k to go to CPUs.

I want to choose between spending $1k on cheap CPUs and overclocking the hell out of them for immediate gratification or $5k on premium chips for stability, etc. Artificial price scales do not please me. When the difference between the Opteron and the Athlon X2 is just a jumper on the chip.. I can't support those business practices. Its unethical.

On the other hand if the difference really is stability, I think we should know how unstable the cheap stuff really is.. Or at the very least let's be honest here. What is the real reasoning behind this scheme? Is it honestly justifiable?

And in other news

[plusser]

Mayor of London Ken Livingston introduces a GHz charge on microprocessors used in London as he gets confused by the fact that AMD are to launch 4x4, as he thinks that they take up too much space and are bad for the environment.

Intel leading with heat and watts

[minion]

People are forgetting though that its not just desktop CPUs that AMD took such large pieces of marketshare away from Intel.
 
There have been a few benchmarks (I believe one was on Anandtech's site) that have shown Intel Xeons running in 64bit mode performed slower than the same processor running in 32bit mode. Now, I know, we're talking about copying larger data segments around, because the address space is larger, so a bit of a slowdown in some areas are expected. But when they're talking 5% slower, thats a bit.
 
We replaced 3 Dual Intel Xeon servers (2.8GHz Xeons) with 4G of RAM each, with a single AMD Dual Opteron server, running in 64bit mode for MySQL. This system is immensely faster than the old Xeon systems. MySQL shows upto 23% performance increases in SELECT commands on 64bit vs 32bit on the AMD. On the Intel, it was a performance loss.
 
As far as heat output, the air coming out the back of this server feels cooler, not to mention that it replaced 3 servers with one.
 
People need to focus on the server market, and not the desktop market to see the real king in the (x86) CPU wars. Lets not forget hypertransport, and seperate data paths for memory and IO, whereas the Xeon has a shared 800MHz FSB (now 1066 with the newer rendition).

Re:Intel leading with heat and watts

[Anonymous Coward]

Actually the newest Xeons have a 1333 Mhz effective FSB and 666Mhz FB DDR2 to go along with it.

Re:Intel leading with heat and watts

[Anonymous Coward]

Wow, you clearly don't have much clue about MySQL. During that upgrade, did you upgrade your disks? How did your 3 MySQL severs were configured? Did you have replication going?

This reminds the blind-guy and the elephant story.

Re:Intel leading with heat and watts

[C_Kode]

You noted how you changed (3) Intel servers with (4) GBs of ram for a single AMD server yet you left out a TON of information about the AMD. What and how are the disk connected compared to the Intel boxen. Secondly, how much ram does this 64-bit AMD have? (16GB?)

We had (2) IBM servers (Dual AMD 64-bit Opteron) with 12GB ram each running 32-bit RHEL3 and Oracle 10g. Because it was 32-bit RH it was only using 4GB in each server. We upgraded the RHEL3-64 and Oracle 10g 64-bit (using all 12GB of memory in each box) and we got about 140% improvement on the same hardware.

What was the difference? 8 more GB of ram each. The fact that a single server has 12GB of ram and all queries happen on a single server makes a HUGE difference than have (3) servers with only 4GB of ram as the database can cache more data in memory.

While I don't know your *true* setup, I can say that a single server with a TON of ram will kill many servers with only a little bit of ram on simple select statements. CPU doesn't do a whole lot on select statements compared to what it will do on say stored procs or all kinds of subselects/joins/aggregate functions in your select statements.

Re:Intel leading with heat and watts

[RightSaidFred99]

That would be an insightful comment..if it were made 1 month ago.

Woodcrest changes your equation completely.

Re:Intel leading with heat and watts

[rjstanford]

How are you measuring your performance? Total query bandwidth database SELECTS are inherently bound by the speed of a single processor. More agressive multi-processor design often increases throughput, but for sheer performance a 2-way machine with faster chips beats a 64-node beowolf cluster of slightly slower ones 99.9% of the time. Databases are also often secondarily (or even primarily) disk bound. If you replaced 3 machines with 1, I'm betting that you moved from a network-based disk system to a locally attached one, right?

Re:Intel leading with heat and watts

[Cyno]

Let us not forget clock multiplier locks, single CPU desktop locks, and extremely high price deltas for very very little performance gain between chips.

AMD has not made a friend of this customer. They would have to beg me to stick with them now. They will truely have to beat Intel on price/performance AND overclockability. A C2D 2.4 can overclock well above 3.0 for around $320. At 3.0 the C2D beats the best Opertons for 90% of desktop workloads. I'm interested in video encoding, gaming, compiling and a little data processing. I think I can suffer with a few extra seconds on my builds or home web server requests for the benefit of having a 3.0Ghz dual-core chip for around $300.. I said AMD should have had this out by now, but they won't.. they're stoopid.

What incentive can they give me to purchase a 4x4 now? How can they convince me they're on my side and not their own? I don't think they can.. I don't think they want to.. AM2 is a disappointment.. another XP/MP setup for me to pay for, huh? Fuck that!

Re:Intel leading with heat and watts

[Anonymous Coward]

boxen

Please stop using this word.

Re:Intel leading with heat and watts

[minion]

Sorry if my post was too vague earlier.
 
The boxes we were using were Dual Xeon 2.8GHz servers, 4G of DDR RAM, and 4 x 73G 15K SCSI disks in a 0+1 RAID array. We had 3 of those servers running like that.
 
The new Dual Opteron server is 2 x model 252s, with 8G of DDR RAM (4G per proc), using node-interleaving memory configuration, with 6 x 73G 15K SCSI disks in a RAID 0+1 array, with 2 x 73G 10K SCSI disks mirrored for binlogs.
 
  Our application for MySQL is an ASP app, with each customer having their own database. So, there was no replication between the servers, each server had a unique data set specific to that client.
 
The new setup is able to handle all of those databases on just the one server.
 
  The datasets combined, on disk, consume slightly more than 70G of data. So the databases are quite large, and we are not storing BLOBs in any of the tables.
 
I don't claim to be the best hardware guy out there, but I do keep a fairly close eye on my datacenter, and I can tell you that there was a noticable difference in power usage and heat output from that rack when we switched to the AMD system.

Re:Intel leading with heat and watts

[minion]

AMD has not made a friend of this customer. They would have to beg me to stick with them now. They will truely have to beat Intel on price/performance AND overclockability
 
Overclocking is something that I would never consider for a server that was mission critical. We're talking apples and oranges here. You're looking for desktop performance, I'm looking at server performance, all that comes along with trying to power and cool a room full of servers, while keeping the customers happy.

Re:Intel leading with heat and watts

[Molochi]

What you don't like Brian Regan?

http://en.wikipedia.org/wiki/Brian_Regan

Oh... wrong boxen.

Re:Intel leading with heat and watts

[Cyno]

Why are you talking servers with Conroe and Athlon64? Should you be looking at Woodcrest and Opteron as Intel and AMD suggest for stable server use?

Yes, overclockability matters to me on the desktop. If AMD wants my money they'll pull out the bolt cutters and take off the locks. Or they'll dump billions in process tech and pray they can catch up to Intel in time for the holidays. Its really up to them.

What have they got to lose? I overclock and blow my CPU and rush down to Fry's to grab a new one? Or I purchase the sweet spot ~$300 chip and overclock it to match the performance of their $1000 chips.. but before we had cpu multiplier locks we didn't have $1000 consumer class chips. Evidence that the lock is an artificial monopoly for both Intel and AMD.

Its about time AMD drops the anti-trust/monopolistic practices and gets back to the real tech. No more making excuses for their sorry ass benchmarks. No more hoping lower latency RAM will make a difference. No more fanboyism from me. What incentive do I got anymore?

I want to help them, I want to be on their side. I like them more than Intel. But now they're overpriced. Its going to be hard for them to be competitive again. It might take years.

Re:Intel leading with heat and watts

[john_uy]

well anandtech has a benchmark for the woodcrest (xeons) vs opterons.

and the xeons wins even though they don't have the hypertransport bus. the latest 5000p server chipset is quite good. it has two cpu busses (one for each socket.) good memory reliability features (sparing, mirroring, etc.) and hosts of other stuff.

we're off to ordering these woodcrest systems already and plan in replacing all the servers we have with them.

This is just dual dual-core

[ArcherB]

Correct me if I'm wrong, but isn't this just two dual-core AMD processors on a single board? What's to stop Intel from releasing a dual-socket board and slapping two Conroes in it (provided the chip supports it)?

Re:This is just dual dual-core

[UID30]

appears on the surface to be as you have described, but from what i've read they are also going to support 3rd party co-processors in the extra slot ... for say ... a dedicated physics or encryption engine. not that i think either of those are particularly useful except in overspecialized situations...

i'm not an intel fan ...but i've got to tip my hat to them. they've put out a chip that actually looks "worth buying" for a change. amd needs to update their fab to 65nm and increase on-die cache sizes ... imo that will do a lot to close the performance gap again...

Re:This is just dual dual-core

[Azarael]

I believe that the difference would be AMD chips support hypertransport while Intel's don't. The theory is that with the hypertransport bus the communication between processors will be almost as fast as if they were on the same die. If Intel had to use the FSB, then they'd be limited by the available bandwidth.

Re:This is just dual dual-core

[Toveling]

Conroe doesn't support more than one processor. That's why the speculation says Conroe won't be in the "Mac Pro" - the lack of a dual processor setup. The new Xeons, on the other hand, do it gladly.

Re:This is just dual dual-core

[geekoid]

"a dedicated physics or encryption engine"
like games, and sending private information over the internet.
But who want to do those things with a computer?

Re:This is just dual dual-core

[ricera10]

They're doing it with upcoming Mac Pro from Apple. Two Woodcrest based processors in the same computer in the high-end configuration.

Re:This is just dual dual-core

[twitchingbug]

Like another reply said, the Xeons (Woodcrest) can to multiple chips. Of course, you pay the Xeon price for those too...

-don

Re:This is just dual dual-core

[Pranadevil2k]

Because AMD's processors have the memory controller on-die, and HyperTransport instead of the old shared frontside bus, AMD processors have consistently scaled far better than Intels. This is why Opterons took over the server market, and if prices drop enough it is certainly possible that it could save AMD from floundering in the desktop market as well.

Re:This is just dual dual-core

[Anonymous Coward]

Keep in mind that with AMD's memory controller on chip:

1) a dual socket system gets 2 memory controllers, likely translating into something in the range of 1.5-2x the memory bandwidth, and likely able to load balance well (think "memory access pipelining" that would work similar to CPU instruction pipelining). I wouldn't be surprised to see an L3 cache fitting into the picture somewhere, especially since AMD still works well on smaller L2 caches.

2) designing the memory controllers to play nice with each other would have been a very development costly and complex task for AMD. I can't imagine them putting that much effort into a gimmick.

3) Intel could do the same, and with far less effort, but no benefit at all. Their quad core would be a more effective next step, and one that's on their roadmap.

Now it's possible that AMD could just be ditching the onchip memory controller, but I find that highly doubtful. After all, the onchip memory controllor is pracically the basis of their cpu organization and design philosophy over the last few years. Bear in mind just how long a development cycle is, and it becomes difficult to envision ANY scenario which would convince AMD or Intel to suddenly strike out in a different direction from the one already half-developed.

It's true that this new platform of itself is nothing special. Neither was the first onchip memory controller, until the dual-cores that needed it came along. AMD had a more profound design in the works then, and I would suspect that it does now as well. At the same time, HyperThreading seemed to be a flop in that its development led somewhat to a dead end that had to be redirected before progress could resume (It's tough to say whether HT was originally supposed to lead to dual core or just got presented that way when the roadmap included/announced it). Nevertheless, Intel had a long-term design. The trouble was they had to SLIGHTLY shift their direction of development, and the result was 2 years spent getting back on track. They've done so wonderfully, but in their comeback I see no evidence of AMD's roadmap being similarly thwarted. Rather, Intel is just trying to do what AMD is doing, only better. I wouldn't be surprised if AMD came back on top within a year. Lately, AMD has seemingly taken PR and competition handling tips from Apple, which rarely says ANYTHING very far in advance.

There's no mistaking Conroe rocks. If I were to upgrade right now, it would be to an Intel system. I just don't think I'll have to wait too long to have my mind changed, and this platform MAY play a crucial role (IMHO likely will, even though it just sounds like a stupid gimmick now).

FINALLY!

[Sebastopol]

...us Intel fanboys get to see AMD scrambling to polish a turd, the same way Intel had to with the P4 core for the past 4 years.

AMD CEO to Marketing: "Attention marketing team: Full Steam Ahead with the scrambling and spinning in place!"

I'm going to take a few moments to enjoy AMD's panic. Because: a) its been a long time, and b) it probably won't last long.

Re:FINALLY!

[groundround]

I'm just curious what you get out of being a fanboy. I mean, it is a corporation we're talking about. So, unless you work there, cheerleading a group of people employed to create company profits is meaningless.

Re:FINALLY!

[Sebastopol]

You are 100% correct. It is absolutely meaningless. But I do it anyway. Why? Boredom? A psychological need to choose sides in a polarized world? The satisfaction in getting to say "BURN!" 50% of the time? ...or maybe the fact that all fanboys (and fangirls) realize that such flamewars amount to nothing more than a modern, sapien equivalent of savannah cheetah cubs fighting each other during playtime. :)

Re:FINALLY!

[groundround]

STFU!! AMD is t3h R0x0r n00b!!!!!1!1!!111

Wow, that felt great! I see what you mean!!!

Re:FINALLY!

[vadim_t]

Well, it's good either way, isn't it?

Personally, I root for AMD, not because I love them for some reason, but because they're still the underdog, and the market benefits from competition. So my next system will probably be from AMD anyway, unless it gets really crappy performance. I also used to have a VIA EPIA board, but unfortunately it had stability problems, or I'd own more of those.

If AMD goes bankrupt Intel will be able to just sit on their asses, and that wouldn't be good. We finally are starting to see massive improvements in CPU performance again, and I'd like that to continue as long as possible. Let the corporations fight, and us the customers reap the benefits :-)

Re:FINALLY!

[Cyno]

Without us AMD fanboys Intel might not have so much competition...

It makes sense to me to have people fanatic about the underdog, because it helps balance our fragile capitalist system, where the monopoly is a nature progression of the corporation. It will probably eventually lead to fascism when the justice department gets bought out or gives up the fight.

So I think being fanatic about a monopoly is just stupid, unless they give you lots of free stuff or something.

I'm a fan of competition and technology and anyone who competes with their minds instead of their wallets.

Re:FINALLY!

[maxume]

Owning stock is an ok reason to fanboy, I would think.

Re:FINALLY!

[Anonymous Coward]

I'm not the OP, but it's not what I get out of being a fanboy, it's my hatred for AMD that fuels my pride. I have had nothing but trouble with AMD procs, including stability and massive heat issues. I broke down and bought an AMD laptop thinking "maybe they've improved". Noper. Just today I tried encoding a 600 MB video file using Windows Media Encoder. It said it was going to take 1.25 hours on the AMD-64 3000+ (2GHz) laptop, so I said screw it and started the process on my 2.53 GHz Intel desktop. It took 9 minutes to finish the thing. Quite a difference... Even though the Intel is only about 25% faster, it finished the encode in about 88% less time. (I thought 64-bit extensions were supposed to be the hot shit?)

I guess you get what you pay for... :/

It was all GPU

[Atroxodisse]

The point was that Intel is hyping the new processor for gaming but you really don't need the best processor for gaming. Might as well drop $180 on a good processor instead of $800 on the best because it won't make a difference.

Re:It was all GPU

[Iamthefallen]

Right, you don't need the $1000 CPU. But a $300 Intel CPU would apparently go a lot further than a $300 AMD.

Re:It was all GPU

[ben there...]

Right. As someone was saying in the other article, they didn't test 800x600 (where you would have seen the difference) because most people who spend $500 on a CPU and $500 on a GPU wouldn't play at that resolution.

But my opinion is that if they're testing the CPU, they should test at 800x600, simply to factor the GPU out of the equation. If they're testing for a particular game that's one thing, but they're supposed to be testing the CPU. Even if they pulled the 800x600 out into a separate table for comparison of atypical scenarios, fine, but they should still show it.

If a new generation of GPUs suddenly comes out, you'll be happy you bought the more powerful CPU, regardless of current GPU limitations.

Re:It was all GPU

[mindmaster064]

Complete crap, and completely untrue. Example of the day: You can easily play all half-life derived games on a computer that has 1 ghz. You may not be able to max the res, but you certainly play them acceptably. RAM and GPU are ALL that matter, and for games that use lots of textures fast disk can be handy too. Plunking on the processor is a complete waste of your money because it doesn't make the game any faster (most of the work being in the rendering/paging ram, dependent on DMA xfers, AGP fast writes, etc.) Case in point: I have a five year old PC and get 190 fps in counter-strike 1.6/condition zero. I could probably even play source since I have the ram to do so, again... All I've upgraded is RAM, and GPU. I specifically wait until the GPU drops out to the level of around $150, because before then you are paying the "early adopter" sin tax. Most counter-strike players will play the game at 800x600 even though their hardware could easily do 1600x1200+... We care more about the fps than the res, since fps=response time, extra pixels do not noticably effect accuracy other than to make your mouse travel more. You must realize that the NEW NEW NEW hardware is 2x too much for any current games. Buy your GPU and CPU two years behind and save a bundle, but still frag like a champ! The side effect is that you will also be buying the hardware when the games can finally make use of it.

Just remember that no matter what you read on the net your gaming experience doesn't get any better after about 100 fps, so paying money to have this feature is IHMO worthless and ultra-high resolution adversely effects the usefulness of mice (yes, even the super-cool 2000+dpi mice). You will not frag better for spending the money so skill up, and quit thinking your PC sucks. :)

This was news.... A month and a half ago

[EconolineCrush]

http://techreport.com/onearticle.x/10091

Let me be the first to say to AMD

[Jakhel]

Go! Go! Go!

while you're copying intel...

[mseidl]

Why you copy AMD a lot lately, why not! copy the best thing of them all, HT! HT as in hyptertransport, not hyperthreading. Im curious to see the new 'xeons' compete with am2 opterons. While the new architecture is seeming much faster than AMD's offering at the moment. The FSB is still old and flawed. The AGTL+ max bandwith of the CD2 is maxed out at ~8gb/sec, where as HyperTransport maxes out at ~14gb/sec.

You are Right: AMD may Die

[Anonymous Coward]

Competing against Intel requires entering niche markets that Intel ignores. The reason is that no company has the money or the army of slave engineers that are needed to compete directly against Intel.

Look at what happened to Transmeta. It specialized in low-power processors for laptops because Transmeta could avoid direct competition with Intel in that market. Then, about a year after Transmeta released Crusoe, Intel gunned for that niche market. Transmeta vaporized and its crew of Stanford Ph.D.'s fled.

As for AMD, it had been hoping to avoid direct competition with Intel by focusing on low-power desktop/server processors. Intel had been focusing almost exclusively on (1) performance at any energy cost in the desktop/server market and (2) ultra-low power for laptops. Now, Intel is refocusing on maximum per-watt performance across the entire processor market.

AMD is in deep trouble. AMD has already reduced its revenue guidance for the rest of the year.

Still, there is hope for AMD. If Fujitsu bought AMD, Fujitsu could transform AMD into a formidable competitor in the market for x86 processors. Just look at the SPARC64. It is competitive with Power4 and Itanium.

Obligatory

[Anonymous Coward]

Begun these core wars have

*ducks*

Ah, but you can fudge the number of cores...

[Dr. Zowie]

... as did Intel with their previous round of patched-up dual-core machines. The reason AMD's multicore is so much better than Intel's is because AMD provided a much better caching architecture. Intel's 64-bit multicores could be compared to a large V-8 engine stuck behind a tiny VW carburetor -- totally starved for data. AMD's multicores effectively shared one anothers' L2 caches (a big win), and achieved lower latency on RAM fetches (another big win).

If the two giants start to compete on core count, you can bet your family farm that there will be fudging going on over cross-communication, latency, and RAM bandwidth.

Re:Ah, but you can fudge the number of cores...

[Bitmanhome]

Not at all. AMD's cores have unique caches per core, plus they share the same memory bus. However, AMD put some effort into the crossbar. While Intel's early multicore offerings were essentially the same, there was no crossbar, so bus contention was a little worse.

The Core 2 series now has shared L2.

Re:Ah, but you can fudge the number of cores...

[Joe The Dragon]

Amd's k8l will have shared L3 with unique L1 and L2 cache per core Intel's quad-cores will be 2 dule core linked buy a FSB sound like a repeat of there dule cores.

Re:Ah, but you can fudge the number of cores...

[Carewolf]

Actually you yourself is slightly wrong. The AMD cores do have unique caches as you state, but HyperTransport allows any core to read from another another cores cache. So the separate caches works almost as a combined cache.

It is however not as fast a specifically designed combined cache, so the bonus is only that it's glueless and scales with no redesign to more cores, where Intel has to design a new combined cache for each new number of cores.

Misleading title...

[dtjohnson]

There doesn't seem to be any AMD counterstrike yet other than hot air. It would be a shock if AMD spokespeople said anything other than that they were 'supremely confident.' What else can they say...that they are facing several quarters of deep price cuts, low margins, and they're scared to death about their stock options? The original P4 delivered a pretty big smackdown on AMD that took them two years to come back from and the Conroe Core 2 Duo looks like it's going to do the same thing. AMD still has the better fundamental architecture, though, just like they did against the P4 with its 26 pipeline stages and power-sucking 'netburst' architecture, so in the long run the AMd direct connect stuff should win out but that's not going to put food on the table for the next year or so.

Re:Misleading title...

[Anonymous Coward]

The original P4 put a smackdown on AMD... WHAT WORLD ARE YOU FROM... The origial P4 sucked ass.. People were stupid (like Intel expected) and though at 1.5ghz P4 was linerally faster than 1ghz Athlon and P3, when in reality it maybe beat out the previous generations in a few benchmarks.. but usually was the same... AMD always has something up their sleeves that people who have a clue will purchase cause its better technology...

Re:Misleading title...

[Dastardly]

Counterstrike seems like the wrong work to me also. More of a rear guard action.

I have not been following this that closely, but if I recall correctly AMD has updated cores coming out early in 2007. And, I would hope they had some idea of what Intel had coming, so they would be prepared in 2007. So, we might end up seeing a leap frog process over the next few years. INtel will be fastest for 6-12 months, then AMD for 6-12 months, then Intel, then AMD... Until one of them stumbles and tries a design that does not perform.

Re:Misleading title...

[Cyno]

They could have bought my loyalty by not separating the Operaton and Athlon64 lines..

I bought an MP system back in the day. Today its cheaper to buy a new dual core system than to upgrade the CPUs in that MP system because its AMD's proprietary server tech.

I don't need it. If AMD don't want to work with me, if they should me they don't like me by locking me out of their tech I'll find someone else. If I'm going to support just another monopoly I might as well support Intel. I thought AMD was different. I thought they were on the consumer's side. But now I can't see how they are. They're on their own, until they offer me the olive branch in the form of a dual-core AM2 @ 3.0 Ghz for no more than $300. Can't do it? Tough, Intel can.

Re:Misleading title...

[john_uy]

well i beg to differ. the core 2 has a better architecture now. just see how anandtech is able to overclock (air cooled) to 4ghz with just a good cooling system. without doing anything, intel will be able to release faster processors as amd tries to tweak itself out. right now, amd processors are slow to ramp up their clock speed. now who is having problems?

Re:Misleading title...

[Anonymous Coward]

I'm sorry, perhaps I was asleep for the last 3 days, but since when did architecture start to mean potential for overclocking?

The 4x4 motherboard excellent for servers?

[Futurepower(R)]

Why do they call it an "AMD 4x4 enthusiast platform"? It seems to me that the 4x4 motherboard would be excellent for servers.

Re:The 4x4 motherboard excellent for servers?

[Anonymous Coward]

You can already buy 2 200 series dual core opterons (265,270,275,280,or 285) and a dual socket 940 motherboard and have 4 cores.

Re:The 4x4 motherboard excellent for servers?

[Futurepower(R)]

Yes, but that didn't answer the question.

Re:The 4x4 motherboard excellent for servers?

[fupeg]

It's not new technology for servers and people who buy servers aren't impressed with lame marketing terms like "4x4".

Re:The 4x4 motherboard excellent for servers?

[LurkerXXX]

Yes it does. If it's labelled 'opteron 2x series dual core', it's marketed for servers. If it's labelled '4x4', it's marketed to enthusiasts.

Re:The 4x4 motherboard excellent for servers?

[Anonymous Coward]

They've been doing this with opterons for a while, the only thing new is bringing this tech to the desktop CPUs.

Intel responds with Core 4 Quadro

[poopie]

Intel will see AMD's 4x4 with their Core 4 quadro and raise them with their uber-secret

Mega-Core-8-octo Pentium-Z MMVII Pro ultra-thread 999 energy-star

Re:Intel responds with Core 4 Quadro

[jank1887]

string bet, bad form.

Title

[PuppiesOnAcid]

After reading the title, I was expecting to see projectile CS:Source CDs shattering Intel processors.

Re:You are Right: AMD may Die

[GoatMonkey2112]

AMD is in a much better situation than they have been in the past.

Their server chips will continue to sell well. Opteron is still very competitive in multiprocessor systems.

There will still be people buying AMD processors based on price and past performance. If you've got some market share people will come back to you for upgrades.

AMD has other sources of income than just CPUs. Their flash memory is the most obvious one.

AMD made a name for itself as being a low cost alternative to Intel years ago. This trip into the high end is a new thing and it made them a nice pile of money to invest in the next generation due out next year.

All of that being said, I'm still going to be buying a Conroe. But your predition of the company going under is a major exaggeration. They will most likely be back and strong around a year to a year and a half from now.

65nm

[Azarael]

My question is, how would the comparison stackup once AMD finally releases 65nm chips? Everyone knows that Intel has the best fabs, but I'm curious to see what happens when AMD catches up further in that area.

Re:65nm

[Kjella]

Well, on a guesstimate based on Intel's 90->65nm chips, I would say it could bring the Athlon64s closer but the Core 2 architecture is a winner in itself. Anandtech showed that the new chips could hit 4GHz on air cooling. The K8 architecture, even on 65nm is probably out of headroom and needs an upgrade.

Re:65nm

[Aadain2001]

About the time AMD does, Intel will be moving towards their 45nm or 40nm or whatever their next smallest size is. Intel has the best fabs for a reason: they invested a LOT of money into fab R&D just after the bubble burst. They called is their One Generation Ahead strategy. While everyone else was trying to cope with the loss of capital and drop in stock prices, Intel want to make sure that they came out one generation of silicon manufacturing ahead and stayed that way. While they have in some respects (physical size of transisters) they have missed some advances in some other areas (SOI). In the end, Intel will probably stay ahead of AMD until they hit the physical wall on the size of transistors (can't go smaller than an atom!).

Re:65nm

[chrisinsocalif]

That is incorrect, Okay, let's nail down the transition to 45nm. AMD will have 65nm chips available Decmember 2006. Intel claims that they will have 45nm at the end of 2007. What this really means is that they will have 45nm chips available at the beginning of 2008. AMD will have 45nm chips available mid 2008. FAB 38 will be online at the end of 2008. This will give AMD the capacity for 30-40% of the market. Intel finally released a fast CPU that is only 20% of their CPU production, the rest is Netburst. Intels Advantage will fall when 65nm K8L is released and the K8L should take the performance crown back from Intel. Also Intel has a problem scaling up do more than 2 processors due to their FSB. Also when people say Intels FABs are superior to AMD's that is also incorrect. Intel is about a few months ahead in terms of 65nm. In all other areas AMD is ahead. For example, AMD is moving to second generation SOI while Intel still doesn't use SOI, AMD is moving to 3rd generation Strained Silicon while Intel is at 2nd. Intel still doesn't have the capabilities of APM. If AMD runs a good test the chip can move into production that day. If Intel does a good test it takes 2 months to duplicate it in a manufacturing FAB.

Re:65nm

[Anonymous Coward]

"Intel is about a few months ahead in terms of 65nm. In all other areas AMD is ahead. "

Sure, if you only call 12 months "a few." Doesn't Intel's advantage in the transition from 90nm to 65nm outweighs any other fab technologies that AMD is using?

While I'm at it, why is it that AMD fanboys can't just accept that Intel is back on top? They all say, "well, just wait until K8L sometime next year or 2008!!!"

To which I relpy, "Well just wait until Intel's new chip in 2048!!!!"

How much futher into the future are you willing to go?

Re:65nm

[pradeep mojo]

Woah! and AMD also has the "easy buttom" and the "death star" in its pocket. Anything else you want to throw in there... AMD' fabs are running 3rd gen strained Si and Intel's is a 2nd gen version. Sure buddy...whatever you want to say. Sleep tight..its all gonna be OK. -P

Re:65nm

[jdb8167]

I've got some bad news for you. AMD's K8L is seriously delayed. It looks like the K8L won't be out until 2008.

No AMD K8L processors until 2008, say sources

The 65nm chips in 2007 are going to be the underpowered and unimpressive AM2 versions of the Athlon 64 X2 lineup. Unless AMD has something up their sleeves (always a possibility) they aren't going to catch up with Intel for a while.

Re:65nm

[chrisinsocalif]

According to this article, the K8L is on schedule (reported after the digitimes article) http://www.theinquirer.net/default.aspx?article=32 948 I am just saying its quite possible if AMD gets the K8L out fast enough, it has the potention to take the performance lead again.

Re:65nm

[Anonymous Coward]

AMD is desperately hurting for some decent PR lately. If they had a whole new chip coming out that soon that could compete, I'm sure we'd all know about it. At the very least they'd be coming out and saying that the new chip hasn't been delayed. Right now all they can do is try and hype this 4x4 BS and cut prices so that people still have a reason to buy AMD.

Every other source on the matter seems to keep mentioning summer of 2007 or early 2008 as the release date.

What I gathered....

[Rendo]

From reading the article is that AMD is aware of Intels changing business practices and are now preparing their next line as soon as possible to combat the possible shift in market share. Let's face it, most of us know that AMD has been killing Intel for years and it's about time Intel finally pull their heads out of their asses and start working on something WORTH the money. Not only will this make more competition, it will give us those very nice price cuts that I can't wait for. :)

Consumer Dual Processor

[LordMyren]

Affordable dual processor boards would be very tempting. Unfortunately, I suspect most will be gimped, and only have memroy and a northbridge hanging off one processor. A $250 dual processor board with a memory and a PCI 16x off each processor would have a huge impact on countering Inte's Duo2 strike.

Actually

[necrodeep]

The HyperTransport 3.0 Specification provides 41.6 GB/s of 'aggregate bandwidth' - nice, huh?

Re:Actually

[tomstdenis]

There are a lot of other goodies behind HT3.0. I dunno how many are public so I won't say. But just say "speed isn't the only improvement".

There is no reason why Intel hasn't chosen to do their own HT "like" network. Other than they would have to aboutface on yet another Intel "truth". That the FSB [with their uber overclocked goodness] is the way to go.

Tom

Re:Actually

[Anonymous Coward]

Do you live in a cave?

Intel will introduce integrated memory controllers and routers in 2008-2009 (CSI).

Re:Actually

[tomstdenis]

I'd wait until it's actully in hardware you can buy at the local computer store...

Tom

Healthy Competition

[Stompp]

All I can say is, "keep one-upping each other!" The more competition (not marketing!) we see, the better we, as consumers make out. So what if the performance gains are negligible (in certain areas) the more they release, the cheaper some of these older (still extremely viable) chips get!

Age old fight: Intel vs. AMD... you want to know who wins? Us.

Not really that great

[bberens]

It depresses me that innovation in the processor market is adding additional cores. The geek in my loves it, but the consumer in me isn't impressed. What I (and I presume most other regular users) want is the ability to double-click on my Word/Evolution/Eclipse/Firefox/IE/etc icon and have it instantly display on the screen. What I don't need is to be able to run multiple programs just as slow as I could run one program 2 years ago. What's the holdup? Is it bus speed? Hard drive speed? Memory speed? Will I ever have zero (apparent) latency between running apps and seeing the result? The problem with PC makers is that if they ever do reach the holy grail of zero (apparent) latency, then they will have to decrease the life expectancy of their products in order to continue to make a profit. Maybe I'm slightly off topic and maybe I'm just bitter, but the latest and greatest PC today just doesn't seem to massively outperform the latest and greatest 3 years ago in any way meaningful to the end user. My compile times are faster, but for the most part, users simply can't/don't tax their processor.

Re:Not really that great

[Dastardly]

What you describe is mostly caused by disk speed. You could try 4 disks in RAID 0. Just back up the stuff you care about because now you have 4 devices that can cause loss of all data instead of one.

MRAM may ulitmately result in unnoticeable load times depending on what you can fit on them. But, I suspect disks will continue to be the price leader. givne th echoice of 1TB of hard disk vs 1GB of MRAM you will ikely still be using the hard disk and maybe have the MRAM hold the OS>

Re:Not really that great

[KingMotley]

It's hard drive speed in that scenario. Double click your icon, and watch your CPU usage. It doesn't hit the CPU that hard at all. Your hard disk light is probably on non-stop. Try opening firefox. Wait until it completely loads. Close firefox. Now open firefox again. Now it should have loaded almost instantly (If you have enough RAM). Once the data is in windows disk cache, most applications load instantly (There are exceptions, like games that pre-render some graphics, or uncompress textures and load them into the video card, etc).

Re:Not really that great

[Surt]

It's hard drive speed. Everything else in your system is hundreds to thousands of times faster. Apps like word load in the range of 300-400 MB at startup, which equates to roughly 20 seconds of a conventional hard drive's read bandwidth, plus many, many seeks which cost you in the range of 10ms each.

Re:Not really that great

[courtarro]

I have worked on two friends' Dell Pentium D-based desktops (these are dual-core) and it's amazing how much closer they are to that zero-latency mark. It's likely that these two machines have faster harddrives, ram, and FSB settings, so those are all factors as well. However, if you played with a top-of-the-line system you'd probably agree that we're quite close to that goal. I'd never experienced that kind of low latency.

The other problem is that programs will always continue to bloat until they hit that threshold of "acceptable" latency on each new machine. Whereas version "1" of a program might only load plugins as you use them, version "2" comes out after processors are 4x faster and the developer decides to load all the plugins at the start since the delay is only 1 second longer. Later, version "3" preloads even more stuff, etc. It's a vicious cycle (just look at Vista).

Re:Not really that great

[TTK Ciar]

How about approaching the problem from another direction, and always keeping your applications open? Works for me. I've had this Mozilla process running since June 28th, my mail client running since June 11th, OpenOffice running since July 6th, and various others.

(It helps, of course, if you run versions of software which are stable, which often means foregoing the bleeding-edge latest versions of your favored software. A stable operating system that will keep running for months without a reboot might help too.)

-- TTK

Re:Not really that great

[strikethree]

The problem of "instant activity" programs is that a small hit in performance here and a small lack of optimization there is no problem at all. It is far better to abstract more and have type safe languages. Java and C# are really as instant as a hand optimized assembly program. I mean who can even notice a microsecond here or a millisecond there? Right?

Gotta throw away C. (Have you tried running Windows 95 on a 2ghz processor?)

strike

What benchmark will they use now?

[Amiga500_Rulez]

If anyone has looked at the C2D specs, the MHz race is over (for now). AMD stopped publishing their MHz and relied on the relative performance to the P4 counterpart. Now with Intel releasing the E6300/6400/6600 what will AMD do? We used to have MHz, MIPs, bogoMIPs, FLOPs, etc.. What will be the new benchmarking standard?

Re:What benchmark will they use now?

[geekoid]

PONIES! OMFGLOL!
kbye

Re:What benchmark will they use now?

[Anonymous Coward]

The article header would have us believe "Valve software" is the new benchmarking standard. ;)

Re:What benchmark will they use now?

[Solder Fumes]

...What will be the new benchmarking standard?

I have an idea: Benchmarks.

Re:What benchmark will they use now?

[labradorx]

I propose to institute an FB benchmark(FanBoy). There will also be a corresponding pwned!!! standard.

Pretty much

[Sycraft-fu]

I like my dual core system but really, had I not needed a new motherboard to get PCIe, I'd still be running a single core P4 2.4GHz and doing fine. I'm not opposed to the push to mroe cores, I think software makers need to start thinking mroe multi-threaded, but I don't think it's worth going overboard on it until the software starts catching up. Give it another 6-12 months. Between the dual core PCs, the 3 core X-Box 360 and the Cell in the PS3, I'm seeing multi-threaded thinking taking a huge upswing. Then maybe it's time to start looking at 4, 8 or more way computing on the desktop. Until then, I think it's just trying to show off your ePenis.

This, of course, doesn't apply to research/simulation types of situations, but that doesn't sound like the target here.

Only one comment...

[SixDimensionalArray]

..isn't competition grand! Look at the innovation these two companies are making at a breakneck pace! Woohoo!

Re:Only one comment...

[pradeep mojo]

Oh definitely agree. But not the 4x4 crap though. I'll wait for the real thing from AMD... -P

Counterstrike

[HoboMaster]

AMD Launches Counterstrike Against Core 2 Duo, but is defeated, citing lag problems as the cause of death.

AMD Launches Counterstrike Against Core 2 Duo

[zaphod_es]

Tut tut! Another sloppy editor. What is wrong with the industry standard:
AMD launches Core 2 Duo Killer

Wake me up when it's really a 4x4.

[Kadin2048]

I would have been a whole lot more impressed if they had actually made a "four by four" machine, instead of just making up a nonsense term for what's nothing but a regular dual-socket, dual-core setup. At most, I'd call that a "2x2." Four-by-four? It doesn't have four of anything in it, certainly not four of anything by four of anything else. That's just misleading.

Two cores per processor times two processors ought to be called a 2x2, and a 4x4 ought to mean a four-socket mobo with four quad-core processors, for a total of 16 cores. Similarly, what they're calling an "8x8" ought to be called a 2x4, or maybe a 4x2, since it's four processors times two processors per core.

For an 'enthusiast' product -- which they're apparently hoping to sell to people who have a clue -- that's a stupid way to name it. Plus, as multi-processor, multi-core systems become more prevalent in the future, it would be nice to have some clear nomenclature to describe them. AMD is just starting everyone off on the wrong foot by calling their dual-core/two-way systems "4-by-anythings".

Re:Wake me up when it's really a 4x4.

[Dare nMc]

>actually made a "four by four" machine, instead of just making up a nonsense term
I completely agree with your post. but I will point out a 4x4 pickup has 4 driving wheels out of 4 wheels, and a 2x4 has 2 driving wheels out of 4.
so a 4x4 processor (uses a stupid analolgy but..) has room for 4 cores, all 4 supplied. so the 2x4 would be a dual core in one slot, or 2 single cores thus room for 4 "cores" but only 2 supplied thus 2x4. the 8x8 thus the first 8 describes the number of "cores", the second 8 would describe a board having room for 8 "cores."

Re:Wake me up when it's really a 4x4.

[Dastardly]

How about 4 cores and 4 memory channels?

Re:Wake me up when it's really a 4x4.

[TheRaven64]

Look, the name is perfectly simple to understand. It's 2 AMD Althlon 64 X2 CPUs. Put that together, and you get...uhhh...

Anyway, ummm, I'm sure it does make sense really...

Re:Wake me up when it's really a 4x4.

[Gherald]

Do 4x4 cars have 16 wheels?

Re:Wake me up when it's really a 4x4.

[Eugene]

the 4x4 platform is called because it's 4 CPU core and 4 GPU core (dual nvidia 7950).

Re:Wake me up when it's really a 4x4.

[fitten]

Yeah, and even with the upcoming price cuts, that's a bloody fortune to pay to play. A system like that is bought only by those with much more money than sense (and those who feel they have to make up for something else by having a huge system). The number of people who will buy into that fanboism market is tiny compared to what even the $316 Core 2 Duo with a reasonable graphics card (X1800 or single 7800GT type) or even a simple AMD X2 3800+ with a similar graphics card is. Even then, those are smaller than the group of people who will buy $200 graphics cards like the X1800GTO and the 7600GS range.

It's called "flailing about just in order to be seen on a day when the tide turns".

Re:Wake me up when it's really a 4x4.

[salimma]

Maybe they're letting us know we need four RAM modules for optimal performance ..

Re:Wake me up when it's really a 4x4.

[courtarro]

So when are those 8x8 trucks going to be available to the general public?

Or for that matter, when are those 64-core mobos going to be available?

Re:Wake me up when it's really a 4x4.

[sebol]

>>So when are those 8x8 trucks going to be available to the general public?

if 8x8 truck is 8 driving wheel out of 8,
This is the truck:-
http://bigfoot4x4.com/images/bf3-3.jpg

Re:Wake me up when it's really a 4x4.

[Anonymous Coward]

The fanboi market is lucrative -- I mean, who are the people buying $800 leather jackets? $250 Air jordans? $45,000 sports cars? It's all about penis power.

Sure you can

[wbean]

Sure you can. All you have to do is dig out those Dos 6.0 disks. Both your boot and application startup times will be blindingly fast. I've been tempted to try it :)

Clarification

[Kadin2048]

Before somebody jumps in and responds to this line: "Four-by-four? It doesn't have four of anything in it, certainly not four of anything by four of anything else." and says 'but it has four cores total in it!', yes, I realize that.

What I should have said was that it doesn't have any single part with four of anything in it, so it's not as though they're doing a "four-way" something times four of them, which is what "4x4" logically suggests, IMO.

This will need New FX CPU's = $$$

[Brit_in_the_USA]

I read elsewhere that this 4x4 baord will only take special FX designated CPU's with new pin count and two sets of hyper transport links to support CPU-CPU and CPU-chipset comms. You will not be able to slot in a regular "enthusiast" CPU such as the present FX or X2 as there is only one hyper transport lane per CPU.

This will become a VERY expensive package with premium CPU's and a premium Motherboard.

Fanboyism

[otherone]

is just another type of advertisement, abeit free.

This is going to compete?

[jonesy16]

Maybe it's just me, but I don't see how doubling up the number of processors/cores will help AMD to compete with the Core 2 Duo and Core 2 Duo Extreme processor offerings. The benchmarks posted earlier today clearly show that Intel's latest offerings leave AMD's processors in the dust both in terms of cost, efficiency and performance. What possible "enthusiast" is going to see the benefit of running 4 cores vs. a dual core? I mean, gamers won't see a performance boost from that until games are truly multi-threaded, and even then they'd have to be optimized for more than 2 threads. Regular desktop apps don't challenge the CPU as it is. Not to mention that putting in 2 AMD 64's is going to generate twice the heat which is the area that AMD fanboys used to brag about the most. So somebody please explain to me in terms an "enthusiast" would understand how 2 slower, hotter, more expensive AMD 64's is worthy competition for a single Core 2 Duo, and then tell me that AMD has nothing to worry about . . .

Memory Bandwidth

[turgid]

A single (single or dual core) super-duper fast processor may be good at some benchmarks but consider this.

Plugging in two Athlon 64s/Opterons doubles the memory bandwidth due to the NUMA (Hypertransport) architecture. That intel processor is choking on an old-fashioned 1980's-vintage front side bus. If you have two, they're both fighting over that bus.

This is why Pentium multiprocessor systems don't scale well. You get a bit of a benefit with a second processor, with very small and diminishing gains with additional processors.

Re:Memory Bandwidth

[KingMotley]

NUMA isn't a hypertransport technology. There are many Intel multi-processor machines that have NUMA and multiple banks of memory. Even the old Data General P2 server we had used NUMA.

Re:Memory Bandwidth

[turgid]

You're right. Hypertransport is just the physical interconnect, but the Athlon 64/Opteron supports NUMA on the processor. Pentiums need external (slow) hardware.

Opterons have many hypertransport point to point links. Pentiums have a single shared front side bus.

Re:Memory Bandwidth

[KingMotley]

I'll conceed that there are benefits to Hypertransport over Intel's FSB technology. Even Intel itself is planning on implementing some of the same types of technology down the road (Although the current plan is to not use HT).

However, the FSB/non-integrated memory controller does afford Intel the flexability to quickly change whole processor families to being able to support different memory technologies (RAMBUS/SDR/DDR/DDR2/?). They only change the northbridge, and the whole processor line is good to go. Looking back, it may have been a good thing (or bad) considering they were using Rambus and needed to quickly switch to DDR/DDR2.

In large servers however, you won't find the issues that you mentioned. Each processor can/will have it's own memory bus. It's only at the low end servers that you see memory busses being shared, and even then only on cheap server board implementations. Cheap dual-processor servers is probably the largest market sharewise in the server space, so being able to build cheap dual (and quad) processor machines is a fairly big deal, and something AMD currently has an advantage in. Step up to 8x, 16x, and 32x processor machines however (where prices generally tend to skyrocket), and you'll see Intel Servers with memory busses for each processor, multiple northbridges (or what would be similiar to a northbridge). Because of this, at the high end, AMD's performance advantage because of hypertransport is negligable (again).

I'm not arguing that Intel's method is better, because it really isn't. But it's not as bad as many seem to think when they don't look beyond the entry level server area. Intel never said (I don't believe they did) that integrated memory controllers were a bad idea, nor would better interconnect be beneficial. They said that it wasn't yet time, and they prioritized other things higher. So far, I'd have to agree with that. I haven't seen any overwhelming reason to have had an integrated memory controller or hypertransport in the past few years. The time may very well be approaching when we do need it, but we didn't need it when AMD delivered it. So long as Intel does deliver it before I need it, I'm good with that.

Re:You are Right: AMD may Die

[Courageous]

Agreed. They still command the IO advantage in the midrange server community. And they still have some of the best EE talent in industry, made possible from a whole slew of defections from intel and generally community elan over the last four-five years or so. Meanwhile, Intel's been spilling their strategy beans for months now.

AMD's biggest problem is not being 65nm yet. However, the process alliance on that front is quite strong, with lots of collateral interests. AMD, IBM, and (more recently), SUN. They're obviously late on the 65nm front, but they have things held close to hand they're not showing yet, and when they get to 65nm, it will be a *better* 65nm than what Intel has.

C//

one word

[Anonymous Coward]

headshot!

-m10

If you've never really *used* an mp system....

[Savior_on_a_Stick]

Then you really aren't qualified to comment.

I'm not trying to be catty here - just that this is one of those driving experiences you must experience to appreciate.

No, mp isn't required for general gaming - but keep in mind that just becauswe one games, it doesn't necessarily mean that they *only* game.

Let's say you have a high end box you use for rendering. It has a fast cpu, high end graphics and a speedy file system.

Would it make sense to build another box to use for gaming? Likely not.

I had to build a fat box once to process web server logs.
It really *required* dual cpu's and lotsa ram, and would have happily gobbled up 4 cpu's.

Having multiple cores also allows one to come closer to multiprocessing, rather than mere task switching.

Keep in mind too that certain high end apps are well known for gobbling all available resources.

Having 4 cores and running an app that will access two of them means that you can still do other work even when two of the cores are cpu-bound.

Dual-processor machines are nothing new

[GWBasic]

This 4x4 thing is a bunch of marketing BS. Back when I entered college ('99) people were running desktops with dual Celrons. Putting two CPUs in a desktop is old news.

or AMD may hire Intel layoffs

[Wry Cooter]

what about The Intel layoffs?

What parts of the company do you think they represented?

Who do you think they will go work for?

My wild guess is, they were the sales force needed to touch
base with Dell, and some of these layoffs may end up working
for AMD in the same capacity, as account managers for Dell
at AMD.

So...

[Runefox]

When can I see my 16-core hyperthreading AMD Opteron system with 32GB of RAM, 4 high-end GeForce cards in SLi and ten 500GB HDD's?

Mmm... Overkill...

Improvement can often be a slap in the face...

[Archie Binnie]

I find it an interesting thought that the statement from the article quoted, "while imitation may be the sincerest form of flattery, improvement can often be a slap in the face" is true for two competing hardware companies, the same could never really be said of contributions to open source software. You don't generally see too much complaint about open source developers using and building upon existing projects, or even outright borrowing the best features of competing projects (e.g. Gnome/KDE, Linux/BSDs).

Re:You are Right: AMD may Die

[pradeep mojo]

"the 65nm front, but they have things held close to hand they're not showing yet, and when they get to 65nm, it will be a *better* 65nm than what Intel has" Better 65nm than what Intel has? Care to substantiate that and pass the crystal ball to me when you are done. -P

Everyone not getting it

[charnov]

Okay, I'll give it the Slashdot norm, but nobody gets what this is. It a hypertransport socket for not just another CPU, but ANYTHING you would want to connect directly to memory and CPU. No PCI or other slow bus.

There are already Xilinx cards available because this has been used in Cray supercomputers for a while (the Opteron ones anyways). This means AMD can counter ANYTHING Intel puts out because you can just slap a $20 speciality DSP on the mobo which could easily be 100x faster than that Intel chip at whatever small set of functions it needs. Video cards are already in the works for this along with all kinds of audio and video stuff. I seem to remember one manufacturer has a RAID processor. The possibilities are endless.

Re:Everyone not getting it

[BFaucet]

Wow! I never thought of such an application.

I'd really like to read up on some of these projects. Do you have any articles or links with more information?

Re:Everyone not getting it

[MrEd]

I am interested in purchasing your theory.

Hmm.

Re:Everyone not getting it

[walshy007]

I have news for you: xilinx cards can be put in nearly anything, the xilinx card your most likely meaning is the one amd showed with a virtex 4 chip on it, xilinx fpga's are very awesome, that being said, with the right design to put on them, you can put them onto basicly anything, I just wish there were more suppliers of their products so the fpgas weren't so damn expensive in small quantities

As a multithreading-happy programmer...

[Anonymous Coward]

I say hooray!

Is that the major difference?

[Futurepower(R)]

Is that the major difference in raw CPU power?

Does the Core 2 Duo support 64 bit instructions

[lamikr]

I know that the X64x offers 64 bit instructions by default and also the 32 bit instructions for the backward compatibility. Is the Core 2 Duo offering the same or is it more a 32 bit system? (I mean will 64 bit Linux distros be able to run on core duo 64)

Re:You are Right: AMD may Die

[Bobsledboy]

AMD uses silicon on insulator technology in their fabs, Intel doesn't. SOI provides better performance for a given size, hence AMD 65nm is "better" than Intel 65nm, this is also why Sony, Nintendo and Microsoft are all using it in their next gen consoles.

Cheaper = good

[billcopc]

If AMD's 4x4 or 8x8 concept means the average code monkey can get their hands on a multicore rig within their budget, I'm all for it. It's lonely up here with a quad dual-core opteron rig and no software to take advantage of it. "emerge -e world" was cool the first dozen times, but now I need a better screen saver :)

Seriously kids, learn to write SMP-aware code everywhere, it's not that hard if you plan accordingly. You will thank yourselves when everyone else's apps crawl on the inevitable multicore systems of the imminent future.

So that probably means...

[Ayanami Rei]

You have a VIA KT3xx chipset. Or an AMD 760.

The AMD 76X pros were much better. I made the mistake of buying some 760s when the came out on the market, I should have waited 3 months.
We got the 76Xs when we bought addt. units and yeah, they were much better (faster FSB speeds too).

It's kinda sad, we've EOL'd all of our 1Us based on that platform. I'll sure miss those guys. *tear*

Re:So that probably means...

[Cyno]

Yeah, its was the AMD 76x, before the pros I think..

It was a really nice system, other than the stability issues. I wish I could have afford to upgrade the CPUs to 2+ Ghz. It could have been really nice.. if AMD..

but nothing I say here will change their minds..

Fukkin signed.

[Ayanami Rei]

Honestly, I don't know why people put up with "Enthusiast" boards or the FX lines of chips because it's all overpriced crap.
ASUS makes workstation boards with one or two sockets and they let you play with your FSB and often allow you to play with the CPU multiplier as well, even with Opterons.

A lower-priced 265HE (dual core) with a FSB boost is very stable on air cooling and sips power. Insane memory bandwidth + decent CPU performance = stability / cost / performance wins.

I mean...
A 265HE is about $370 right now. *2 = $750, tops, with shipping.
Comapred to like an FX-anything which is at least $750, or more.

Re:Fukkin signed.

[Cyno]

Yes!

Go buy a 1.8 Ghz x 4 Opteron Server today for only around $1200..

Not such a bad deal, really, but single thread and gaming performance would suck compared with these new chips..

We really need 2.2+ Ghz Opterons for that price to be competitive with Woodcrest (2.0 Ghz @ $350 + 4MB cache)..

If times weren't a changin I would think AMD still kicks ass, but right now they seem stagnant. I want to be able to say without a doubt AMD is still the best bang for the buck, but I can't say that anymore, so I'm upset.

Re:You are Right: AMD may Die

[misen]

AMD has no flash anymore -- they spun out the Spansion joint venture last year. AMD is entirely a CPU company now.

Capitalism at work

[Anonymous Coward]

Sounds like AMD may have had a performance edge and a better idea of what the latest architecture should be, but then Intel stopped driving towards a cliff and started putting their expertise and manufacturing power into something closer to what AMD had in mind. If AMD wants to take away more of Intel's market share instead of gradually losing what they've gained (if any), they'll have to one-up Intel again. Profit!

Wake me up when AMD stops making false named CPU

[Anonymous Coward]

I agree with what you said, and have to comment as far as I remember AMD as a company they have always 'played' with thier CPU names to trick people into thinking things are better than they really are.

counterstrike

[john_uy]

well i guess if amd cannot win technically at the moment, they fight legally. i seems they've been launching antitrust suits left and right. hmmm... is amd losing its focus and using legal tactics instead of better engineering (not that i think intel did not violate any antitrust laws)?

if you can't beat 'em, sue 'em.

Sorry slow to respond links below

[charnov]

http://www.hypertransport.org/products/productdeta il.cfm?RecordID=65
PathScale Infiniband card. Lowest latency infiniband neworking card in existance (1.5 microseconds).

http://www.supercomputingonline.com/article.php?si d=11429
Xilinx card

Articles about HTX and 4x4 (Torrenza) tie in:
http://www.xbitlabs.com/news/cpu/display/200606070 74412.html
http://www.amd.com/us-en/Weblets/0,,7832_8366_5730 ~109409,00.html
http://www.tgdaily.com/2006/06/02/qanda_amd_vp_ran dy_allen/

There are many more, but this is a start.