Slashdot Mirror
Dell Set to Introduce AMD's Triple-core Phenom CPU
An anonymous reader writes "AMD is set to launch what is considered its most important product against Intel's Core 2 Duo processors next week. TG Daily reports that the triple-core Phenoms — quad-core CPUs with one disabled core — will be launching on February 19. Oddly enough, the first company expected to announce systems with triple-core Phenoms will be Dell. Yes, that is the same company that was rumored to be dropping AMD just a few weeks ago. Now we are waiting for the hardware review sites to tell us whether three cores are actually better than two in real world applications and not just in marketing."
Enable that other core!
Just callin' it like I see it.
Making 3-core machines out of 4-core CPUs will do wonders for their yield. So many chips get trashed because of single tiny failures, this will allow them to keep any chip with any number of failures as long as they are limited to just one of the cores. The same sort of benefit Intel saw by using Pentiums with bad cache segments to make Celerons, or nVidia saw when disabling (supposedly) bad pipelines to turn 16-pipe GPUs into cheaper 12-pipe versions.
I am sure some units will make it through the process with a functional-enough fourth core to be useful to "overclockers", but I think the majority will have actual problems. That is, unless there is no 4-working-core version of this processor for the known-working ones to be sold as?
One concern... How do they keep thermal load even if 1/4 of the die is not running?
So, does one have to purchase 1.5 Vista licenses?
Microsoft has declared for all their products that a processor is defined as a physical processor in one socket. No matter how many cores it has, it is a single CPU for licensing purposes. Also you don't have to buy more licenses to run more processors, you have to buy different versions. Last I checked it was 2 processors for workstation versions, 4 for server, 8 for advanced server and 32 for datacentre. Not sure if that's changed.
At work we have purchased a dual processor system with a quad core CPU in each that runs Vista. All 8 cores show up and are usable by software.
Works for razors - 2 is better than 1, so 3 has got to be better than 2. I'm not switching from Intel until someone comes out with 5 - count 'em, 5! - micro sharp cores...
When Dell bought Alienware (who used AMD CPUs) Dell began using AMD.
3 cores will be better if you have a use for them. It's that simple. That answer will hold true for any arbitrary number of cores. Basically you need to have a number of threads equal to or greater than your number of cores that each need a lot of CPU time. This could all be from one program that's heavily multi-threaded and CPU intensive, or it could be from multiple applications running at the same time.
For most things, no 3 cores isn't really going to be much benefit at this point. While there are now multithreaded games out there that make use of 2 cores pretty well, they don't really scale past that at this point. I imagine that'll change as time goes on since quad core processors are getting more common, but it hasn't yet. As for desktop apps, well they don't tend to use much power so it won't help much. I suppose it might help responsiveness in some cases a tiny bit, but I doubt it.
However for some professional apps it can help. Cakewalk's Sonar makes use of multiple processors quite handily. Every effect plugin, every instrument, all run as a separate thread so it can easily use a large number of cores. I've seen it run on a quad core system and it distributes load quite well across them. I don't imagine anything would be different with 3 cores, it'd just have one less to use.
[For those too young, the reference is the 1975 SNL parody about the Remco Triple Track Razor - done just after twin-bladed razors first appeared.]
It must have been something you assimilated. . . .
Ironically, the main advantage of dual-core has nothing to do with applications taking advantage of that second core -- in fact, just the opposite.
Dual-core means that for most cases, I can run a video encode, a backup/compression process, a long-ish compilation (of the sort that doesn't like 'make -j2'), etc -- not so much all at once, as I can fire off any background process and not worry about it, as I have a whole other core to use. It's shameful -- Amarok will occasionally use 100% of one core, and I won't notice for hours.
Having more than two cores wouldn't benefit me a lot right now. I wouldn't mind it, certainly -- I've been playing a bit with things like Erlang, which should be able to scale arbitrarily -- but I think the real applications are only just catching on to the idea that threading is a good thing. I imagine it's still going to be a lot longer till a quad-core machine is useful for anything other than, say, running virtual machines, as most programming languages do not make threading easy. (Locks and semaphores are almost as bad as manual memory management.)
While I'm playing crystal ball, I'll predict that the first application of multicore will be things which were already running on multiple machines in the first place -- video rendering, for instance. Not encoding, rendering.
The second application for it will be gaming. This will take longer, and will only be the larger, higher-quality engines, who will simply throw manpower at the problem of squeezing the most out of whatever hardware is available.
I suspect that the old pattern will be very much in effect, though -- wherein gamers will buy a three-core system and unlock the fourth one (if possible), then use maybe one core, probably half of one, with the video card still being the most important purchase. If there's a perceptible improvement, it'll be because their spyware, IM, torrents, leftover Firefox with 20 MySpace pages and flash ads, etc, won't be able to quite fill the other three cores.
I'd like to add that for most people, including me, one core is plenty if you know how to manage your processes properly -- set priorities, kill Amarok when it gets stuck in that infinite loop, and get off my lawn!
Don't thank God, thank a doctor!
Different OSes have different methods for managing threads. In the case of Windows it shuffles them around as it sees fit. If you have three apps all using 100% of a core then yes, they'll get stuck each on their own core. You can also force it in task manager, where you can tell Windows which cores a given process is allowed to run on.
In general most modern OSes do a pretty good job moving things around. It isn't necessarily an app per core situation since many apps don't use much power and thus can all run on a single core. Also a single multi-threaded app may run on multiple cores at the same time. In general the OS will move things to try and get all threads as much CPU as they want, and to try and have CPU left over for new tasks.
It is getting more common for companies to physically disable the section on a chip that isn't supposed to be used. I'm not sure how it is done but I imagine just burning the traces with a laser would work. I'm going to guess AMD will be doing this with their 3 core systems. It servers 2 purposes:
1) Reduces complaints. You'd get people who would enable a defective core and then bitch that their system didn't work, especially since it could be somewhat random when failures happened.
2) Allow them to have a cheaper part. Yields may improve to the point that there are few defective cores, however there may still be demand for the cheaper part. Thus disabling 1 core allows them to continue selling both.
I think I remember reading an article on Tomshardwareguide where they tried running one dual core, and a single core CPU in the same system for 3 cores. While they got it to boot the OS, a lot of applications failed to run.
I'm guessing there is a lot of code out there that's looking for power of 2 number of cores. A program might run fine with 1,2,4,8, or 16 cores, but if you do some kind of odd number I wouldn't be surprised if several applications just refused to run. It will be interesting to see what kind of compatibility testing AMD has done with this new processor.
In the end though, this just seems like another last ditch attempt by AMD to marginally compete on the lower end market with Intel. Intel says they have no need for 3 core chips since their yields are so much higher.
You're sold a three core chip, it has three working cores.
Which part of that is "defective", misleading, or unfit for purpose?
How many dual core chips are really four core chips with two failed cores? Do you know? Face it, it's just the number three which bugs you, and that's pretty childish...
No sig today...
Is that a cheap attempt at humor?
Or maybe you don't understand manufacturing.
Not a shyster; no suckers.
(It would be interesting to pit an AMD Triple-core against Intel's Quad-core.)
Computer chips have billions of transistors, capacitors, resistors, and interconnects. All of them have to work to make the chip work.
Even in the says of tubes (valves), the manufacturers tested their product, then set aside the best to sell at a premium.
Intel used this technique on their 486SX processors. When the FPU on a 486DX tested defective, they could disable it and sell it as a 486SX. They probably still use the technique with multi-core processors. It would be stupid and wasteful not to.
Hard drives hold billions, even trillions of bits. All have to work. Drive makers have always mapped out defective sectors. Now they do it transparently. Flash disks too.
MacDonald's advertises "Billions Served." Imagine if they could say, "Billions served without a mistake."
When is the last time you were able to produce millions of items without a defect?
There are a couple known problems with the first spin of the Phenom die (codename Agena).
The first (and less relevant) problem is the TLB errata. The second (and more relevant to this discussion), is a problem in which core #2 (out of [0,1,2,3]) is lower yielding than the first three. For example, on the same CPU die, cores [0,1,3] may work fine at 2.6Ghz, but core [2] yields only at 2.0GHz. This is a widespread problem, mostly found out through failed overclocking attempts.
Google it yourself and find out..
As I have stated before:
Many of the newest Operating Systems, applications, and games are multi-threaded. Multiple cpu cores just allow modern systems to take advantage of them, when available.
I have a dual quad-core computer, that dual boots Windows Vista Ultimate, 64-bit, and Fedora 8 Linux, 64-bit. Many programs do take advantage of this system, including modern PC games, such as Crysis and Unreal Tournament 3. UT3 does use all 8 cpu cores during parts of the game.
So, even though multiple cores are not necessary, I find it helps in many ways, and many programs. The system seems to perform very smoothly.
"Two things are infinite: the universe and human stupidity; and I'm not sure about the universe." Albert Einstein
I hate it when people tell me this. They have dropped WAY to much effort into the whole 6950 and SC1435 lines. Hell, the new 2970's are out if not already.
My personal opinion is that they still need to be fleshed out though. I am not sure why, but all the AMD systems we have only accept DDR2 unbuffered as well has having issues with very large amounts of ram ( More than 64gigs). I will admit however, they use ALLOT less power and much quieter.
Just making sure your references are noted ;)
"Old man yells at systemd"
For reference, see The Onion reference, "... We're doing five blades". (Rough language. If you're at a school maybe NSFW). From February, 2004. For the record, the Gillette Fusion with five blades and two lubricating strips was introduced in early 2006.
Hilarious though:
I'm a big AMD fan but three cores are barely better than two. Buy it anyway - AMD needs to live if the computer market is to be bearable at all in ten years. Via makes some interesting stuff too - and they're not afraid to cut the watts and make them small. You can do some very neat stuff with a low watt CPU on a small board.
It doesn't take a great deal of insight to see we're going to 8 cores per processor on the desktop sometime in the next few years. Dual 16 core processors will happen within ten if competition keeps the pressure up. Personally I don't care if every core is on a separate slab of silicon as long as they integrate in the package well. Yields are better that way I imagine. Somebody tell them to get the watts down. Electricity is mostly made from CO2 emissions:
Help stamp out iliturcy.
With one dead core dropped per processor, that would explain the rumours.
Excuse me, but please get off my Pennisetum Clandestinum, eh!
I've hacked a couple of graphics cards by moving a resistor on the top of the chip. One was a GeForce and it came up afterwards as a "Quadro". The other was an ATI 9500 which came up afterwards as a 9700 (more shaders). Both cards worked perfectly for years.
No sig today...
I am TheRaven on Soylent News
Cores don't add. That's problem number one with your confusion. You can't add clock speeds together because you have multiple cores. There's a lot more logic involved, and speed is dependent on a lot of other things in hardware (RAM, bandwidth, etc.). How effective multiple cores are depends on how threaded an application is, and on the quality of the operating system's scheduler. In some workloads, a dual-core might be twice as fast as a single-core, and quad core twice as fast as a dual core. In other workloads, a quad-core may only be 50% faster than a dual-core, and a dual-core might be only 50% faster than a single-core. Again, it depends on a plethora of hardware bottlenecks and software factors.
There's also the fact that clockspeed isn't the only metric - an AMD chip at the same clockspeed as an Intel one may actually be slower overall (or faster at some things and slower at others). This is because what you're interested in is work/second, not clocks/second. Assuming you get the same amount of work done per assembly instruction (since it's all x86 with only minor differing extensions, that's not an outlandish assumption), instructions/clock is a crucial metric. Because of various factors, Core2 Duos can do more instructions per clock than Phenoms. Previously, Athlons were beating Pentium4s at instruction/clock. So clockspeed isn't the only metric, and in fact isn't the most crucial one.
Additionally, most CPUs have only one clock and one voltage setter. So either the entire chip runs at 2.6 GHz, or the entire chip runs at 2.0 GHz. You can't mix and match them currently. Because you need a stable processor, you're only as strong as your weakest link - if one core can only hit 2.0 GHz at a set voltage, you have to make the entire processor 2.0 GHz. If disabling that core lets you hit 2.6 GHz with the 3 "healthy" cores, that may be a more attractive option, depending on the workload. Because a lot of software isn't multithreaded, 3 faster cores are sometimes superior to 4 slower ones. Heck, a 3.2 GHz dual-core is sometimes better than a 2.4 GHz quad-core (for some limited workloads).
Processors aren't designed individually, they're made by the thousands. They start out as silicon wafers. Then they get put in a machine with a whole bunch of lasers and stuff I don't even pretend to understand, which etches a few dozen processors on the wafer. Because of a variety of factors (manufacturing process issue, stray pieces of dust, impurities in silicon, whatever), some cores wind up testing better than others. A processor which can meet the 2.6 GHz benchmarks gets sold as a 2.6 GHz chip. The chip next to it may fail the 2.6 GHz tests, but meet the 2.2 GHz benchmark, and so gets sold as a 2.2 GHz chip. If a dual-core chip has one busted core (some kind of massive defect in one core but not the other), it gets the bad core blasted off and lives life as a single core chip. If a chip has an issue with some of its cache, then it gets half the cache disabled and is sold as a Celeron.
It's not a hassle to manufacture this extra stuff, whether its cache or cores. It's actually more of a pain in the butt to completely re-tool all the machines to make a pure triple-core. If you look at the economics of it (and I've only done that from the homework standpoint), most of the cost is the fixed cost of buying the machines and setting them up just right. After that, the goal is to get as much out of the chips you manufacture as possible. The choice you're making is between selling a chip with features disabled for a lower price, or tossing it in the trash.
Each chip has 4 cores, but with the slower core enabled, the chip can only hit 2.0 GHz. Without having to deal with the slow core, the other 3 can run faster (at 2.6 GHz). Obviously, AMD would prefer to sell the chip as a quad-2.6, but they can't. They can sell it at the speed it can hit with 4 cores (2.0 GHz), the speed it can hit with 3 cores (2.6 GHz) by disabling a core, or throw it out as defective.
You are living in the past on that quote.
AMD used a per core cache on older designs. On the new design they use both a per core cache AND a shared cache. So on a quad core that has 512k per core and 2m shared the cache for a chip with one core disabled is (512x3)+2048/(512x4)+2048 or 7/8. So instead of disabling 1/4 of the cache they are disabling 1/8th, but because they disable 1/4 of the cores when disabling 1/8 of the cache it actually helps the cache per core ratio instead of hurting it.
Tri core Phenoms get 1195k of L2/L3 cache per core in that example. Quad core Phenoms get 1024k. So the tri core gets 16% larger cache based on that logic.
Besides that math is wrong/too simplistic because you are only considering L2 and L3 cache. Each core also has 128KB L1 but it seems in vogue to ignore it. It makes the math simpler especially when you get to 45nm and below when you bump that L3 cache up every time the process improves. 6MB L3 plus 512kb L2 on a tri core vs 6MB L3 plus 512kb L2 on a quad core gives you a 15/16 ratio vs the 3/4 ratio. The bigger the L3 the better the advantage for the tri core.
2560 vs 2048 in the 6MB L3 cache scenario, the 16% advantage becomes a 25% advantage at that node.