AMD Announces Triple-Core Phenom Processors 334
MojoKid writes "AMD has officially announced their triple-core Phenom multi-core processor offering, suggesting a triple-threat of processors, from dual-cores to triple-cores and native quad-cores coming to market this year. While the term symmetric multi-processing (or SMP) suggests a balanced approach of multiple cores in an even number of engines working together on a single workload, AMD offers that an odd number of processors can slice at that workload just as efficiently. Time will tell how this architecture will scale amongst various multi-threaded applications and real-world usage models. AMD is definitely moving to make use of these quad-cores that don't quite make the cut by testing them fully as triple-cores and realizing some revenue, rather than throwing them away."
For the cleanest, most comfortable shave ever! (Score:5, Funny)
Re: (Score:2, Offtopic)
Is it wrong that I'm thinking of building a water cooled laptop with 8 cores, a RAID5 with ~2TB of usable space, and a 24" monitor? I'm imaging a computer roughly the size of a large pizza box. Woot!
Re: (Score:2)
It isn't wrong unless you are thinking of taking it through airport security. In which case it would be hard to tell what could happen that wouldn't be expensive to you.
Re:For the cleanest, most comfortable shave ever! (Score:5, Funny)
Re: (Score:2)
Re:For the cleanest, most comfortable shave ever! (Score:5, Informative)
Re: (Score:3, Interesting)
Heck - I would love to see a Lisp-machine based on current technology. I could even help design one.
I thought Lisp machines were originally stack-based, which was hard to extract ILP from, which resulted in the death of the architecture. The last Lisp machines were Alphas with a pure software Lisp implementation. I'd be really interested in how you would plan on building a Lisp machine with current technology. Possibly go for simple in-order cores and make a Termite-machine, rather than a pure Lisp machine?
Re: (Score:2, Funny)
With apologies to the Onion (Score:5, Funny)
But you know what happened next? Shut up, I'm telling you what happened -- the bastards went to four cores. Now we're standing around with our cocks in our hands, selling three cores and a fan. Cooling or no, suddenly we're the chumps.
Well, fuck it. We're going to five cores.
Re:With apologies to the Onion (Score:5, Informative)
Mod parent up, please, and while you're doing that, read this:
http://www.theonion.com/content/node/33930 [theonion.com]
Re: (Score:3, Informative)
Re: (Score:3, Informative)
No, the thing that's awesome is that eighteen months after the Onion article was published, Gillette actually did it [cnn.com].
Re: (Score:3, Interesting)
Re: (Score:3, Funny)
Would someone tell me how this happened? Intel was the fucking vanguard of computing in this country. The Core Duo was the processor to own. Then the other guy came out with a three-core processor. Were we scared? Hell, no. Because we hit back with a little thing called the Core Trio. That's three cores and an fan. For cooling.
But you know what happened next? Shut up, I'm telling you what happened -- the bastards went to four cores. Now we're standing around with our cocks in our hands, selling three cores and a fan. Cooling or no, suddenly we're the chumps.
Well, fuck it. We're going to a beowulf cluster.
Don't buy yet (Score:5, Funny)
Re:Don't buy yet (Score:5, Funny)
Well, technically this triple core CPU does - in binary.
Business as usual (Score:2, Insightful)
Re: (Score:3, Insightful)
Re: (Score:2)
Re: (Score:2)
Re:Business as usual (Score:5, Funny)
Re:Business as usual (Score:5, Insightful)
Re: (Score:2, Interesting)
I wonder if you could build a whole computer from factory rejects.
It's been a while since I've done any PC building (switched to mac a few years ago), but I was always of the understanding that if you're not buying top-shelf components, there is a pretty good chance that it is a factory reject of a better component. This is especially true with CPU's, of which the slower processors are just faster processors that didn't make the cut and are systematically underclocked until they pass. Same thing with graphics cards, thats why you can usually buy a cheaper card and open
Nothing new here. (Score:5, Informative)
Why Yes. Yes it does. From HERE [wikipedia.org]:
Re: (Score:3, Interesting)
Re: (Score:2, Interesting)
It's still three cores on a single chip, and thus "triple-core". The architecture and functionality of an individual core doesn't matter, so long as it's capable of working as a single core (thus the PS3 is not an "8-core" or "7-core" system, since the Cell units are not functional as independent cor
Someone has been brainswashed (Score:5, Informative)
It's about multiple cores processing simultaneously. Check the article I link to, even the damn example diagram has 3 cpu-s.
Re:Someone has been brainswashed (Score:5, Informative)
However, AMD is definitely moving to make use of these quad-cores that don't quite make the cut, by testing them fully as triple-cores and realizing some revenue, rather than throwing them away.
The triple-core Phenom is an actual Phenom architecture, it's not 4-core rejects. Jesus Christ, NEVER accept submissions from hothardware.com anymore!
That's the worst one in months.
Re: (Score:2)
It's too late, the crowd here is already heading to a couple hundred snarky comments to "facts" that weren't facts to begin with.
I'm assuming this chip would use the same style layout as the chip on the Xbox 360, but you probably won't find that sort of thing out here.
A very real reason for using triple-core (Score:3, Interesting)
Without seeing further details I sincerely doubt that these are quad-core chips with one dud core. I suspect AMD has actually used their technical brains here and given us the fastest non-(overly)complex multi core setup.
Of course, if it's the bean counters in charge, then it's possible it's a failed quad core (though I s
Re:A very real reason for using triple-core (Score:5, Informative)
Firstly, for any general multi-node graph, it's entirely possible for three, four, eight, or any number of nodes to be only one hop away from each other. See fully-connected mesh [wikipedia.org]. For the four-node case, imagine a 2D square, connected on the four sides, plus two links connecting the "diagonals" of the square. In that topology, each of the four nodes are only one hop away from each other. Of course, as the number of nodes increases, the cost of fully connecting them increases, as does the processing cost to multiplex and process transactions into the node from the (n-1) incoming links, but with only four nodes it's entirely possible to create a fully-connected network.
Wiith AMD multi-core processors, all of the cores communicate using a fully-connected crossbar switch in the on-die northbridge - meaning all cores on the die are one "hop" away from each other, including the four-core case. What you're probably thinking of is a multi-socket system that only has two coherent links per socket - that would prevent you from making a fully-connected coherent interconnect for a 4-socket system.
Re: (Score:3, Informative)
Firstly, for any general multi-node graph, it's entirely possible for three, four, eight, or any number of nodes to be only one hop away from each other.
Firstly, we're not talking about any general multi-node graph.
We're talking about CPUs & AFAIK, the the traces can't cross one another..
Unless they commercialized some 3D process @ 65nm that I didn't read about. Wiith AMD multi-core processors, all of the cores communicate using a fully-connected crossbar switch in the on-die northbridge - meaning all cores on the die are one "hop" away from each other, including the four-core case. Sooo...
Cpu 1 --> hop --> northbridge --> hop --> CPU 4
Or a
Re: (Score:3, Informative)
We're talking about CPUs & AFAIK, the the traces can't cross one another.
There are around 10 metal layers in a modern IC. Traces can certainly cross one another on different layers.
Cpu 1 --> hop --> northbridge --> hop --> CPU 4 Or am I misunderstanding the definition of a "hop"?
I see what you're saying...but you're counting a processor interface as a hop. How many hops are there between cores in an Intel system then? CPU 1 --> hop --> front side bus --> hop --> CPU 4? No one counts hops that way. Your definition of hop would be like me saying there are two hops from your computer to your router, one between your processor and your network card over the
Re:A very real reason for using triple-core (Score:4, Informative)
Before you call me incorrect, please take 2 minutes to look at some lecture notes from an intro VLSI course:
http://www.cse.sc.edu/~jimdavis/Courses/2005-Fall%20CSCE%20613/CSCE613-Week10-Chapter-04-05.pdf [sc.edu]
You can clearly see on page 3 (slide 6) that metal1 and metal3 are directly on top of each other. As I stated in a different post, you're confusing metal layer/wire routing in an IC with entire logic devices (transistors/gates/flops). Let me repeat it again for you: metal layers in an IC can cross.
Re: (Score:2)
Re:Someone has been brainswashed (Score:4, Informative)
Re:Someone has been brainswashed (Score:5, Informative)
The symmetry in SMP does not refer to the capabilities of the processors. It refers to the relation between the processors and memory.
In symmetric multiprocessors, all processors access the same shared memory uniformly. That is, memory access delays depend neither on what memory zone nor from which processor it is being accessed.
In contrast, in NUMA architectures (non uniform memory access), each processor holds a portion of the shared memory that it can access very quickly. A processor can also access the portions of other processors but this incurs potentially large delays.
At the end of the spectrum, asymmetric multiprocessors combine processors with different capabilities. Here, asymmetric indeed most probably refers to the fact that processors are different. However, while most (all?) actual implementations using a NUMA architecture do use identical processors, they are never said to be symmetric because of the memory access.
Re: (Score:3, Informative)
The symmetry in SMP does not refer to the capabilities of the processors. It refers to the relation between the processors and memory.
The symmetry refers to all the capabilities of the processors, including their access to memory. SMP means that all processors in the system are interchangable from the OS's point of view - that cannot be the case if any characteristic varies between the processors. A system with different processors that have the same access to memory (such as an unexpanded Amiga 1000) is not considered an SMP system.
Re: (Score:2)
No, that just crappy Intel SMP. Not true for SMP in general.
Re:Someone has been brainswashed (Score:4, Informative)
Wikipedia [wikipedia.org] would disagree with you: "Symmetric multiprocessing, or SMP, is a multiprocessor computer architecture where two or more identical processors are connected to a single shared main memory."
SMP implies that there is a shared memory address space and that the cores can execute similar binaries. NUMA implies separate banks of memory dedicated to specific CPUs -- for example, AMD Opteron. However, most vendors still call the Opteron 'SMP' when used in a multi-CPU configuration because the "independent" banks of memory are mapped into the same memory address space (visible from all CPUs) and there is neglible penalty for executing tasks on either core regardless or location of code or data ***. (*** note: memory banks shouldn't be completely ignored for memory intensive high-performance computing applications and indeed on certain OS's like Vista, it is possible to allocate memory with CPU affinity or to schedule tasks with CPU affinity on an Opteron to alleviate NUMA crosstalk between the CPUs).
ASymmetric MultiProcessing (ASMP) implies dissimilarity in either the processing units (different binary opcodes) or disjoint memory accesses. Using a physics-accelerator or a generic-GPU programming with a main CPU is asymmetric processing even if the accelerator can access the same memory as the CPU (i.e. from cheap "shared-memory" GPU such as those integrated on cheap motherboards or to more powerful ones such as the GPU in the XBOX360). The CELL in the PS3 is not SMP because the PPU and SPU can not execute the same binaries and the cores are asimilar even though all cores have some method of accessing the main memory with a shared address space (although the SPUs also use a DMA read/write to main memory rather than direct access which would doubly qualify them as ASMP - but even without this memory difference, they would still be ASMP processing).
Re: (Score:2, Funny)
Re: (Score:2)
The S in SMP is symetry as in "several identical" parts anyway, not as in "a power of two".
A star with 5 equal arms, equally spaced, is perfectly symetrical.
Re: (Score:3, Funny)
Your data gets turned through 120 degrees as it moves from core to core, irritating, but better than having it come out sdrawkcab as sometimes happens with dual core processors.
Re: (Score:2)
Obviously in the case of three cores, it's not going to be literally "symmetrical".
He goes on to say "AMD offers that an odd number of processors can slice at that workload as well, just as efficiently", so it's not like he's spreading FUD or something...
Fourth Core Unlocking (Score:2, Interesting)
Re: (Score:2)
Re: (Score:2)
Re:Fourth Core Unlocking (Score:5, Interesting)
1. The core is there and locked off via microcode like the extra quads on a cut-down GPU (e.g. Radeon x1900GT vs. x1900XT) and can be enabled with a microcode flash.
2. The core is there but the fuses that connect it electrically to the rest of the die are blown, so it is there but not able to be enabled.
3. The core was never there as the die only has three cores in it in the first place- you have a fully-functional piece of silicon, so there is nothing extra to enable.
Either way, it's really long odds you'll get a free core enabled. Nobody has been able to even upward-unlock the K8's multiplier and I know for a fact that is set in microcode (some guys on ExtremeSystems got a JTAG and found that out but not how to change it.) They will probably use the same method they used to disable one core on a dual-core die and sell single-core Manchester and Toledo-die chips and AFAIK nobody has unlocked any of those. I bet they have a few of the X3s be X4s with a bad die, but the X4 is a darn big chip at nearly 300 mm^2 and the cost reduction by using a native 3-core die would be mighty attractive to them so I guess that most will be #3 then.
Single, double, triple, and quad (Score:4, Interesting)
This implies that they have a way to use all four cores independently. Maybe they can't; maybe one core is "special", like the "master" core that has to be working for anything to work. Also this implies that the cores can detect that their sibling(s) aren't working and switch to a mode in which the sibling is not used at all.
Also, a question - when a core doesn't function properly, is it somehow disabled completely so that it doesn't use any power? Or maybe a core that isn't being fed any instructions doesn't use any power anyway?
Re: (Score:2)
Rather than detecting the failure, they probably have a strap on the chip. It could be a simple wire in the package, or it could be on the surface where they can zap it with a laser to disable a core that didn't make the grade. IIRC that's what Intel did with the cache to make a celeron.
The strapping would give them more options in the case that one core bins slower than the other 3. Derate the chip or zap out the slow core.
It's a great way to raise their effective yield and keep prices down.
Re:Single, double, triple, and quad (Score:5, Informative)
The idea of post-factory error detection isn't so far-fetched. If a chip passes QA, the sorts of defects you'll see later in its life are likely to be thermally induced, and the likelihood that the defect will manifest prior to loading of the BIOS is very low. You're not using the MMU or the FPU at all, you're not using much of the cache, you can be running at your minimum power setting, and you're not doing it long enough to heat up much. If a core gets marked bad due to an excess of MCEs, similar to how many systems can mark DIMMs bad on excessive multi-bit ECC errors, the BIOS simply doesn't need to bring it online at boot time. Even if core 0 is the faulty one, you can probably load just enough of the BIOS to bring a good core online and finish booting, since you're not straining it enough to cause thermal problems, and you're only using a tiny fraction of the instruction set and die transistors. This sort of High Availability feature probably won't make it to the desktop right away, but as core counts keep increasing, it's inevitable.
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Not at the cost of supporting all the different SKUs. Supporting all the parts with the different test programs, piece parts, and managing inventories can result in less profit than just scrapping the silicon.
Re: (Score:2)
With single core processors, 90% work 10% don't.
duel core: 81% have both cores working, 18% have one working core, 1% have no working cores.
tri core: 73% 3 cores work 20% have 2 cores working 7% have one working, a tiny fraction have no cores
quad cores: 66% all cores work, 25% have 3 cores working 8% have 2 1% have 1, tiny fraction has no cores working.
Note though this is just an example. In RL theres probably defects that could affect multiple cores aro
Re: (Score:2)
Just a binned part? (Score:4, Informative)
And why should ``symmetric'' imply even? It merely implies that all cores see memory with the same class of service. And, in reality, aren't most AMD multiprocessors cc-NUMA machines, not SMP?
For most workloads, if they are fairly multithreadable, 3 processors available will be just fine. I know of very few workloads that require an even number of processors, and even if it were the case that the task were split into an even number of threads, the OS should have no problem scheduling on a reduced number of processors.
Hey, doesn't the XBox 360 have a 3-core PPC in it?
Re:Just a binned part? (Score:5, Informative)
This is what the article authors suggest, but no, it's a separate architecture. While I suspect it's possible a subset of the 4-core Phenoms to be relabelled as 3-core Phenoms, the bulk of 3-core Phenoms will be built as 3-core parts from the very start.
And, to add insult to injury, this is a quad-core Phenom on the picture, since it's all the authors of the fine article could find. In other words, they are idiots.
Re: (Score:2)
Manufacturing Yield vs. Marketing Perception (Score:2, Interesting)
Yet I can't help but won
Re: (Score:2)
Re: (Score:2)
Re:Manufacturing Yield vs. Marketing Perception (Score:4, Insightful)
Re: (Score:2)
Re: (Score:2)
So in this case, you might be buying a chip where 1 core only ran at 1.8 GHz, and the rest run at 2.4 GHz, so it's a 3x 2.4 GHz because it didn't cut it as a quad core part.
At any rate, it sure beats junking parts that are still very effective
Re: (Score:2)
Not necessarily, the cost to support additional parts may be more than just scrapping the silicon.
It costs just as much to manage a 4-core part as a 3-core but the 3-core would be sold at a lower price. Further, as yield on the 4-core improves you start running into issues supplying 3-cores run into cases where to support SKU demands you sell working 4-core parts as 3-cores and earning less for that good silicon.
Re: (Score:2)
Even? What the hell? (Score:4, Informative)
SMP Doesn't Suggest Even Numbers Of Processors (Score:5, Informative)
SMP refers to the fact that all the processors are identical and share the same memory (in contrast to NUMA designs like multi-chip Opteron systems). However, I've seen more and more people refering to cache coherent NUMA designs like multi-core opteron and the upcoming CSI based intel systems as SMP systems which, while a stretch of the definition, is at least reasonable.
Suggesting that SMP has anything to do with having an even number of processors is just DUMB. It may be the case that SMP systems usually have an even number of cores (I don't know) but that's not what the writeup or article seem to be saying.
SMP? (Score:3, Insightful)
The day is coming... (Score:2, Funny)
Magic number 3....simple when you know how! (Score:3, Funny)
AMD has a process which can put X number of transistors on a chip.
One of their cores needs Y transistors.
A qualified engineer with years of training in advanced mathematic divided X by Y and got the number "3".
So... the chip got three cores.
Mystery solved!
Which is why (Score:2)
*Triple* Core? Scoff. (Score:2)
Phenomx3 + GPU? (Score:2)
Is that part of their roadmap?
AMD's bad decisions.. (Score:2, Interesting)
This seems like almost as bad of a decision as Intel made going with Netburst. The difference being Intel could afford to make a big mistake. AMD doesn't have that luxury, and their new luxury might be going bankrupt. Sure, Intel went with Netburst because it solved several
486SX (Score:2)
Re: (Score:3, Interesting)
Re: (Score:3, Informative)
Quite good, perhaps, but for less money you can certainly get better performance out of Intel. As much as I have loved AMD for the last decade, Intel is completely eating their lunch at the moment and Phenom and Barcelona are not going to save them.
Really? I built an Athlon 64 X2 6000+ system a few weeks ago and the comparable Intel chips seemed a lot more expensive. The Core 2 Duo E6700 seems to perform about 5-15% better, but costs nearly twice as much ($320 vs. $170 at Newegg).
Re: (Score:2)
Re: (Score:2)
However, Intel sells them binned at low frequency mainly because there is a need for low cost processors, and they don't want to sell fast processors for cheap (they sell slower processors for cheap).
Amd owns the low-end. (Score:4, Insightful)
Also, you're absolutely right that we should hope AMD doesn't get gobbled up. The current Intel stuff, it seems to me, is a direct result of AMD dominating the price/performance ratios for so long, and even, recently, doing well with performance/watt. So even if you don't end up buying AMD, having them as a constant threat means Intel will be forced to compete.
Re:I wonder how many GHz these can crank out (Score:5, Insightful)
After all their stuff:
1) Actually works (and is reliable compared to other computer stuff - RAM, HDD, motherboards, etc)
2) Is cheap
3) Is available in sufficient quantities
4) Performs ok
Only prob is Intel is now significantly ahead of them in many areas.
That's what you get for being in a high tech commodity market where lots of buyers actually go by specs and price and not by covenience or brandname.
If AMD was number two in the orange juice, soda pop or cooking oil market with just 15% share they'd still be making money. And they could sell the same standard juice/soda/oil for years without investing billions in fabs and processes.
AMD has lots of smart people working for them.
It's better to be number 9 in good industry than number 2 in a crappy industry.
Kids, learn from this. That's why smart parents discourage you from trying to earn a living as a movie star or singer, the number #10000 star/singer in the world doesn't make as much as the number #10000 lawyer/doctor.
Re:2^n = 3, where n belongs to Z is not possible (Score:4, Informative)
Supposedly 3 core is actually pretty nice in some ways, as each core has a direct link to the other two. On a quad core system, each core is linked to two others, so sometimes it takes two hops to get messages from one core to other, slowing things down.
Re: (Score:2)
Re: (Score:3, Informative)
Re: (Score:2)
Not even that.. (Score:4, Informative)
All that said, SMP has nothing to do with an even number of processors/cores. It just means each processing element of a system is roughly equivalent. So you have a choice of three parts to schedule something on, the scheduler can know all three are equally capable and the heuristics for processor selection are straightforward. ASMP typically has specific roles for each part (i.e. a dedicated processor for interrupts, etc etc)
Re: (Score:3, Insightful)
Re: (Score:2)
Re:2^n = 3, where n belongs to Z is not possible (Score:4, Informative)
Another reason why powers of two are popular with multicore chips is that powers of two can be laid out into rectangles. If your multicore design is basically a copy-and-paste job with a little glue logic, it's a lot easier to lay out the cores. With something like the Cell, 8 is a nice number of cores since it allows you to have two rows of four. Three is just awkward.
The Cells found in the Playstation 3, however, did not have 8 SPU cores, they had 7. This is because most of the die space is the SPUs and you can dramatically increase yields if you only expect 7 of the 8 to work. If a single SPU has a manufacturing flaw, you just disable that one and sell pop the chip in a PS3. If none of them do, you sell it for more expensive blades.
AMD and Intel have been doing this for a while. Chips with flaws in the cache have some of the cache disabled and are sold more cheaply. In addition AMD chips are designed with three hypertransport controllers. If only one works, the chips are sold as Athlon 64s. If two work, they are cheap Opterons, if all three work, they are expensive Opterons (exactly how expensive depends on how many flaws there are in the cache area). Similarly, with the dual core lines flaws in one core result in them being marked down as single-core chips.
Intel, currently, sell quad core chips containing two separate dies. If either die has a flaw, it is sold as a Core Solo and not put in a dual-die package. AMD, however, are going to be making single-die quad-core chips. Selling three-core versions allows them to make use of the ones with a flaw in one core. This should help keep their yields high (and thus their costs relatively low), since it means that they can sell flawed chips almost irrespective of where the flaw is, just marking it down as a cheaper part.
No that rule only applies to fingers (Score:2)
Re: (Score:3, Funny)
Re: (Score:2)
It was used in a couple different Honda cars... (Score:3, Informative)
Next question please...
Re: (Score:3, Interesting)
Lister have been making 3 cylinder (and 2 cylinder, and 1 cylinder) diesels for years.
Re: (Score:3, Informative)
Re: (Score:2, Informative)
AMD's multi-core processors use a fully-connected crossbar switch in the on-die northbridge to communicate. There is only one "hop" between each core.
What you're thinking of is a four-socket system whose interconnect network is not fully-connected - it's only the edges of a square, and there are two missing links between the "corners" of the square. That is certainly a legitimate topology for a four-socket system, with the limitation you pointed out (two hops to get to the opposite node), but it doesn
Three dollar bill (Score:5, Funny)
Re: (Score:3, Funny)