backtop


Print 138 comment(s) - last by Wolfpup.. on May 11 at 4:40 PM


The Cell Broadband Engine is uber powerful at folding proteins
PlayStation 3's Cell Broadband Engine is better at folding proteins than the Xbox 360's triple-core CPU

The PlayStation 3 has recently flexed its floating point muscles by leading all processors at Folding@home. Sony’s game machine has been performing more work than any other system on helping to understanding protein folding, misfolding and related diseases such as Alzheimer’s and cancer.

A topic up for debate between the console circles is how the PlayStation 3’s arch-nemesis, Microsoft’s Xbox 360, would perform on Folding@home. While there is no Folding@home client for the Xbox 360, Vijay Pande, Associate Professor of Chemistry at Stanford University and Folding@home program lead, told Pro-G that the CPU inside Microsoft console would definitely lag behind the PS3’s Cell Broadband Engine.

“We are simulating key processes in protein folding and misfolding in Alzheimer's Disease. PS3's are performing aspects of these simulations, and doing so about 20 times faster than a typical PC,” Pande said. When asked if the Xbox 360 could be of use to the Folding@home program, he answered, “Possibly, although the cell processor in the PS3 is much more powerful for our calculations than the CPU in the Xbox 360.”

Mind you, the comparison may only be strictly between the two CPU architectures of each console. The Xbox 360 uses a custom triple-core PowerPC-based CPU, while the PlayStation 3 uses the Cell Broadband Engine composed of seven processor elements.

While the Xbox 360 CPU may not be tops in terms of Folding@home, its ATI-developed GPU could be much better. ATI Radeon GPUs running Folding@home are outperforming the PlayStation 3 on a per capita basis. Though whether or not the Xbox 360 will ever get its crack at helping to cure a disease remains to be seen.



Comments     Threshold


This article is over a month old, voting and posting comments is disabled

Does it matter?
By jabber on 5/4/2007 4:52:40 AM , Rating: 5
After all they are games consoles first.

Bit like saying the one cars cup holders can hold a bigger cup than another car. Its not really its primary purpose.

At the end of the day if they both end up with a Folding app then its good work all round!

Keep up the good fight.




RE: Does it matter?
By behemothzero on 5/4/07, Rating: -1
RE: Does it matter?
By behemothzero on 5/4/2007 5:21:47 AM , Rating: 2
Sorry, I meant "protein folding is not the primary reason they were made".


RE: Does it matter?
By h0kiez on 5/4/2007 8:32:00 AM , Rating: 5
Might matter to you. Games are what matter to me.


RE: Does it matter?
By caqde on 5/5/2007 1:14:11 AM , Rating: 3
I'll wait until the day the research effects you to see if it still doesn't matter. Games may matter to you now. But when you can't play the game or a close friend, relative, or loved one (wife/husband) is effected by one of the diseases that the research from folding at home could cure you will care. Mind you games matter to me too and I know they were meant to play games first, but this kind of research is important to a-lot of people and could also be important to any one of the people reading on dailytech now or if not at a later time in their lives.

Just think do you want Alzheimer's Disease?


RE: Does it matter?
By noxipoo on 5/9/2007 12:57:47 PM , Rating: 2
so i take it you bought a ps3 to fold and only fold for the fight against cancer?


RE: Does it matter?
By OblivionMage on 5/4/2007 3:51:42 PM , Rating: 3
The Xbox 360 has graphics that are either superior or equal to that of the PS3, more memory bandwidth, is much easier to program for, has 512mb of shared RAM, rather then 256 for video and 256 for processor. And is also much cheaper, in my opinion this thing about the PS3's being able to use Folding@home so well does not strike me as a very important reason for the average consumer/person.


RE: Does it matter?
By bjacobson on 5/8/2007 10:30:16 AM , Rating: 2
That's the problem Sony created for themselves. They designed a supercomputing chip to fill needs a general purpose CPU would be much better suited for.

If they had used 2 of the PPC cores (of which the Xbox360 has 3) AND had the SPEs...then the PS3 would truly be superior.


RE: Does it matter?
By DaveDee on 5/9/2007 8:24:53 AM , Rating: 2
I was under the impression, that the PS3 also has 512 shared memory.

256 XDR, 256 DDR3, and either the cell or the RTX can access either of them.

the 256 XDR is also considerably quicker than the 360 memory (by a factor of 10).


RE: Does it matter?
By Wolfpup on 5/10/2007 4:44:20 AM , Rating: 2
The PS3's GPU has full access to all 512MB, and because it's not unified, the PS3 has basically double the memory bandwidth of the 360 (though not 10x).

The CPU would just have that 256MB space, and I don't know if that's a real issue for Folding @ Home right now or not. On my PCs, each instance of F@H uses between 1-110MB, depending on the work unit. Most are probably around 10MB. But since the PS3 would just be using all of it's cores on one unit, I wouldn't think the memory demands would go much past 100MB, at worst (plus the extra RAM used for the graphics, etc.)


RE: Does it matter?
By Proteusza on 5/4/2007 5:41:10 AM , Rating: 4
What a crap article, with a sensationalist article and a load of conjecture.

Really bad reporting.

The guy says he THINKS the XBox 360 CPU will be slower, he doesnt actually know, and suddenly the headline is that the PS3 is more powerful. Load of crap.

Why dont they make a client, and then we can see?

Anyway, I happen to think with each of the 360's cores being more powerful than an SPE of the Cell, and each being multithreaded, the performance wont be very large in any one direction.

Between the suspiciously high 8800GTX benchmark figures in the 2900 XTX article and this, DT is really losing my support.


RE: Does it matter?
By ciparis on 5/4/2007 5:59:18 AM , Rating: 1
quote:
conjecture


Fact. But don't let that stand in your way (and by all means don't bother going to the F@H site for an explanation of why). And of course continue to ignore the relevance of the part after the "for" as it applies to suitability for gaming in today's market (which is the only reason I can think of for someone to get worked up over this article -- which I'm sure was the intent).


RE: Does it matter?
By Proteusza on 5/4/2007 6:19:41 AM , Rating: 4
How is it fact?

They have not run it, they can not prove it.

This has nothing to do with gaming, it has everything to do with bad reporting.

a better title would be "Engineer believes PS3 more powerful than 360 for FAH."

It is conjecture and only conjecture.


RE: Does it matter?
By Murst on 5/4/2007 10:11:52 AM , Rating: 1
You're right. There's no way to prove it... to you.

But, I'd also be willing to bet that a Ferrari would beat a Dodge Neon in a 1000 mile race. In fact, I'd consider it rather obvious, but I guess you would argue that unless the race between a Ferrari and a Dodge Neon was actually ran, it is impossible to make the statement that a Ferrari is faster than a Dodge Neon in a 1000 mile race.

However, humans have this abilty called deduction. They can look at similar, relevant factors and deduce from them that, in fact, a Ferrari would be able to beat a Dodge Neon in a 1000 mile race, even though that race was never ran. Its exactly the same with the PS3 and Xbox. It can be deduced from known factors (although not known to me, but I'm sure an expert in the field can do it) that, in fact, the cell processor is better at F@H than the PowerPC-like processor in the 360.

As to your question about "facts"... there are many things that we consider to be facts in the world that came about via deduction.


RE: Does it matter?
By OxBow on 5/4/2007 10:22:11 AM , Rating: 2
Are we dealing with "known known's, known unknowns, or unknown unknown'" ;)


RE: Does it matter?
By Murst on 5/4/2007 11:49:54 AM , Rating: 2
Its definetally a "known unknown" or an "unknown known". I'm learning towards the "unknown known" as that is the one which probably would lead to a solution through deduction or inference.


RE: Does it matter?
By Proteusza on 5/4/07, Rating: 0
RE: Does it matter?
By oab on 5/4/2007 11:35:22 AM , Rating: 5
Well, yes, there would be a way to prove it, have the F@H people spend a year and a half working with the MS people trying to tap into the raw performance of the X-Box 360GPU. I'm quite sure if you were willing to fund such extensive work, then they would be more then willing to port their software to the new system.

However, that is a completely unreasonable request.

The PS3 processor is more tailored to doing the types of calculations that work for the F@H people, more then the X-Box 360 processor. The 360 proc is a fairly standard general purpose processor. It has 3 cores, each capable of running two threads, for a total of 6 threads. But, it is general purpose, it is designed to run multiple different types of calculations as fast as possible, without degrading the performance of other types (i.e. it does fast integer and floating point calculations, in addition to the other types of things it does), while the PS3 does have one general purpose core, it also has those 7 specialized vector processors.

Vector processors were originally designed for scientific calculations (i.e. the Cray), not for general purpose calculations. They are designed to run operations on multiple bits of data at one time. Originally, your GP proc in your system was not. It has been added with SIMD support (i.e. SSE1/2/3/4, and AMDs 3D Now!+), but they arn't vector processors.

As it turns out, vector processing runs F@H super fast. It is also why GPUs perform calculations so much faster for the F@H people then regular CPUs do, videogames are highly dependent on vector processing (3d graphics), which is why everyone needs a 3d card in their system to be able to play the latest and greatest quake or unreal game and not have it run like a slideshow.

The article did point out that the GPU of the 360 would probably have very high F@H calculation rendering speed as well, certainly that is the case, because it is a GPU, by its nature it would run F@H calculations better then the CPU. Is it worth the F@H teams time to actually develop a 360GPU client? After all, its basically an x1x00 series, they already did that for the PC, and a GPU cannot perform all the calculations that the PS3 can. It can't perform explicit solvent equations, while the PS3 can, it can only perform implicit solvent equations.


RE: Does it matter?
By oab on 5/4/2007 11:35:26 AM , Rating: 2
Well, yes, there would be a way to prove it, have the F@H people spend a year and a half working with the MS people trying to tap into the raw performance of the X-Box 360GPU. I'm quite sure if you were willing to fund such extensive work, then they would be more then willing to port their software to the new system.

However, that is a completely unreasonable request.

The PS3 processor is more tailored to doing the types of calculations that work for the F@H people, more then the X-Box 360 processor. The 360 proc is a fairly standard general purpose processor. It has 3 cores, each capable of running two threads, for a total of 6 threads. But, it is general purpose, it is designed to run multiple different types of calculations as fast as possible, without degrading the performance of other types (i.e. it does fast integer and floating point calculations, in addition to the other types of things it does), while the PS3 does have one general purpose core, it also has those 7 specialized vector processors.

Vector processors were originally designed for scientific calculations (i.e. the Cray), not for general purpose calculations. They are designed to run operations on multiple bits of data at one time. Originally, your GP proc in your system was not. It has been added with SIMD support (i.e. SSE1/2/3/4, and AMDs 3D Now!+), but they arn't vector processors.

As it turns out, vector processing runs F@H super fast. It is also why GPUs perform calculations so much faster for the F@H people then regular CPUs do, videogames are highly dependent on vector processing (3d graphics), which is why everyone needs a 3d card in their system to be able to play the latest and greatest quake or unreal game and not have it run like a slideshow.

The article did point out that the GPU of the 360 would probably have very high F@H calculation rendering speed as well, certainly that is the case, because it is a GPU, by its nature it would run F@H calculations better then the CPU. Is it worth the F@H teams time to actually develop a 360GPU client? After all, its basically an x1x00 series, they already did that for the PC, and a GPU cannot perform all the calculations that the PS3 can. It can't perform explicit solvent equations, while the PS3 can, it can only perform implicit solvent equations.


RE: Does it matter?
By oab on 5/4/2007 11:36:25 AM , Rating: 2
accidental double post, sorry. ignore this one.


RE: Does it matter?
By Proteusza on 5/4/2007 11:40:45 AM , Rating: 2
Someone posted a link at the bottom of this page that explains things well, in terms of the Cell vs a GPU etc.

In any case, considering the fact that the 360 has such a large install base, and that it has a powerful general purpose cpu coupled with a powerful gpu, dont you think that would make a better choice for a F@H client? I mean, okay so the cell may be faster, but with 3 XBox's to a PS3, and the GPU of the 360, dont you think if they were really concerned about the total performance rather than the per client performance they would have gone after the 360?

Regardless of which is actually better in F@H, no matter if GPU's are brought to the table or not, I still think this article is terrible because of its misleading and attention grabbing title.


RE: Does it matter?
By adt6247 on 5/4/2007 2:18:22 PM , Rating: 2
Both still have relatively small install bases, and though the PS3 is off to a slow start, and I predict it will finish in last place, the gap will be narrowed.

Quite frankly, the CPU of the 360 is pretty lousy for this type of workload; it doesn't have much FPU horsepower, its cores are general-purpose in-order cores, and have no vector functionality. A GPU client would be the only thing that makes sense.


RE: Does it matter?
By Alexvrb on 5/4/2007 8:59:06 PM , Rating: 2
Vector? What about the VMX units? Cell is also in-order.

Xenon may not be as fast, but it can do more types of work.
Xenos can't do as many kinds of calculations, but it is certainly faster.

The difference is, the 360 has BOTH of those chips. If they were given huge sums of cash and enough time, I'm sure they could put together a pretty impressive FAH client.


RE: Does it matter?
By Xeeros on 5/6/2007 10:51:10 PM , Rating: 2
also keep in mind that Microsft advises a max of 6 hours of gaming and no offense but the air coming out of the back of a 360 after a long game session could cook an egg(not really but it is hot) and F@H if it were on 360 we could hypothesize that the thermals would be in-line with the thermals of the unit during game play since F@H uses all available cycles. Infact my only reason for using the F@H on my PS3 is that after a constant month of running the unit it still is whisper quiet and not at all hot in the back of the system. Now when they shrink the cores and reduce the thermals (idk maybe an actually cooling solution and not a PC cooler in a console) then im totally fine with running both 24/7 with the exception of when I want to do something.

But yes if you wanna get into a pissing contest I think that the 360's GPU might have an edge on the Cell Broadband engine as far as scores are concerned but in the end they would both be curing disease and benefiting human kind so really does it matter at all?


RE: Does it matter?
By Munkles on 5/4/2007 2:52:03 PM , Rating: 2
More so than just the complexity,

The cars will only preform as the driver instructs thus, If you put a "slow" driver against a "race-car" driver the race car driver in the neon just may win; furthermore, because the driver of the ferrari may not be familiar with how to drive/handle such a machine they may crash it or blow their engine out.

Likewise these processors are dependent on the software to be properly coded and optomized to take advantage of their architecture.

In general I will readily admit that the cell had more raw potential but to flat out profess CELL IS FASTER on no concrete evidence on a tech site is nuts.


RE: Does it matter?
By Murst on 5/4/2007 3:58:08 PM , Rating: 3
That's rather silly. We're comparing F@H, nothing else. Its rather well known that the strengths of the cell processor are exactly the features that F@H demands.

I doubt anyone here has claimed that cell is, in broad terms, faster than the 360 cpu. In fact, just as with any competing technology, you will find benchmarks which will put either side at the top.

However, we are talking about F@H, and the comparison is only between the cell and the 360 CPU (not the gpu!). And I think that claiming that the 360 CPU is faster or equal to cell, in terms of F@H, is fanboism at its finest.

Again, this has nothing to do with what processor is better for playing games or running any other software (and in fact, I don't even care, I don't own either system).

If you look back on DT, there was a post about F@H and some numbers... although the 360 cpu was not in there (it was, however, dominated by GPUs). This was a comparison of OSs, but behind every OS is a cpu. Unless you would claim that the 360 cpu is also much better than top processors of the current generation for the desktops at F@H, you'd easily arrive at the conclusion that the cell is faster at F@H than the 360 cpu.


RE: Does it matter?
By Murst on 5/4/2007 3:58:35 PM , Rating: 2
RE: Does it matter?
By geeg on 5/4/2007 2:11:14 PM , Rating: 2
the banner should be green


RE: Does it matter?
By nerdboy on 5/4/2007 6:42:47 PM , Rating: 2
Depends on how many times the ferrari has to stop for gas.


RE: Does it matter?
By theapparition on 5/5/2007 8:22:17 AM , Rating: 2
Very flawed analogy.
With some Ferrari's getting 6-8mpg, the time required to fill up on a 1000mile race may be signifigant. Reminds me of "The tortoise and the Hare" scenerio. Quick, but one has to rest, while the other is slow and steady.
You'd be better off saying 1/4mile race.

The second reason it's flawed, is that we have real world testing data for each. We've test the hypothetical Ferrari, and we've tested that Dodge. We don't have to make a deduction, since real test data exists. That is something we don't have with these consoles.

On to the Cell vs. PowerPc debate, I do think the PS3 will be superior to the Xbox360 in the F@H client. The cell is basically 7 DSP's coupled to a weak management processor, while the Xeon is three powerfull general purpose multithreaded processors. General purpose processors may not be the best for handling massive amounts of floating point calculations, something the SPE's are fantastic at. The SPE's, on the other hand are very inefficient at handling general purpose instuctions. So, for gaming performance, things are close, but I'd guess that the PS3 would have an edge at F@H.

Still, make a client for XBox. Since there are signifigantly more Xbox's in use than PS3, the small performance gap could easily be eaten up by the larger number of systems.


RE: Does it matter?
By Locutus465 on 5/6/2007 7:56:07 PM , Rating: 2
More accurate would be Ferrari v. Lamborgini.... Or at least Corvett/Viper... Lets face it, both consolse are plenty fast.... I don't doubt that it might be true that the PS3 is faster that 360 for this, and if the FAH people by a bunch to run the app great... But as far as console wars go, this isn't any more than a moral victory.


RE: Does it matter?
By peternelson on 5/10/2007 7:34:37 AM , Rating: 2
This guy is one of the leaders of F@H project, he should know.

They know their source code and the libraries it uses like Gromacs.

They know what instructions are used to perform the calcs.

They know how many instructions are completed per cycle by x360 versus Cell.

They know what parts of their code are repeated loops or rarely taken branches.

They can simulate the performance using standard computer science techniques, without ever writing (and certainly not releasing) a client for x360.

Therefore don't minimise the value of his opinion.

On the other hand I'd very much like to see them focus attention on a Nvidia gpu client, particularly with some of the newer nvidia products coming out later this year and their architecture for gpu based processing. At the moment F@H are missing out on about half the world gpu power by only supporting ATI.

X360 could contribute, but many do run hot and I wouldn't like to be responsible for failures and fires from people running folding on x360 24x7 in summer. After the die shrink it may be possible but I think he is right the contribution won't catch ps3.


RE: Does it matter?
By Wolfpup on 5/11/2007 4:37:42 PM , Rating: 2
Yeah, I'm kind of expecting either a Direct X 10 client that just runs on any DX 10 hardware, or something specifically for Nvidia's architecture. I can't wait to see how it performs. ATi's old architecture blows away Cell, so I can't imagine how fast an 8800 is going to be! :)


RE: Does it matter?
By Lonyo on 5/4/07, Rating: 0
RE: Does it matter?
By someguy123 on 5/4/2007 6:42:03 AM , Rating: 1
no...its actually much more like the PC than the ps3. the 360 is 3 PPE cores while the ps3 has a PPE and 7 SPEs. the 360 is much more comparable to a "tricore" desktop cpu than to the cell processor.


RE: Does it matter?
By Proteusza on 5/4/2007 6:50:05 AM , Rating: 1
If having three of the same cores makes it similar to a PC, then I suppose maybe.

But in actual fact they use similar instruction sets, and are both in order cores. PC's use vastly different instruction sets and architecture, and are very complex out of order cores typically with good superscalar execution.


RE: Does it matter?
By jrb531 on 5/4/2007 10:57:49 AM , Rating: 4
The Video Chip in the 360 outperforms the PS3 in folding but, of course, they are comparing "CPU's" so this is nothing but apples to oranges.

Pure BS!

-JB


RE: Does it matter?
By GlassHouse69 on 5/4/07, Rating: -1
RE: Does it matter?
By the truth on 5/4/2007 2:21:09 PM , Rating: 3
we can tell by what u say you are bias to MS. if you read capability of the CPS's u would say that they say Cell is 215 giga-flops and xenon is 115.2 giga flops. If u went to folding's web site you would understand that even though a gpu is very fast at folding but they can use it only for limit types of folding. If u look at folding u will see that there are 82% more PC cpu folding then PS3 but ps3 is giving 68% more computing power. And i like the way u think let use millions of dollars to figure out that the PS3 is better at what we need to do. If you ever run a company remind me never to support u with you business sense it will last maybe a week. Let use you logic and not looking at any referencing when i write a comment. the xbox 360 is slow.


RE: Does it matter?
By Angelus897 on 5/4/2007 4:05:43 PM , Rating: 1
Wow, that means a lot of people are using PS3s for folding.... Eh, I don't blame them.


RE: Does it matter?
By the truth on 5/4/2007 7:14:27 PM , Rating: 2
sorry about the spelling errors kinda write that fast with out proof reading. But if you looking at the folding website they have stats on how many of each processor are being used. Latest numbers as of 5/4/07. And if somepeople would check it out they would see that person making a quote could make a educated guess base off them having data on powerpc processors. Since the xenon processor is very similar to the G5 processor in Macs
OS Type__________|Current TFLOPS| Active CPUs| Total CPUs
Windows__________| 184 | 193674 | 1686765
Mac OS X/PowerPC__| 9__ | 4808__ | 12344
Linux____________| 46_ | 26966_ | 224553
GPU_____________| 57_ | 963___ | 3061
PLAYSTATION®3___| 567 | 33136_ | 112244
Total____________| 878 | 270427 | 2137774


RE: Does it matter?
By the truth on 5/4/2007 8:29:45 PM , Rating: 2
sorry about the spelling errors kinda write that fast with out proof reading. But if you looking at the folding website they have stats on how many of each processor are being used. Latest numbers as of 5/4/07. And if somepeople would check it out they would see that person making a quote could make a educated guess base off them having data on powerpc processors. Since the xenon processor is very similar to the G5 processor in Macs
OS Type__________|Current TFLOPS| Active CPUs| Total CPUs
Windows__________| 184 | 193674 | 1686765
Mac OS X/PowerPC__| 9__ | 4808__ | 12344
Linux____________| 46_ | 26966_ | 224553
GPU_____________| 57_ | 963___ | 3061
PLAYSTATION®3___| 567 | 33136_ | 112244
Total____________| 878 | 270427 | 2137774


RE: Does it matter?
By RyanHirst on 5/4/2007 4:15:31 PM , Rating: 2
Proteusza: there is no one more knowledgeable about F@H coding than Vijay Pande.
I agree that this fact isn't really "news," but PandeLabs has hand-coded clients for x86, x86 + MMX, x86 +SSE/SSE2, x64, DirectX9, PPC and Cell. Do note that PPC is on that list.

Now, different clients perform different calculations. An ATI R80 GPU can perform calculations an order of magnitude faster than a general-purpose CPU, but only on highly parralelizable calculations that can be described by DirectX9.

The point is: The Xbox360 is of interest, and potentially a competitor for the PS3, but NOT because of the Xenon processor (which is why the article is misleading. The Xbox 360 is of interest, but because of the GPU - which is a 48 unified shader architecture like R580). It is exactly the PS3's specialized architecture-- the SPEs-- that provide such rapid calculations for F@H. Its massive THEORETICAL output lies directly in line with the F@H client's ACTUAL demands. You can feel it all day long, but for the code Pande uses, each of the Xbox 360's cores are NOT more powerful than an SPE.

To get an idea of how specialized the F@H code is: My new 2.13Ghz 2MB cache C2D, at stock speed, calculates between 2-4 TIMES as fast as on my Dual Opteron 2.2Ghz rig.

You imply that Vijay Pande knows no more about the real-world implications of his own code than you or I learn from browsing a quote.

Your doubt does not oblige PandeLabs to make a client so YOU can "see"; it is not a test of whether or not sufficient data has been gathered to draw the stated conclusion. The only problem with the news quote is that space was not dedicated to asking the question, "why is that so?", and reporting the answer. But that's a hard complaint to level at a site that collects and posts brief headlines, abstracts, and summaries.

A similar complaint arose when F@H did not release a native 64-bit windows client. PandeLabs had stated that there was no real performance benifit to porting their code to 64-bit; meanwhile, the existing code ran fine on the 64-bit OS. People who knew NOTHING about the subject at hand jumped all over the quote, with statements like, "how can he possibly say that scientific calculation will not benifit from 64-bit code, when that's exactly what 64-bit code helps....." Well, the FP unit on any current cpu is ALREADY capable of receiving and executing double-precision 64-bit FP operations, from 32-bit code. The F@H client already used the highly optimized FP code (e.g. SSE) for super-32-bit calculations. Porting to a 64-bit OS would have only aided 64-bit integer calculations, a tiny fraction of the F@H calculations.


RE: Does it matter?
By fijillian on 5/4/2007 11:11:39 AM , Rating: 1
LOL. Gaming console 1st then maybe i will think about folding 2nd. if time away from gaming can be found.


RE: Does it matter?
By panda10 on 5/4/2007 1:12:56 PM , Rating: 3
Of course it matters. If anyone buys a car without bringing a 32oz big gulp with them- well they're just in for a world of hurt later. I mean there's nothing worse than driving with one hand and holding a 2lbs soda in the other for more than 5 minutes.

I wish I knew about this before I purchased my PS3 a few months ago- it would have made the decision much easier. Thank goodness my initial premonition was correct- and now I can fold proteins in style.


Of course it is
By ciparis on 5/4/2007 5:52:56 AM , Rating: 2
This isn't really news. Or at least it shouldn't be, not to anyone who has been paying a little bit of attention. It's exactly the sort of thing Cell is ideal for, and yes, it is MUCH more suitable for these calculations than the 360's CPU is. This is no surprise.

It's useful I suppose in making a headline sure to grab denials from clueless partisans unaware of what's being argued, though, which in turn is sure to incite responses from the other side.




RE: Of course it is
By cochy on 5/4/2007 6:02:24 AM , Rating: 2
I agree this is not news. There's no F@H client for the 360 and no one has apparently run F@H on the 360 CPU. They just are guessing it would be slower, which it probably would be. So why is this on Daily Tech is the question? Gratuitous plug for the PS3?


RE: Of course it is
By Proteusza on 5/4/2007 6:05:12 AM , Rating: 2
I suppose the Cell would probably be faster, being able to execute 8 threads at a time. Although the 360 can execute 6 at a time, hardly a slouch. It may come down to things like IPC and cache access time.

Anyway, yes this is non news and a very bad article by DT shamelessly plugging cell.


RE: Of course it is
By mars777 on 5/4/2007 10:53:40 AM , Rating: 2
You are generalizing too much.
Let me explain:

quote:
I suppose the Cell would probably be faster, being able to execute 8 threads at a time. Although the 360 can execute 6 at a time, hardly a slouch.


Xenon has "hyperthreading". That means it can execute two threads per core. Assuming the two threads are raw FPU calculations (as this is what F@H is). Barely a single thread of those two will fill the FPU of Xenon making the other wait till the FPU is free. Just a single thread stresses the core more that enough on the FPU. If the other thread was doing some integer math then yes it would be some improvement, but this is another application, not F@H. You can count the Xenon as a 3 core FPU processor (since it is RISC and doesn't have vector instructions (SIMD)). Nevertheless you have cache coherency overhead while doing it this way.

On the other way on the Cell you have a different situation:
You have a general CPU wich takes care if 7 other vector units have its OWN instruction cache filled. The SPE units operate at vector operations which means they can execute more than 1 FPU operation per cicle and have it's own data cache to store results. And can execute it's instructions in out-of-order mode. Then the general cpu (PPE) collects SPE cache and in it's own and gets the results of more that 7 (at least 14 according to my opinion) FPU operations per clock.

So if we take into account that we are not talking about number of threads but FPU operations:

Xenox 3 vs Cell > 14

This is all according to me, but as an advanced Assembly programmer I think what i have written could be generally correct.


RE: Of course it is
By mars777 on 5/4/2007 10:59:40 AM , Rating: 2
Conclusion:

Number of threads means nothing if there are no MATH execution units to be utilized.

It like flooding a single core procesoor with more and more threads when it is at 100% utilization.


RE: Of course it is
By Proteusza on 5/4/2007 11:01:52 AM , Rating: 2
Actually the Xenon does have Vector instructions, and both cores are in order. Check out http://www.anandtech.com/showdoc.aspx?i=2453&p=1

So we have two in order cores, with SIMD instructions. How can you be sure that only one FPU will be used at a time in each core? Do we even know how many ALUs and FPUs each core has?


RE: Of course it is
By mars777 on 5/4/2007 11:41:56 AM , Rating: 2
I've read the article you gave.
Now from what i understand these instructions can help but you must compile specifically for them. Like writing code for the SPEs.

But what i didnt' know is that all SPE and PPE in both processors are in-order.
This gives more performance hit on the Xenon. This even means that instruction level parallelism is unarchievable on Xenon so the Hyperthreading got in there only to fill FPUs and ALUs which were unutilized.

So Xenon is 3 if using 6 threads and is 1.5 if using 3 threads and is 0.5 if using 1 thread.

So maybe it is (assuming F@H has a lot of threads)

Xenon 4 vs Cell > 12

Still more that 6 vs 8 :P


RE: Of course it is
By Proteusza on 5/4/2007 11:50:42 AM , Rating: 2
quote:
This even means that instruction level parallelism is unarchievable on Xenon so the Hyperthreading got in there only to fill FPUs and ALUs which were unutilized.


Not really. All you need is a good compiler. They are all 2 issue super scalar cores, so you can execute two instructions at a time as long as they dont have any dependencies.

Although I realize compiler optimization isnt what it should be, and being in order its probably a good thing they arent more than 2 issue - any more would see that power wasted. In that case you are probably right about the FPU utilization.

Sidenote: I always wonder how powerful processors would be if Intel had one RISC instead of CISC. Imagine a 4 issue out of order RISC core with OoO exec, good branch prediction, and an int pipeline of length 8-16!

Sidenote 2: big up to you for actually reading the article and arguing intelligently, a lot of people here couldnt be bothered to read anything that doesnt agree with their way of thinking.


RE: Of course it is
By Wolfpup on 5/10/2007 4:39:40 AM , Rating: 2
There's apparently two VMX units per core, but each handles separate types of instructions. There might still be a speed up because of "Hyper Threading", but still, Cell's 7 SIMD units are going to beat out the 3 in Xenon (except for whatever reason only 4 SPEs are actually being used for Folding at Home...maybe some of the others are processing the graphics? If so, it's too bad you can't shut the graphics off and have 6 or 7 cores running F@H...)


parralellism
By bubbacub616 on 5/4/2007 5:31:33 AM , Rating: 3
As I understand (well to be honest I don't really!) folding@home computations work optimally on a parallel architecture. The larger the number of cores the better. So makes me wonder that if they managed to hook it up to a modern GPU (i.e. not an out of date one like in the new consoles!) you should be able to get amazing folding results (with 128 stream processors etc.) What they should really compare is the net number crunching ability of the whole console - CPU + GPU - which would let us know console is more "powerful" letting console owners know how big their e-penis really is - which is the point behind all these articles on the PS3/360's 'supercomputer' special powers!

if I have made a fundamental error in my assumptions please ignore




RE: parralellism
By PlasmaBomb on 5/4/2007 5:40:21 AM , Rating: 5
The X19xx series of GPUs from ATI are excellent folders, although they aren't quite as general purpose as a cpu (therefore they don't fold the same units). The GPU in the 360 should be fast enough to run with the PS3, whereas its GPU is not capable of running FAH simulations.


RE: parralellism
By bpurkapi on 5/4/2007 5:52:21 AM , Rating: 3
the original design of the cell chip was to configure multiple chips to become a super broadband linked computer. The reason other than games why cell is important is because sony can take your spare processor cycles and loan them to the highest bidder. This was the basis of the cell architecture, to be able to be a broadband linked super computer which sony could manipulate for various purposes. It plays games but the full repercussions of such a system will point towards a sony controlled super computer. And of course mass licensing of the blue ray media standard. Folding at home is really just marketing for medical fields which are looking for upgrades. The cell was supossed to enter the mainstream eventually and specifically the medical fields so the investment does not stop and start at consoles but goes towards the medical fields.
(I blame any lack of clarity on alcohol)


RE: parralellism
By bkm32 on 5/4/2007 10:16:22 AM , Rating: 2
Huh, huh, huh! He said, "e-penis". Huh, huh, huh!


misleading headline
By phusg on 5/4/2007 5:21:34 AM , Rating: 2
You could just as well have gone with the headline, "XBox more powerful than PS3 for Folding@home" with the sub-heading, "just waiting on Microsoft for support". Which at least is less misleading than the current one.




RE: misleading headline
By Proteusza on 5/4/2007 5:45:50 AM , Rating: 3
They could have said "Wanted: Microsoft Sponsorship. Will plug your products for food"


RE: misleading headline
By phusg on 5/4/2007 7:17:17 AM , Rating: 2
Exactly. Actually, with hindsight I'm not too displeased with this headline. Hopefully someone Microsoft has seen all this publicity going to Sony and finally decided to get their a*se into gear and release folding@home support for the XBOX. It would be a great publicity stunt for them if the XBOX was faster than the PS3, which theoretically it should be (although for less types of work unit).


RE: misleading headline
By jabber on 5/4/2007 7:28:49 AM , Rating: 3
While they are at it they could put some effort into giving us some more useful content on the European version of Xbox Live too.

Had the same old media rubbish on there for nearly a year.


RE: misleading headline
By bkm32 on 5/4/2007 3:29:54 PM , Rating: 2
Now, imagine an X360 client for F@H. It works wonderful. All you have to do is keep your X360 "on" for an extra amount of time. Now, imagine the service life of your X360 declining exponentially and the RROD being a staple of your next-gen gaming experience.

Until MS solves the X360s overheating problems, this is a non-issue. No X360 owner in their right mind is going keep their box "on" any longer than they have to due to the overheating issues of the X360.

Sorry, dudes. I love the X360 as much as the next guy (maybe not as much as that one dude at the top), but I wouldn't subject my machine (or my gaming experience) to such pain.

I say let the PS3 owners run down the lifecycle of their $600 investment. It'll produce a nice headline: "F@H actually folds PS3s worldwide"


Shock Horror
By Zurtex on 5/4/2007 6:38:24 AM , Rating: 2
A Cell CPU is very good at to doing lots of mathematical parallel operations, basically what it was designed for, versus a general CPU that was designed to be well balanced given any type of operation it may have to deal with.

And as the author points out anyway, GPUs are well known to out perform any kind of CPU in this type of work, I would be willing to bet that the PS3 GPU could out perform the PS3 Cell CPU, also that the 360 GPU could outperform the PS3 Cell CPU.




RE: Shock Horror
By PrinceGaz on 5/4/2007 8:41:51 AM , Rating: 5
The PS3 GPU is based on the GeForce 7 design which has been found to be quite poor at GPGPU work. It would probably be slower then the PS3's Cell CPU.

On the other hand, the XB360's Xenos GPU is based on the Radeon X1900 which has already been proven to be excellent at running F@H.

The article is correct when it says the PS3's Cell processor is faster at F@H than the XB360's Xenon processor, but it would almost certainly be slower than the XB360's Xenos GPU which is what the XB360 client will probably use (if Microsoft allow the client to be created). Sony should be careful about over-promoting the power of the PS3 through the F@H client, because it will seriously backfire if the XB360 F@H client is considerably faster (which it probably will be provided it uses the Xenos rather than the Xenon for the number-crunching).


RE: Shock Horror
By Zurtex on 5/4/2007 9:03:42 AM , Rating: 2
Actually I disagree, the GeForce 7 design is often under rated in GPGPU applications because the lack of real effort in to solving many of the problems attributed to it and its interactions with high level GPGPU languages.

I happen to know a bit about this, as my house mate did his masters in physics project on programming particle physics engines on to a GeForce 7900GT, I helped him from time to time a little with the maths involved. A lot of the problems found with the GeForce 7 series has to do with the Kernal of whatever language your working in and can be adapted to actually work quite well.

Currently they have it working dozens of times faster than the best of Core2Duo CPUs in specific applications on sheer processing capabilities.


RE: Shock Horror
By bkm32 on 5/7/2007 1:31:24 PM , Rating: 2
Awesome, unbiased post, dude. Kudos


Nice bought and paid for article Sony!
By jrb531 on 5/4/2007 10:55:25 AM , Rating: 3
Damn! I guess I'll buy a PS3 now because it can "fold" better when I'm not playing. ROTFLMAO!

What an utter POS propaganda article!

These are game machines. Care to see how the original Xbox would "fold" vs the PS2? If the xbox "folded" better would people have stopped buying the PS2?

What a bunch of crap. They do not even have a program to "fold" on the 360 and yes the 360 uses an advanced ATI video chip that outperforms the PS3 but noooooo lets come out with some BS article to try and sell PS3's because of protein folding???

Is Sony that desparate that they have to pay for this type of article?

Going to cool down now.... I do not give a rat's backside how the PS3 or 360 "folds" I care about how they play games and the fact is that the 360 gives a similar game experience for alot less money... unless, of course, you fine the need for a Blue Ray Movie Player or have a $2000 television.

-JB




RE: Nice bought and paid for article Sony!
By mars777 on 5/4/2007 11:03:51 AM , Rating: 2
quote:
Is Sony that desparate that they have to pay for this type of article?


You are assuming DT staff is getting paid to post pro Sony?

This article has a sensationalistic title (and i cannot approve that) but what you are assuming is very bad.

You should be rated down to hell or prove your claims.


By jrb531 on 5/4/2007 11:18:15 AM , Rating: 2
So the person who wrote this article should have to prove their claims?

LOL

-JB


By jrb531 on 5/4/2007 11:23:46 AM , Rating: 2
BTW Suggesting that Sony "fed" this article has nothing to do with suggesting that DT had anything to do with it... unless they wrote it and if this is true then they need to retract this POS article!

-JB


By Hawkido on 5/4/2007 11:36:33 AM , Rating: 2
It's not a bought and paid for article. Sorry.

It is MS's fualt they don't have a folding client. The Stanford Professor is probably taunting MS into making one.

As to the Xbox360's GPU being better, No. Faster? Yes.

Read the F@H Faq on folding. The GPU's are insanly fast. However they can only fold the simplest protein chains. The PC's can fold the far more complex ones, but at such a slow rate it is frustrating. The Cell-BE can fold the middle ground protein chains at ludicrous speed. Thus allowing the PC's to focus on the tougher chains, and the GPU's can focus on the smaller chains.

As to the Xbox being better or faster at folding, I would say this. The 3xPPC cores in the Xbox360 are probably capable of crunching the tougher chains much like the PC. No Flashy numbers to show off, just progress in the very sluggish end of the research, thus MS doesn't want bad PR and won't release the F@H client they developed, because they can't get the modified x1900 GPU to process small chains without exposing their platform to being hijacked for other purposes (new webserver anyone?).

MS doesn't have a nifty OS platform for developers to use to develop software other than games. They did this because of what happened to the xbox (XlinuX Server, I mean). The PS3 has a Linux OS for it, and thus anything that can be made for Linux can be ported to the Cell platform without Sony's intervention. So, Sony has made then more independant-developer friendly platform. On the MS side if it doesn't go throught the DirectX interface then it doesn't happen on the Xbox360. Thus, it is nothing more than a game machine, that can play movies (through the DX interface).


Facts and Agendas
By pareklund on 5/4/2007 6:50:24 AM , Rating: 3
Well, the article is certainly not a poster child of correctness. Here's my take on it:

* At present, the PS3 is infinitely better than X360 at F@H protein folding, since there is no F@H client for the latter.

* The Cell processor is theoretically a lot more powerful in terms of F@H protein folding than the triple core Power chip in X360. This, due to the fact that the Cell has 7 SPU:s that are designed for SIMD/vector processing, which greatly accelerates F@H calculations. I.e. the most important factor is not the number of cores, but the fact that the SPU:s in Cell will beat the SIMD capabilities of the PowerPC Altivec instructions hands-down. Otherwise, there wouldn't be a reason for their existence, since the latter is already present in the Cell PowerPC CPU.

* Conversely, there is a F@H client for ATI GPU:s, a variant of which is found in the X360, whereas, AFAIK, there is none for nVidia GPU:s. The reason for this, according to prof. Pande, is that the shader architecture of ATI is superior to nVidias (at least in the generation preceding nVidia G80) for the needed SIMD calculations. Also, given the fact that ATI GPU:s are superior to Cell processors chip-by-chip in terms of F@H calculations, an X360 running a hypothetical F@H client would probably beat the existing PS3 F@H client.

So, why then the angle of the article (apart from the fact that an attempt to correctness seldom makes for a good article heading ;)?

My guesses:

* DT wants more readers and posters to sell more ads.
* Sony obviously wants more positive press.
* Prof. Pande probably wants Microsoft to open up their network to make an X360 useful. Even though the X360 GPU should be well suited to F@H, it will be pretty useless, if Microsoft won't allow access to F@H servers.

/P




RE: Facts and Agendas
By DingieM on 5/4/2007 9:14:33 AM , Rating: 3
Don't forget to mention that the Xenon incorporates heavily modified altivec engines and are considerable faster than the "standard" altivec engines found in the PowerPC G5, for example.

F@H could choose to let the Xenon (CPU) and the Xenos (GPU) work together due to the balanced integration of those two.
Since Xenos has memory export functionality and can read and write directly to the cache.
This would put the PS3's F@H performance to shame.

Would I be running F@H at my XB360?? Hell no I won't leave the machine on for days, weeks, months and wait for a really big energy bill.

We humans are destroying our own planet, consuming waaay too much energy and then we should leave our computers running constantly for F@H???


RE: Facts and Agendas
By KevinW on 5/4/2007 11:05:16 AM , Rating: 2
The Sony demos of Cell processors I have seen suggest that the bulk of the FLOPS available in the PS3 Cell processor come from the RSX graphics engine and not from the SPUs something like a 12:1 ratio for the GPU to all the other processors combined. I would imagine that is where the folding client gets the bang for the buck.


Gpu
By scrapsma54 on 5/4/2007 1:08:49 PM , Rating: 3
So great, ps3 has got a powerful cpu, but so did ps2. Its the same crap as last time. Now if ps3 can tout its gpu power, then ok sony did something right.




RE: Gpu
By scrapsma54 on 5/10/2007 8:34:08 PM , Rating: 2
For instance, go to folding @ homes web site, score each cpu by dividing the teraflops by the active cpu's. You will find The Gpu's are more efficient. My x1950pro runs these at 2.2seconds per frame, thats fast compared to the cpu client. I think sony is scared to show their gpu power, because they are ashamed they picked such a current gen gpu. Because of this I think XBox can use its based between the Rv600 and has the same shader architecture as x1950pro.


RE: Gpu
By Wolfpup on 5/11/2007 4:40:19 PM , Rating: 2
The 360's GPU is much less powerful than ATi's SM 3.0 architecture, let alone a DX 10 part. It may do pretty well at Folding @ Home (or maybe not...we don't actually know it could run it at all).


More nonsense PR...
By Verran on 5/4/2007 9:24:23 AM , Rating: 3
Yeah, we freakin' get it. The PS3 folds well. As of yet, there's little mention of how it GAMES. Clearly, Sony is NOT interested in talking about quality of games and gameplay (I wouldn't be either, if I was them), so let's stay on the subject. How about benching the PS3 on the its ability to fold clothes, or mow my lawn, or inflate a tire on my car. How does it do there?

Oh yeah, no one cares!

Folding is great. Cure my cancer. That's fantastic. But here's a radical idea. How about comparing the PS3 on GAMES?!?!




RE: More nonsense PR...
By Spyvie on 5/4/2007 3:27:04 PM , Rating: 2
Sony’s PR machine has been hard at work since the launch

Shortly after the release of the F@H client for the PS3, the local FOX affiliate here in Denver did a fluff piece about the new Sony game machine helping scientists cure diseases. There was no mention that hobbyists have been Folding on the PC platform for a couple of years, but they did end the piece by saying “This works because the PS3 is 7 times faster than a PC”


RE: More nonsense PR...
By elmikethemike on 5/9/2007 7:46:54 AM , Rating: 1
Probably cuz folding PS3s stomp the crap out of all the other PC's combined. I mean its not even remotely close.

Give sony props for the PS3. It folds like a monster. But you wont cuz you're gay for the 360.


Horrible Blog
By Cobra Commander on 5/4/2007 8:20:37 AM , Rating: 3
Thanks for passing judgment on the Xbox360 based on some guy's speculation. This is lame .




RE: Horrible Blog
By elmikethemike on 5/9/2007 7:43:49 AM , Rating: 1
Yea, that professor guy with his PHd doesnt know what he's talking about at all. In fact he should just go back to his rocket science protein folding and stfu.


RE: Horrible Blog
By Wolfpup on 5/10/2007 4:46:25 AM , Rating: 1
Um, yes, he does. He's been doing this for over 7 years, has gotten this ported to a ton of different platforms, etc., and they've published over 100 papers. Obviously he knows exactly what he's doing.


Relevance
By oopyseohs on 5/4/07, Rating: 0
RE: Relevance
By Proteusza on 5/4/2007 9:21:03 AM , Rating: 3
Nobody ever said F@H is a waste of time, they only said the article is misleadingly titled.

Its all well and good to solve human problems like disease, but it seems there is no cure for dishonesty.


RE: Relevance
By cochy on 5/4/2007 3:16:50 PM , Rating: 2
quote:
let alone an article that is concerned with the best method of folding proteins to fight disease - be considered a waste of time?


It's not concerned with the best method of folding because the Xbox 360 doesn't have a F@H client. So yes this article is a hypothetical waste of space.


Some of you guys can't be serious...
By CupCak3 on 5/4/2007 12:42:05 PM , Rating: 3
Anyone who's really into folding and the technology behind it has known extactly what this article said for the past year. The 360 can't even come close to the floating point power of the PS3. I'm a HUGE folding advocate with a folding farm of my own but will never own a PS3 b/c I already have a 360.

Another question you have to ask yourself is WHY THE HELL WOULD YOU WANT YOUR 360 TO FOLD??? It can barely handle gaming without breaking. You would be able to heat a small house with the 360... before it died with the dreaded red ring of death

I do think this article has to be one of the most useless articles ever written by DT...




By jabber on 5/10/2007 8:28:31 AM , Rating: 2
I have to agree there.

With MS not being able to admit their 360 failiure rate (in other words its BIG), would they risk putting out an app that will mean potentially thousands of 360s running away at 100% nearly 24 hours a day? They must have done some testing and have some idea of the possible carnage that would happen.

If they do bring out a folding app for the 360 it could well mean its designed purely for the 65nm version (if it ever happens) with a re-designed power setup.

If I was MS I wouldnt risk it and as a consumer I wouldnt either.


How good is the PowerPC core at folding?
By Bladen on 5/4/2007 5:05:01 AM , Rating: 2
I have the impression (for some reason) that any one PPC core isn't that good at folding, when compared to a modern X86 core, but the SPE's in the Cell are good at folding.




By Proteusza on 5/4/2007 5:47:11 AM , Rating: 2
SPE's are stripped down PPC's, its simply the sheer number of cores that is helping, and the fact that they are all RISC architecture CPU's.

The X360 has 3 PPC's.


Move along people, nothing to see here
By TravisO on 5/4/2007 9:55:50 AM , Rating: 2
So where's the news here? Not even Microsoft would claim the 360 is more powerful than the PS3. By the way, did you hear about the price change of rice in the China market?




By bkm32 on 5/4/2007 10:23:33 AM , Rating: 2
Rice is good...and good for you.


IF DIET PEPSI COULD CURE CANCER THEN. . .
By Crazyeyeskillah on 5/4/2007 6:15:16 PM , Rating: 2
I'm sure we would be reading about it in the news all the time, as it would have benefit to Coca Cola, and Cancer is something that can be equally agreed to which people would like to see cured. The fact that the ps3 is such a versatile machine should really pay homage to IBM's brilliant cell chip. Sony is VERY lucky that they are recieving as much press by contracting out to IBM for this technology. While i won't buy a ps3 to 'fold', since i don't bother folding on my home pc while it's idling, Most of these articles are merely tech talk meant to entertain and inform. Comparing this to other consoles for the purpose of Folding is rather rediculous imo.

You my caucasian.




By EclipsedAurora on 5/5/2007 9:39:30 PM , Rating: 2
I think u have mixed up a bit background of Cell. Cell is originally drafted by Sony and Toshiba, and IBM was the one who did implementation only! The original architecture of Cell was based on Emotional Engine used in PS2, where in EE actually has some special purpose "DSP" which can run some operation (e.g. TnL, physic vectors and MPEG decoding) as very fast rate. It was Sony and Toshiba present the idea to IBM, which inspired IBM that seperating SIMD into another small specialized customed "core" can let the hardware running SIMD at a speed few times faster than exsisting general purpsose CPU today. And finally give birth of a jointed project between 3 companies.

So the "SPE" of Cell actually came from "VP0" and "VP1" core inside Emotional Engine used in PS2.


Reminds me of the old Mac days
By Nekrik on 5/4/2007 5:43:56 PM , Rating: 3
this seems pretty similar to the old Apple claims about their machines being oh so much more powerful than PCs, they would then go to use Adobe apps, specifically compiled for the Mac, and show them benchmarked against non-native/optimized versions of Adobe running on a Windows machine.

Of course one architecture might be better at a certain task than another. Chances are that the other acrhitecture will then excel at other types of tasks. While we're at it why don't we compare CPUs with and without math co-processors to see how well they compare at arithmetic operations.




Bad comparison,..
By zaki on 5/5/2007 5:28:20 AM , Rating: 2
It is inevitable that this comparison would be made because ps3 is fighting for the top spot with the 360 and the wii. so naturally if there is a big benefit for the ps3, people will write about it. i think sony has picked up a different market segment than xbox 360 owners because the ps3 owners do actually care about this stuff (proof: they are participating in folding @home)
where as 360 owners/fans want to shun anything non-gaming related on the ps3.
personally i feel that if you are going to call something next-gen then expect a few innovations. microsoft is welcome too, im sure 360 supporters wouldnt be dissmissing folding at home reports for their console.




RE: Bad comparison,..
By EclipsedAurora on 5/5/2007 9:31:21 PM , Rating: 1
U are right, Sony and Microsoft is taking different approach. But I have to disagree with you that the 360 has so called "innovation". I think wii should be the true winner in terms of innovation!

Anybody who is professional in the industry should know that Sony's involvement of Folding@Home is her first step towards the distributed computing market, where she may find one of the fastest growing area thoughout the whole computer industry. Current if an fresh graduated computer engineering/science or related area student with their thesis working on distributed computing, they can earn nrealy 3 times more than students doing any other area of interest!


No sh!t sherlock...
By sxr7171 on 5/5/2007 11:48:07 AM , Rating: 3
The cell was optimized for this kind of work. What I want to know is which is better at gaming, and I have seen no evidence of superior gaming from PS3 to what I've seen in Gears of War. So if anyone wants high points in F@H and has $600 rotting in their pockets, you know what to do. If you're looking for a good gaming console with actual games worth playing, you know where to go.




Cell vs Xenos
By Sureshot324 on 5/6/2007 7:42:29 PM , Rating: 2
The Cell may be more powerful overall, but how many game engines can fully take advantage of a 7 core CPU? Not many, and I think in most cases the xbox 360's 3 core cpu will be better.

Most games are GPU bound anyway since these consoles are running at HD resolutions, and the Xbox 360 has the more powerful GPU.




RE: Cell vs Xenos
By elmikethemike on 5/9/2007 7:41:04 AM , Rating: 1
This whole argument is retarded. If no one can prove the Cell is better than the 360s CPU, then no one can prove that the 360s GPU is better than the PS3.

All these fanboy comments like the one I'm replying to crack me up. I bet if you owned a PS3 you'd say the exact opposite. Oh, but you don't cuz your mommy and daddy won't buy it for you.

GTFU.


.
By boing on 5/4/2007 7:12:23 AM , Rating: 2
It's good to see the gaming industry doing something worthy but arguing about which games console is better at protien folding is like arguing about which of your girlfreinds was the best at changing your spark plugs, it's missing the point.

This generation more than any other seems to be so obsessed with tribal loyalties and juvenille pissing competitions that they will dive into any debate no matter how irrelevant just to 'prove' their superiority.

Buy these companies products, use them, enjoy them, then turn them off and go enjoy something else, don't let it mean any more than it should.




By rdeegvainl on 5/4/2007 7:41:53 AM , Rating: 2
I have seen numbers showing that the PS3 has actually used F@H, but so far I haven't seen anything showing that the 360 folded any or part of any protien. Though they could fix that, seems to me to be a bug in the coding. Maybe they could optimize it.
note: this was sarcasm.




CPU?
By fparis on 5/4/2007 7:42:10 AM , Rating: 2
Hey!, nobody seems to consider that the main difference is not in the CPU architecture but in the DMA controller... CELL is a "broadband" supercomputer because of it's concurrent, high througput DMA capabilities.




good benchmark...
By Randum on 5/4/2007 7:48:02 AM , Rating: 2
thats a good benchmark for two gaming consoles.....what!?




RE: good benchmark...
By OxBow on 5/4/07, Rating: 0
Apples to oranges?
By hardwareking on 5/4/2007 9:45:29 AM , Rating: 2
Cell is a specialized CPU and it is a floating point operation crunching monster.Bad for games,good for folding

IMO the Power PC based Xenon wudn't standa chance,since like other people said its half way b/w cell and a desktop proc

But the xenos gpu wud a be a completely different story
As of now i believe a x1900 xt matches a PS3 in terms of peak flops available?
So if a client shud ever come out for the 360,it'll prolly use the xenos GPU and kick some cell a**
And there are close to 11 million 360's out there too
imagine all the folding that cud be done




Blatantly Obvious
By Flunk on 5/4/2007 10:10:02 AM , Rating: 2
No one should be surprised by this. Folding@Home is exactly the type of application that the Cell processor excels at. massive linear computations are what the Cell was designed to do.

Still, when if I had a choice in programming a game for either chip I would choose the Xenos because it's much simpler to design a 3+ thread game than a game with 7+ threads 6 of which have to run on reduced functionality cores.




BIG TALK
By Jackyl on 5/4/2007 10:24:34 AM , Rating: 2
They can talk all they want about finding cures for this and that, and simulating everything. I want to see progress and actual cures for these diseases. This folding project has been going on for years, with billions of calculations performed. Yet there are no cures and all the calculations still remain a mystery. All this folding program does is make people feel good about themselves.

Also for all the conspiracy theorists....
For all we know, we could be calculating data for something else besides protein folding. Same with the SETI program. You really think they are going to release to the public their findings, if they find something?




Perspective
By Houdani on 5/4/2007 11:15:06 AM , Rating: 2
quote:
...PS3's are performing aspects of these simulations, and doing so about 20 times faster than a typical PC... [emphasis added]
It is important to note that the Cell processor is better at doing some types of work, whereas a CPU (much like the XBox360 processor) is better at other types of work. Knowing this allows us to keep things in perspective.

Folding@Home has different types of "work" which needs to be done. Cell just happens to be really good at one type of that work. CPUs and GPUs are really good at other types of work. This allows each camp (Cell, GPU, CPU) to have bragging rights all at the same time. They just so happen to be bragging about different things.

Here's a pretty good summary article which describes why the PS3 is so good at folding:
http://arstechnica.com/news.ars/post/20070326-why-...




Tomorrow's headline
By cochy on 5/4/2007 3:25:25 PM , Rating: 2
PS3 can fold more proteins than D-link routers.

Maybe if the 360 was designed for, had the ability to, or cared to, use F@H, this could be considered news.

Since it's not might as well compared apples to apples.




By darkpuppet on 5/4/2007 4:05:36 PM , Rating: 2
I see a lot of bias posted in people's comments in this thread, and not once has anyone considered all the aspects when they make their comparisons. So far everyone has found the angle they want to project, and post facts that support only their argument.

please consider...
-It's not CISC vs RISC (news flash, CISC processors have pretty been RISC with translators tacked on for YEARS now)
-SPE's aren't stripped down PPC. The PPE on the Cell is a stripped down PPC.
-traditional benchmarks for applications written on general purpose computers meant to run either ATI or nVidia aren't a true indication of purely optimized performance on either

technically speaking CPU wise, the Cell should outshine the xbox CPU in almost every regard and should be able to beat general purpose cores too if what i've seen of the new edge servers carries over to real life scenarios.

seriously, the only toss-up is really in the choices of memory interfaces and GPUs... both have advantages and disadvantages, and you're not going to see a real winner until GPU and CPU constraints are really hit... comparing what you know of traditional games on traditional computers may have applied to the original xbox, but it's not as relevant on the 360 because, once again, the architecture is different.

In the end, I guess it boils down to who you want to win.. MS or Sony, but I don't think it's so easy to pick a F@H or even next gen winner at this point as so many people are doing.

on a side note, I'm pulling for Sony, but I just find the tech so much more compelling personally.




Closed Archtecture
By Slaimus on 5/4/2007 5:33:13 PM , Rating: 2
Seeing how MS is not even allowing homebrew software like F@H to run, I am not sure why this is relevent.

Maybe this will cause MS to actually open up the hardware and let people run F@H.

As for Cell, of course it will be faster these these types of tasks. The PS3 was originally supposed to be just the Cell, doing regular calculations on the PPE and 3D calculations on the SPEs. The SPEs were not fast enough, so they tacked on a G71 chip. Seeing how fast GPUs are at F@H, there is no reason to doubt that Cell is faster.




Interesting logic...
By thestereotype on 5/4/2007 6:23:23 PM , Rating: 2
It seems to me that the fact that there are more X360s out there would mean something about the total folding power... IDK though.




Unfair comparsion!
By EclipsedAurora on 5/5/2007 6:51:36 AM , Rating: 1
>>>While the Xbox 360 CPU may not be tops in terms of Folding@home, its ATI-developed GPU could be much better. ATI Radeon GPUs running Folding@home are outperforming the PlayStation 3 on a per capita basis. Though whether or not the Xbox 360 will ever get its crack at helping to cure a disease remains to be seen.

This is completely a "non-sense" bais to favour M$. If u take a deep look into StanFord's Folding@Home client FAQ, the only the ATI X1900 series GPU perform well in their code since the number of shaders and driver implementation in the X1900 series. Other ATI GPU DO NOT have good results as well

quote:
The R580 (in the X1900XT, etc.) performs particularly well for molecular dynamics, due to its 48 pixel shaders. Currently, other cards (such as those from nVidia and other ATI cards) do not perform well enough for our calculations as they have fewer pixel shaders. Also, nVidia cards in general have some technical limitations beyond the number of pixel shaders which makes them perform poorly in our calculations.


http://folding.stanford.edu/FAQ-ATI.html

Also, the GPU client has a very limit kind of calulation in can perform only. In general, still a GPU is not a general purpose design.
quote:
We balance the points based on both speed and the flexibility of the client. The GPU client is still the fastest, but it is the least flexible and can only run a very, very limited set of WU's. Thus, its points are not linearly proportional to the speed increase. The PS3 takes the middle ground between GPU's (extreme speed, but at limited types of WU's) and CPU's (less speed, but more flexibility in types of WU's). We have picked the PS3 as the natural benchmark machine for PS3 calculations and set its points per day to 900 to reflect this middle ground between speed (faster than CPU, but slower than GPU) and flexibility (more flexible than GPU, less than CPU).


http://folding.stanford.edu/FAQ-PS3.html




WoW
By chick0n on 5/4/07, Rating: -1
RE: WoW
By Proteusza on 5/4/2007 8:22:25 AM , Rating: 3
As a programmer trained in CPU design I think its a close call.

I dont have anything against Cell, or Sony.

I just think they should have written a client for the 360, tested it, and then said it was slower.

The actual performance isnt the issue here, because there is no comparison, only guesswork.

If they had figures to prove it, then good for Sony. If they dont, bad on DT for using such a sensationalist headline, when there is no proof.

What if made a headline that said, "Intel's Penryn will beat AMD's Barcelona" ?

And then in the article I got a guy with some knowledge to make a guess. Is that good journalism, or is it attention seeking?


RE: WoW
By GlassHouse69 on 5/4/07, Rating: -1
RE: WoW
By jkresh on 5/4/2007 3:04:16 PM , Rating: 1
First let me say that I bought a 360 at launch, a wii at launch, my brother has a ps3 (which runs folding at home, as do my workstations) and I may get a ps3 when enough interesting games are out (or when I decide to start coding some stuff for the cell). That being said, its very easy for them to figure out that in terms of cpu the ps3 is superior for folding as each of the 3 cores in the 360 is like the ppu in the ps3 and the ppu is crappy for folding, the calculations run on the spu's and 360 doesn't have an equivalent part. Now in terms of graphics the 360 has a much better card for folding and would put up impressive numbers if a client was made for it (based on results from an x1900, x1950 and the extra features on the 360 gpu) but...


RE: WoW
By therealnickdanger on 5/4/2007 8:24:17 AM , Rating: 5
You're not seeing "Sony haters", you're seeing "bad article haters". This is clearly Sony PR with no real substance behind it. Yes, we've all known for some time that the Cell can beat the 360 CPU. Yes, I'm sure that his comments would prove true if tested. However, this article is flamebait because there are no facts in it. Let's see numbers to back up claims, let's see F@H utilizing the GPUs in both consoles as well. I wouldn't release something like this until I had numbers to back it up - no matter how positive I was that it was true. This is a tech site, so I guess we (or at least, I) have some expectations that stories be factual and/or technical. Leave stories like this for The Inq.


RE: WoW
By Goty on 5/4/2007 9:15:24 AM , Rating: 2
The only reasons this is a bad article are because the title is misleading and nothing said by Vijay has anything to do at all with Xenon.

All the statements made by Vijay are true. The Cell would beat out the 360's CPU in F@H calculations because the Cell holds and advantage over it in pure floating point power and in parallelism. The only person who said anything about Xenon in this article was the author.


RE: WoW
By Proteusza on 5/4/2007 9:24:20 AM , Rating: 1
Actually Vijas does directly reference the 360.

Nonetheless, there simply is no proof that the Cell is any more powerful than the 360. Have you ever seen any definitive proof ever that this is true? There has never been an apples to apples comparison between the two, so any speculation is just that - speculation.

Hence we cant take Vijay's comments as true, because we really can only guess.


RE: WoW
By Goty on 5/4/2007 10:55:34 AM , Rating: 2
I said referenced the 360s GPU .

And yes, we can pretty much definitively say that the Cell will beat a triple-core PowerPC CPU ( I said x86 before, I was mistaken).

Notice I did not say the 360, I merely said the CPU.

The reasons for this have already been stated.


RE: WoW
By therealnickdanger on 5/4/2007 11:25:28 AM , Rating: 2
It would be very cool to see a F@H app written to combine the power of the CPUs and GPUs of both systems and see which can come out on top overall. Those SPEs can kick some ass when it comes to this type of number crunching, but the Xenon GPU can completely destroy the RSX. I think we'd find that the two consoles would be nearly a wash overall, but we won't know until we try it.


RE: WoW
By Wolfpup on 5/5/2007 12:36:51 AM , Rating: 2
That's a solid hypothisis, but far from a fact. We don't really know if Xenos has the hardware needed to run this well. The Geforce 7800 line runs it slower than a Pentium 4, which is why they gave up on it, while the x1900 line runs it faster...something about the way they handle floating point math, I think (though I don't remember for sure).

It's a solid guess that Xenos might have the same math capabilities as the x1900 line since it's from the same company, but it may not. It's a very different part with fewer transistors. It may well not have the same capabilities, in which case we have no idea if it could beat RSX at Folding @ Home.


RE: WoW
By Wolfpup on 5/5/2007 12:33:26 AM , Rating: 2
There’s nothing bad about this article. It just quotes that Cell is faster than Xenon at Folding @ Home.

And yes, the lead of the project who’s gotten this ported to x86, PowerPC, Cell, GPUs, and multiple OSs is qualified to make a statement like that. (Not to mention, as has been pointed out previously-the single PPE in Cell is pretty much the same thing as the three CPUs in Xenon.)

It’s a well known fact that Cell has more raw power than Xenon. It’s also a well known fact that both CPUs are terrible general purpose CPUs. In fact, if you read up on this, the work being done on Cell and GPUs is just a subset of the work being done on general purpose CPUs. Those processors are much faster for some types of work, but far, far slower for others that require a good general purpose CPU.


RE: WoW
By Goty on 5/4/2007 9:09:37 AM , Rating: 2
I don't think the Cell B.E. would be able to best Xenon in F@H calculations for one important reason: parallelism. F@H thrives when you give it more execution units, of which the Cell only has 8, or as close to that number as makes no difference. I'm positive that Xenon has more than 8 stream processors (maybe an order of magnitude more), which would already give it an advantage.


RE: WoW
By Lugaidster on 5/4/2007 10:26:58 AM , Rating: 2
Are you ok?? Really, get your fact's straight. The cell architecture is even more parallel than the xenon. Really, read a bit before you post. That's like saying:

"Dude, my 380 HP car is faster than your 400 HP car for one important reason: Horsepower!"

Really, go read an article on both, arstechnica has articles on the inner workings of both the cell and the xenon.

Since the 360's CPU is an in-order There's mostly no gain in adding execution units to the core since there would be no way to feed all of them, that's why the 360's cpu hyperthreads, kinda like the p4 but hopefully a lot more efficient. Besides that, not all execution units provide floating point performance, because there are different types of them inside the cpu.

On a side note, IBM designed both cpu's, do you really think that they would be promoting cell as the ultimate processor architecture if xenon was more powerfull?

Xenon only performs fast in very specific cases, and are even more specific than those for the cell, contrary to what you might think. Since the fact that you lack dynamic execution in that cpu, only in very specific streams of instruction do you get to utilize all the execution units. I really doubt anyone at Stanford would want to put time tuning fah to run fast on xenon.

Still they could make fah run on Ati's GPU. But that's for another moment.

My regards...

PS: Don't Get me wrong I think Xenon is an excellent architecture for what it was designed: gaming. I just don't think it's as good in other applications.


RE: WoW
By Proteusza on 5/4/2007 10:41:25 AM , Rating: 1
quote:
Since the 360's CPU is an in-order There's mostly no gain in adding execution units to the core since there would be no way to feed all of them, that's why the 360's cpu hyperthreads, kinda like the p4 but hopefully a lot more efficient. Besides that, not all execution units provide floating point performance, because there are different types of them inside the cpu.


Sigh.

Both are in order actually. In order cores need not be an advantage if you have good compilers, or hand written code.

In order just means the CPU wont dynamically reorder instructions based on expected execution time, to keep the pipeline full. Much simpler, but as I said, needs good compilers to make sure the code doesnt need much optimization.

Thats not why it hyperthreads, hyperthreading isnt even the right word (its an Intel trademark). Its multithreading, and it refers to one core running two threads at a time, which is different to super scalar execution, in which a core executes multiple instructions at a time. The Cell is a 2 issue superscalar CPU design, meaning each of its cores can execute a maximum of 2 instructions at a time. The Core 2 Duo does 4 if I remember correctly.

In terms of keeping the pipes full, you also need good instruction prefetching and branch prediction, a whole other story.

Adding execution units will help, depending on usage scenario.

quote:
On a side note, IBM designed both cpu's, do you really think that they would be promoting cell as the ultimate processor architecture if xenon was more powerfull?


Maybe Microsoft didnt pay them enough? It takes a significant investment by all parties involved to attempt to make something as ambitious as cell. IBM probably stands to gain more from Cell, because Cell is going to be more widespread. Think of it as IBM's favourite child.

quote:
Xenon only performs fast in very specific cases, and are even more specific than those for the cell, contrary to what you might think. Since the fact that you lack dynamic execution in that cpu, only in very specific streams of instruction do you get to utilize all the execution units. I really doubt anyone at Stanford would want to put time tuning fah to run fast on xenon.


Not true. It will perform well in similar scenarios as Cell. Its also a risc processor, its also based on the PowerPC architecture, its also a multi core architecture. It is by no means a gaming only CPU. It just hasnt been marketed as such.


RE: WoW
By Wolfpup on 5/5/2007 3:29:36 PM , Rating: 2
Actually, no, they both do have Hyperthreading (we're using that term because it's the commonly used term, even if it is the branded name for something else-I think SMT might be the real "generic" term?)

Xenon can "execute two threads" the same way a P4 with hyperthreading does-it stops one, and starts another. It's "more efficient" on the Xenon only because the CPU is so much less efficient th at it's more likely to be sitting there doing nothing-so swapping to a second thread allows it to continue working while the first thread is stalled. The Pentium 4 is a much more efficient design, so doesn't stall as much, so doesn't see the same kind of benefit from Hyperthreading.


RE: WoW
By Goty on 5/4/2007 11:04:17 AM , Rating: 2
I don't think I'm the one who doesn't have his facts straight.

Xenon is the codename for the 360's GPU, not its CPU.

The Cell in the PS3 has, what, 8 SPEs? The 360's GPU has 48 shaders, all of which are able to execute code simultaneously. So, quick math quiz: who here can tell me which number is greater? 8 or 48?

Now, granted the clockspeed of the 360's GPU is quite a bit lower than that of the Cell, but the ability of each shader to execute four floating point ops per cycle (along with the fact that there are six times as many shaders in the GPU as SPEs in the Cell) easily makes up for this difference.

So basically, no, my argument is not misplaced.


RE: WoW
By harshbarj on 5/4/2007 11:54:32 AM , Rating: 2
Dude your on crack. Xenon is the cpu and "Xenos" or "C1" is the GPU.

Please quit being such a fanboy and admit the cell is likely (still to be proven) faster in fah than the Xenon.

Article about the Xenon.

http://arstechnica.com/articles/paedia/cpu/xbox360...

Details about the 360 GPU

http://www.techreport.com/etc/2005q2/xbox360-gpu/i...

Article with the GPU codename
http://interviews.teamxbox.com/xbox/1458/The-Power...


RE: WoW
By Goty on 5/4/2007 7:21:45 PM , Rating: 2
Ah, ok, well thank you for correcting me on the nomenclature there, but I think you're still a little confused.

You keep saying that the Cell would be faster than the 360s CPU in F@H calculations, and I've never disputed that (as a matter of fact, I agreed with it if you read my whole post). I'm saying that the 360s GPU is almost undoubtedly faster than the Cell at the same calculations, as stated (with reasoning) previously.

What's really funny about all of this is that I normally get flamed for being a "PS3 fanboy" when I'm probably one of the more impartial contributors to these boards.


RE: WoW
By harshbarj on 5/5/2007 3:25:22 PM , Rating: 2
It might be true that the gpu in the 360 can beat the cell in fah but that was never the point of this article. It was comparing the cpu in each (as best they could without a benchmark). Now if they release an article that says overall the ps3 is better for fah than the 360 then that could cause a bit of debate (although without a 360 client we will never know for sure).


RE: WoW
By Wolfpup on 5/10/2007 4:19:39 AM , Rating: 2
Also, we have no way of knowing if Xenos would even run Folding at Home well, or at all. It's particular capabilities may not work for it.

It's an interesting idea that it might run it well, but we certainly don't know that.


RE: WoW
By jrb531 on 5/4/2007 11:00:48 AM , Rating: 2
This has nothing to to with hating Sony. The article is pure BS.

The 360 has a video chip that outperforms the PS3 in folding. They know this so they compare CPU's and in the ultimate form of BS... they are guessing because a folding program does not exist for the 360.

Tell ya what....

The 360 beats all "white colored" consoles in folding!

Is this headline any less silly that this one?

-JB


RE: WoW
By akugami on 5/4/2007 1:56:22 PM , Rating: 2
Some of the people here are saying the story is inaccurate and especially the headline misleading. Others are pointing out that even if the PS3 is 1000 times better than a Xbox 360, or even a quad core Intel chip, it's still a games console and no one has seen any games that make it worth the price premium over the Xbox 360 and Wii games consoles.

If someone bought their PS3 for F@H, then great, good for them. Most I would assume bought it as a games console with a handful buying it as a Blu-Ray player with gaming capability.


RE: WoW
By sxr7171 on 5/5/2007 11:54:23 AM , Rating: 2
Yeah true, but what about in gaming? What was it really designed for? So owners can flash bigger and bigger ePenises?


"I'm an Internet expert too. It's all right to wire the industrial zone only, but there are many problems if other regions of the North are wired." -- North Korean Supreme Commander Kim Jong-il

Related Articles













botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki