Laffer said: As a IT professional in Infrastructure and having implemented some HPC Clusters, the discussion must be separated into 1.) What is the better Console and 2.) what is the better CPU Point 1.) is much up to your personal preference and budget but Point 2.) does not need to be discussed. IBM developped and manufactures both CPU's. The XBOX360 CPU was some sideproduct of the Cell and Power development. IBM never used this CPU further or used it for any new development. The Cell is absolutely the contrary. IBM is betting their future on Cell. The PS3 Cell Chip was part of their QS21 Blade Center (2 PowerCell with 8 GB Memory @ 3.2GHz). The PowerCell8i was an improved version, now supporting Double Precision Floating Point natively in the SPU's and supporting up to 16 GB of DDR2 RAM (the QS22 Blade, 2 CPU, 32 GB DDR2 RAM) The NextGen Cell will sport 4 Power6 Cores and 32!!! SPE's!!! The processing Power will kill everything on this planet. To compare DP FP Power: PS3 Cell: with software approx 120 GFLOPS DP, 200 GFLOPS SP PowerCell8i: 200 GFLOPS DP, 400 GFLOPS SP ATI Radeon Firestream 9250 (HD 4870), 200 GFLOPS DP , 1200 GFLOPS SP So the actuall PowerCell8i is matching with 8 SPU's already the fastest available Graphics board in Double Precision FP Ops. IBM has still the issue with unleashing this incredible power, because software libraries optimized for Cell using SPE's are not available and hand coding is needed. The Power7 Generation will just be a Cell processor. Cell was the first mainstream hybrid processor, it is somehow the father of what AMD will do with the 1st FUSION Processor (Bulldozer) and where all allpurpose CPU will go... (like Intel's Sandy Bridge)

Cool!  I'm an IT guy as well (Enterprise Technology division of the company).

Yeah, I have the slide on the 4PPEs + 32 eSPEs.  The performance is suppose to be 1TFLOPs - SP  500GFLOPs - DP.

Some of the improvements to the SPEs.  The 4 PPEs/PPUs are suppose to be based on POWER7 technology, last I heard.  That was before they decided on 4 PPEs/PPUs.

I LOL'd at the line in bold.

 

Deneidez said: Ascended_Saiyan3 said: Deneidez said: So its not many tasks, but one task? Did you know that computing is much more than just floating point operations and that CELL is crap compared to newest GPUs today in flops? Do you also know why games do not run entirely on GPUs today? Thats the same reason why CELL can just beat current cpus in some tasks. I don't say its piece of crap, but its not really breakthrough in computer science. @MikeB I have already answered how. You just ignored my post last time. I did even provide you some pseudocode, but you didn't seem to understand it. (Btw, tomorrow I publish one of my programs(For you anyway. It has been available on irc for a while already.). Check sig. ^^) You bold words you added aren't true.  The Cell is a general purpose processor.  Even Dr. Hofstee, the creator of the Cell, said they are general purpose processors.  Those links show general purpose in the versatility of the applications they are used in...period.  There is NO LOGICAL way around that.  Stop trying to turn a blind eye to the technology just because it's a revolutionary concept.  The GDC 2009 dev PDFs even show the results of that general purpose usage. Did I say that CELL isn't general purpose processor? No I didn't, I did say that most of general purpose processors rape CELL when it comes to general purpose stuff. You seem to be expert in this so tell me what does IoE do for general purpose stuff with lots of branches? (Answer: It might stall processing 'infinitely'. You might ask wtf is IoE. Its something that CELL uses. X360 also uses it, but original Xbox do not use it. More info about IoE & OoOE -> http://en.wikipedia.org/wiki/Out_of_order_execution ) Basic idea is when running program on IoE platform, if theres no resources available for next calculation it will have to wait. On OoOE platform it will do the stuff when it can do it and not to stall. And no sometimes it does stall. You just can't make some programs to work perfectly on IoE. Yeah, I know theres branch predicting also, but its not as good as OoOE.

Why would you use extremely branchy code when ALL processors run MUCH faster without them (Branch Elimination). Anyway, the Cell has branch hinting. If you know what you're doing, there would be FAR less penalties for a branch miss on the Cell. Plus, you are wrong about your statement saying most general purpose processors would rape Cell in general purpose tasks AS MY DOZEN OR SO LINKS PROVE (one or two pages back). There's that blind eye, again. ;)

Yeah everybody knows that 95 % of the multiplatform games look and play better on XBOX360. Nobody can deny or reject facts...

So you agree. The Next-Gen Cell will be out not to far away from now and be VERY powerful. Of course Cell is general purpose. But They took an approach different from the standard Multicore. Todays Multicore CPU's like Phenom and Core i7 are only really used on servers (Opteron, Xeon). The General Purpose Applications are just not much multithreaded. So using the complete Cores requires many users. Games, Visuals, Simulation do not really need GP Operations, but just raw compute power and often this calculations are naturally parallel. So even having 4 cores with SMT, gives you just 8 threads with all the fucking overhead of Cache coherency etc. That is very the GPU jump in, implizit parallel, just working on data streams. 800 Unified Shaders, which are 800 spezialized CPU's. Cell was first on the market to combine these 2 ideas. Use GP CPU where is lots of different things to do and much brain involved, but not much parallel and use high speed number cruncher which offer more intelligence than GPU Streamprocessors. Fuse it together with a high-speed bus and VOILA.

Cell is not the thing to run Word faster than a Core i7 or Phenom. First of all, Cell is an in order CPU design. It is not done to run Word and Excel and Internet Explorer in Parallel.
But Cell is at his best where you have a single Lead application with tons of things to calculate and to synchronize. The PPU's are the brain spending energy to organize the slaves, doing the management work. The SPU's are incredible efficient workers without much brain, but incredibly fast at small jobs (like the Shadersprogramm in GPU).

How do you want to use fully the 3 Cores in the XBOX360 CPU on ONE application? Even if you are threading, most part of the CPU logic is just dead!

Cell has a much more intelligent approach. SPE's are small, because they do not have OOE Units, Fetcher, Decoders, Branch predictors and all this other stuff to run multiple applications at the same time fast.

But the Future Cell, would also match the AMD-INTEL fraction on this part, it does not today, because it is already 4 years old and followed another approach.

AMD and INTEL are trying to go this way in future, because multiplying HUGE Full COREs is a waste of space and money.

Like Cars, the future will become Hybrid.

Ascended_Saiyan3 said: Why would you use extremely branchy code when ALL processors run MUCH faster without them (Branch Elimination). Anyway, the Cell has branch hinting. If you know what you're doing, there would be FAR less penalties for a branch miss on the Cell. Plus, you are wrong about your statement saying most general purpose processors would rape Cell in general purpose tasks AS MY DOZEN OR SO LINKS PROVE (one or two pages back). There's that blind eye, again. ;)

Branch predicting != OoOE and sometimes you have to have lots of branches, if theres enough branches in program even old Xbox would perform better than CELL. Theres also problems with memory in CELL approach. SPEs have way too small memory for some tasks. For example 50MB chunk in memory with random depencies into other points of that 50MB chunk. You would have to use main mem only and that would make processing quite slow.

And about that blind eye. Show me one link that shows that its better in general purpose stuff than PCs today. If you run only simple instructions/integer stuff, CELL wouldn't beat even 3-4 years old average computers and if you put some branches there it would render even slower.

Well, as there seem to be title war. I am going to be master of science in computer science in year or two.

(Btw, first program is available. For linux & windows. Check sig, ty. ^^)

Deneidez said: Ascended_Saiyan3 said: Why would you use extremely branchy code when ALL processors run MUCH faster without them (Branch Elimination). Anyway, the Cell has branch hinting. If you know what you're doing, there would be FAR less penalties for a branch miss on the Cell. Plus, you are wrong about your statement saying most general purpose processors would rape Cell in general purpose tasks AS MY DOZEN OR SO LINKS PROVE (one or two pages back). There's that blind eye, again. ;) Branch predicting != OoOE and sometimes you have to have lots of branches, if theres enough branches in program even old Xbox would perform better than CELL. Theres also problems with memory in CELL approach. SPEs have way too small memory for some tasks. For example 50MB chunk in memory with random depencies into other points of that 50MB chunk. You would have to use main mem only and that would make processing quite slow. And about that blind eye. Show me one link that shows that its better in general purpose stuff than PCs today. If you run only simple instructions/integer stuff, CELL wouldn't beat even 3-4 years old average computers and if you put some branches there it would render even slower. Well, as there seem to be title war. I am going to be master of science in computer science in year or two. (Btw, first program is available. For linux & windows. Check sig, ty. ^^)

That's not true, IF the programmer use the branch hints PROPERLY.  If they don't, you would be right.  However, that would be the programmer's fault and not the processor, because the processor is capable with proper effort.

Don't ask me to provide the links AGAIN, when I went through the trouble of putting them all in one place (one to two pages back).  Just look at the multiple purposes (by definition that's general) that were demonstrated.

What's this about a title war?  BTW, Mike Acton, from Insomniac, tells you to UNlearn what they taught you in school.  He says the UNlearning process is what makes parallelizing your code for the Cell hard.  Have you UNlearned your old practices today? ;)

 

Laffer said: So you agree. The Next-Gen Cell will be out not to far away from now and be VERY powerful. Of course Cell is general purpose. But They took an approach different from the standard Multicore. Todays Multicore CPU's like Phenom and Core i7 are only really used on servers (Opteron, Xeon). The General Purpose Applications are just not much multithreaded. So using the complete Cores requires many users. Games, Visuals, Simulation do not really need GP Operations, but just raw compute power and often this calculations are naturally parallel. So even having 4 cores with SMT, gives you just 8 threads with all the fucking overhead of Cache coherency etc. That is very the GPU jump in, implizit parallel, just working on data streams. 800 Unified Shaders, which are 800 spezialized CPU's. Cell was first on the market to combine these 2 ideas. Use GP CPU where is lots of different things to do and much brain involved, but not much parallel and use high speed number cruncher which offer more intelligence than GPU Streamprocessors. Fuse it together with a high-speed bus and VOILA. Cell is not the thing to run Word faster than a Core i7 or Phenom. First of all, Cell is an in order CPU design. It is not done to run Word and Excel and Internet Explorer in Parallel. But Cell is at his best where you have a single Lead application with tons of things to calculate and to synchronize. The PPU's are the brain spending energy to organize the slaves, doing the management work. The SPU's are incredible efficient workers without much brain, but incredibly fast at small jobs (like the Shadersprogramm in GPU). How do you want to use fully the 3 Cores in the XBOX360 CPU on ONE application? Even if you are threading, most part of the CPU logic is just dead! Cell has a much more intelligent approach. SPE's are small, because they do not have OOE Units, Fetcher, Decoders, Branch predictors and all this other stuff to run multiple applications at the same time fast. But the Future Cell, would also match the AMD-INTEL fraction on this part, it does not today, because it is already 4 years old and followed another approach. AMD and INTEL are trying to go this way in future, because multiplying HUGE Full COREs is a waste of space and money. Like Cars, the future will become Hybrid.

I subscribe to almost all of this.  Looking at Intel's future processor roadmap, all processors are going the way of the Cell in the near future.  Then, it will eventually go to a, basically, all SPU/SPE type configuration.

 

eliasg said: Yeah everybody knows that 95 % of the multiplatform games look and play better on XBOX360. Nobody can deny or reject facts...

 

Fanboys fanboys wat u gonna do wat yo gonna do when they BS you fanboys fanboys.

XxXProphecyXxX said: eliasg said: Yeah everybody knows that 95 % of the multiplatform games look and play better on XBOX360. Nobody can deny or reject facts...   Fanboys fanboys wat u gonna do wat yo gonna do when they BS you fanboys fanboys.

 

Lol go to neogaf to check your self Fanboy....

Ascended_Saiyan3 said: Deneidez said: Ascended_Saiyan3 said: Why would you use extremely branchy code when ALL processors run MUCH faster without them (Branch Elimination). Anyway, the Cell has branch hinting. If you know what you're doing, there would be FAR less penalties for a branch miss on the Cell. Plus, you are wrong about your statement saying most general purpose processors would rape Cell in general purpose tasks AS MY DOZEN OR SO LINKS PROVE (one or two pages back). There's that blind eye, again. ;) Branch predicting != OoOE and sometimes you have to have lots of branches, if theres enough branches in program even old Xbox would perform better than CELL. Theres also problems with memory in CELL approach. SPEs have way too small memory for some tasks. For example 50MB chunk in memory with random depencies into other points of that 50MB chunk. You would have to use main mem only and that would make processing quite slow. And about that blind eye. Show me one link that shows that its better in general purpose stuff than PCs today. If you run only simple instructions/integer stuff, CELL wouldn't beat even 3-4 years old average computers and if you put some branches there it would render even slower. Well, as there seem to be title war. I am going to be master of science in computer science in year or two. (Btw, first program is available. For linux & windows. Check sig, ty. ^^) That's not true, IF the programmer use the branch hints PROPERLY.  If they don't, you would be right.  However, that would be the programmer's fault and not the processor, because the processor is capable with proper effort. Don't ask me to provide the links AGAIN, when I went through the trouble of putting them all in one place (one to two pages back).  Just look at the multiple purposes (by definition that's general) that were demonstrated. What's this about a title war?  BTW, Mike Acton, from Insomniac, tells you to UNlearn what they taught you in school.  He says the UNlearning process is what makes parallelizing your code for the Cell hard.  Have you UNlearned your old practices today? ;)

You cannot predict all branches. Well, I checked those links are there were only mention that CELL can do general processing, but there were nothing about how good it is in it. Other than some marketing speech of course.

Well, not title war. More like title boasting. :)

Anyway same applies for X360 too,

http://arstechnica.com/old/content/2005/06/xbox360-2.ars/7

etc. Read the whole article.