MikeB said: Peak performance Xenon, 77 GFlops. Source: Forbes/IBM Microsoft once claimed the Xenon's peak performance is actually 115.2 GFlops, this appears to be a very common misconception. The Amiga community buddy who wrote that Cell article linked to aboves, provides the following explanation: "The 115.2 figure is the theoretical peak if you include non-arithmetic instructions such as permute. These are not normally included in *any* measure of FLOPs." "If you want to count non-arithemitic peak figures, the usable Cell components in the PS3 will get 268 Gflops (6 SPEs + PPE) - over twice that of the 360." Note the 268 GFlops figure does not take into account the GFlops SPE used by the PS3's CellOS. (which isn't available under Linux)

 

Which would make the Cell an ideal processor if you're decoding multimedia streams or running a web application, but the majority of what your CPU is responsible for in games is not that parallelizable or asyncronous which will limit how close you can get to approaching theoritical performance. If developers can only ever achieve 80% of the Xenon's theoritical peak and 40% of the Cells theoritical peak in game then the processors are as powerful (for gaming) as eachother.

NNN2004 said: i think we will see the true power of Cell in Killzone 2.

 

IMO you will see a game which isn't possible without significant sacrifices on the 360, however they are currently using only 4 of the available SPEs, so there will be plenty of headroom for a Killzone 3 to tap even greater performance, both from the SPEs they are already using and 2 more they aren't using yet.

Maybe for the final game they will use all 6 SPEs (like Resistance 2 uses 6 SPEs, Resistance 1 used ~10% of achievable performance from 5 SPEs), but then would have a lot of CPU cycles to spare on each processor.

i dont see what the big deal is even if it is twice as powerful. It is commonsense at the end of the day. if both the xenon and the cell have multicore chips that run at the same speed, then it would be pretty obvious that if u had twice the number of SPE's on one or the other, then the performance would be something near to being twice as fast.

at the end of the day each SPE is assigned some sort of work. the more spe's you have the more work you can assign to them, meaning more gets done at the same time. e.g. one factory with 10 workers will do more work then one factory with 5 workers.

HappySqurriel said: I would say trading ease of development for theoritical processing power which may never be obtained is a poor trade-off ... Developers have been working with both processors long enough that they're approaching the maximum performance that they will ever see, and both processors are performing in a similar range.

Final devkits for both platforms only went out ~1 month before their respective launches.  This gives developers a fully year head start working with the 360.  It's 2.5 years vs 1.5 years.  Developers haven't been working with the Cell all that long, let alone compared to the how long they've been working with the Xenon.

Also, developers continually say that there is plenty of untapped potential in the Cell, and that they are accessing more and more of it as time goes by.

Really, both processors probably have a good amount of untapped performance left in them.  Just look at games like God of War II that were popping up at the end of the ps2's life.

 

@ NNN2004

He probably got them from Nvidia presentation sheets. The RSX can perform more shader ops per clock than the Xenos can (136 vs 96).

Comparing the Xenos and RSX, both GPUs have their advantages, it's not as clearcut as comparing the Cell and Xenon. For the RSX to operate as efficiently as possible the GPU was (re-)designed to take good advantage of the Cell.

HappySqurriel said: Which would make the Cell an ideal processor if you're decoding multimedia streams or running a web application, but the majority of what your CPU is responsible for in games is not that parallelizable or asyncronous which will limit how close you can get to approaching theoritical performance. If developers can only ever achieve 80% of the Xenon's theoritical peak and 40% of the Cells theoritical peak in game then the processors are as powerful (for gaming) as eachother.

Well, they can use it with simple tasks like physics and stuff(as those are usually embarrassingly parallel), but not really that much graphically as you know today "power = graphics". And when it comes to advanced AI. Making it work on CELL is well... hell, if you can even make it at all. So imo best way how you could enchance games with CELL would be more lively world(moving geometry and calculate for example effect of wind for every object, liquid calculations & so on).

to be honest .... some times i dont beleive this sony guys ... before the ps2 came they said its processor capable of doing 60 milion something ( i forgot what is it exactly but i think one of u guys remember it ) but when its came its appear that its can do only 16 milion ... if anybody remember anything about this plz post it.

makingmusic476 said: HappySqurriel said: I would say trading ease of development for theoritical processing power which may never be obtained is a poor trade-off ... Developers have been working with both processors long enough that they're approaching the maximum performance that they will ever see, and both processors are performing in a similar range. Final devkits for both platforms only went out ~1 month before their respective launches.  This gives developers a fully year head start working with the 360.  It's 2.5 years vs 1.5 years.  Developers haven't been working with the Cell all that long, let alone compared to the how long they've been working with the Xenon. Also, developers continually say that there is plenty of untapped potential in the Cell, and that they are accessing more and more of it as time goes by. Really, both processors probably have a good amount of untapped performance left in them.  Just look at games like God of War II that were popping up at the end of the ps2's life.

 

... Not having a final development kit is an adequate excuse for launch games to have poor performance 18 months after launch it is a pathetic excuse. Beyond that, Epic had the initial development kit for the PS3 at least 18 months before launch and (even though not everything performs exactly the same in an initial development kit) after 3 years with a processor you should be seeing decent performance from it.

MikeB said: @ NNN2004 from where the hell u get this number @MikeB ??!!! He probably got them from Nvidia presentation sheets. The RSX can perform more shader ops per clock than the Xenos can (136 vs 96). Comparing the Xenos and RSX, both GPUs have their advantages, it's not as clearcut as comparing the Cell and Xenon. For the RSX to operate as efficiently as possible the GPU was (re-)designed to take good advantage of the Cell.

 

 Xenos have more power than RSX like we all know. 

@ makingmusic476

Agreed, but also the 360 dev kits were more mature (for example for the 360 launch title Kameo already 85% of available CPU cycles were being harvested, source: Microsoft).

It's not that IBM/Sony did a bad job on their early development tools, the Cell design is radically different compared to previous CPU designs, it's more revolutionary than the Xenon, which has a more evolutionary design, has more in common with past multi-core CPU designs on the PC and Microsoft took many Windows tools with them onto the 360.