Some quotes about PS3 CPU

[V3]

i doubt we will see a 3.6 ghz cell being used in the ps3 , we'd see 3.6 ghz chips in the xenon before that happens .

A complex chip like cell is not going to scale that high nor was it meant to

Well one of the idea of multicores chip is that so you can run each core at very high speed, if need be. And if any cores are overheating, the chip or OS will shut them off, knowing there will be other cores to take over.

I expected the efficiency that you will get from this multicore chip in Xenon CPU or PS3 or Revolution, will take care of any overheating. So it will work out nicely.

 

[SiBoy]

Just like the doubts about the clock speed being high are totally wrong. CELL is being built differently, so when they manufacture CELL they won't run into the same type of problems that hamper old-style CPU architectures.

You're right, but for the wrong reason. The clock speed is high for the same reason it is high on an Intel Pentium.

A great example is MIT's raw where you have a processor designed to combat the problems of wire delay.

This has absolutely nothing to do with Cell. Just another university project, like Berkeley IRAM, etc. Funny how IRAM dropped off the face of the earth.

 

[Farid]

A complex chip like cell is not going to scale that high nor was it meant to

Like V3 already stated it, Multi-Core CPU are perfect for high clock speed.
Also, some of the engineers working on Cell architecture did work on project like GUTS, which was a High-Speed (GHz) architecture.
So, the expected speed that PS3 Cell would have to achieve might be quite high. IMO.

 

[Vince]

Also, some of the engineers working on Cell architecture did work on project like GUTS, which was a High-Speed (GHz) architecture.
So, the expected speed that PS3 Cell would have to achieve might be quite high. IMO.

Hopefully he'll listen when you say it. I've already stated countless times that many of the engineers were taken for their experience on BOA, guTS and power-aware designs. Many of them have been retasked after Cell to IBM's work on Ultra High-Frequency Architectures for use in future generation Power.

 

[jvd]

I guess we will see soon enough eh ?

 

[cthellis42]

AND THEN YOU'LL ALL PAY! OH YES, YOU'LL ALL PAY FOR LAUGHING AT ME!!!!!

 

[jvd]

AND THEN YOU'LL ALL PAY! OH YES, YOU'LL ALL PAY FOR LAUGHING AT ME!!!!!

No we wont . We will continue to laugh at u .


Btw in the new time magzine they talk about the nintendo ds and they talk about blueray . They say blueray is going to cost 1k and they mention hd-dvd comingo ut to compete . No mention of psp though

 

[cthellis42]

No we wont . We will continue to laugh at u .

Well yes, but for DIFFERENT reasons. I have a whole bunch!

 

[Deepak]

Link

 

[Fafalada]

"I guess we will see soon enough eh ?"

Or will we?

 

[MrSingh]

so how much local memory (embedded memory) do you guys expect for each SPU (APU)?

 

[Fafalada]

so how much local memory (embedded memory) do you guys expect for each SPU (APU)?

Local storage was assumed at 128KB per SPU at this point. There may be eDram on the chip as well though - some people expect 32MB+.

Personally I am worried about the implication of there being eDram thouh(especially so much of it), because following certain design philosophies that would probably lead to a cacheles chip.
If that happened you might as well forget about sharing codebases with any other platforms. Although I guess it would be somewhat amusing watching the reactions/outcry that would follow.

 

[nAo]

so how much local memory (embedded memory) do you guys expect for each SPU (APU)?

128k is nice..IF it's more than that it would be even nicer emh...cough.. 8)

 

[nAo]

Personally I am worried about the implication of there being eDram thouh(especially so much of it), because following certain design philosophies that would probably lead to a cacheles chip.
If that happened you might as well forget about sharing codebases with any other platforms.

In some extents that what happen just now with projects that share code between PS2 and other platforms (such as XBOX in my personal experience).
PS2 needs special 'care'..so you find yourself spending a lot of time to tweak here and there (god bless #ifdef )to make your code not run very badly on the PS2..

ciao,
Marco

 

[ERP]

I think most people just write PS2 games and port them.
Good PS2 code generally won't run badly on other platforms.

I personally try to avoid the dreaded #ifdef wherever possible, IMO it's a maintenance nightmare. Of course it is just unavoidable in places.

 

[Fafalada]

(god bless #ifdef Wink)to make your code not run very badly on the PS2...

I have to second ERP here - lots of ifdefs is a fast way to make code unreadable, I try to avoid them when I can. But yeah, using stuff like XBox cpu code directly is the easiest way to break PS2 performance.

I think most people just write PS2 games and port them.

Given how vast majority of PS2 multiplatform titles suffer from CPU performance issues, I have to wonder about that...

Good PS2 code generally won't run badly on other platforms.

That again depends what you consider good PS2 code.

 

[nAo]

Filling your code with #ifdef is bad..but an #ifdef can save your programmer life
(BTW..on classes that differs a lot between PS2 and XBOX we don't share the source code on the same source file)

 

[ERP]

Given how vast majority of PS2 multiplatform titles suffer from CPU performance issues, I have to wonder about that...

We were discussing this issue recently at work or rather lamenting the size of modern teams.

We pretty much do just write a PS2 version and have resources allocated to port to the other platforms (as we develop).

With 4 or 5 programmers you can monitor code quality pretty closely and make sure you don't screw yourself on PS2. My last game had >25 programmers working on it, you have to selectively maintain code quality, and there is a limit to what you can do.

And I'm not so much talking about the quality of the code (that's much less of an issue) as I am the PS2ness of the code, infact a lot of what are considered good programming practices will produce very poorly performing PS2 code.

But the real issue is the vast gulf in general CPU performance between PS2 and (XBox/GCube).

 

[rabidrabbit]

http://arstechnica.com/articles/paedia/cpu/POWER5.ars/5

POWER5 and Cell
I asked if there's any relationship between POWER5 and the Cell architecture that IBM is working on with Sony. Pattnaik didn't seem to familiar with the details of Cell, and he said that there's no relationship between the two designs. He noted that if they shared some similar characteristics, then it isn't because the two teams are collaborating in any way.

 

[Fox5]

Given how vast majority of PS2 multiplatform titles suffer from CPU performance issues, I have to wonder about that...

We were discussing this issue recently at work or rather lamenting the size of modern teams.

We pretty much do just write a PS2 version and have resources allocated to port to the other platforms (as we develop).

With 4 or 5 programmers you can monitor code quality pretty closely and make sure you don't screw yourself on PS2. My last game had >25 programmers working on it, you have to selectively maintain code quality, and there is a limit to what you can do.

And I'm not so much talking about the quality of the code (that's much less of an issue) as I am the PS2ness of the code, infact a lot of what are considered good programming practices will produce very poorly performing PS2 code.

But the real issue is the vast gulf in general CPU performance between PS2 and (XBox/GCube).

PS2 has much higher cpu performance?