IBM-PowerPC
Some interesting info seems to be in this personal webpage:
<a href="http://www.tubbs.net/main.html"; target="_blank">http://www.tubbs.net/main.html</a>;
<a href="http://www.tubbs.net/main.html"; target="_blank">http://www.tubbs.net/main.html</a>;
Comments
<strong>Some interesting info seems to be in this personal webpage:
<a href="http://www.tubbs.net/main.html"; target="_blank">http://www.tubbs.net/main.html</a></strong><hr></blockquote>;
Don't get too excited just yet.
A quote from that guy's resume:
[quote]
PowerPC microprocessor/FPU development ? IBM Blue Gene/L project
ASE Design Engineer ? maintained and debugged Verilog logic in the ASE (Arithmetic, Storage and Exception) unit of the SIMD/FPU extension to the PowerPC 440 used in proposed 200 teraflop supercomputer
PowerPC assembly language programming for microprocessor testing/debugging
<hr></blockquote>
Bye,
RazzFazz
[ 08-02-2002: Message edited by: RazzFazz ]</p>
:eek:
I would sure love to run some GA problems on that machine
The PPC 440 isn't particularly fast, so they must have had an enormous number of them! I wonder if they were experimenting with the IBM technology for stacking chips (which is called the "Cell" technology, I believe)? That would explain why they used a low power embedded chip.
[ 08-02-2002: Message edited by: Programmer ]</p>
This is an interesting research article detailing the future of chip design and it even gives some history of past designs and why they were implemented. the natural progression shows a cell type processor with many processors running alongside each other.
For those interested....
<a href="http://news.com.com/2100-1040-873337.html"; target="_blank">http://news.com.com/2100-1040-873337.html</a>;
WHAT is "new" news though... Moto still can't get their act together (is that old news too?)
<a href="http://www.eetimes.com/semi/news/OEG20020801S0078"; target="_blank">http://www.eetimes.com/semi/news/OEG20020801S0078</a>;
<strong>I believe SONY has recently signed a deal, or at least made it publicly known that they're going to port all of their PC games to the Mac. It seemed kind of odd at first but now it makes a bit of sense. If the PS3 is going to use a PowerPC chip then this deal/arrangement gives SONY a good headstart on learning how to program to the chip's strengths.</strong><hr></blockquote>
I don't think so.
This would only really help them if they were actually writing a large part of their games in straight assembly. Even then, the hardware (sound, gfx) and software (OS, libs, APIs) environments are just too different between both architectures.
Bye,
RazzFazz
<strong>WHAT is "new" news though... Moto still can't get their act together (is that old news too?)
<a href="http://www.eetimes.com/semi/news/OEG20020801S0078"; target="_blank">http://www.eetimes.com/semi/news/OEG20020801S0078</a></strong><hr></blockquote>;
This just means they can already build too much ... it doesn't mean that their act isn't together. What sense would it make to bring the fab online if there was nothing for it to build? Running those things is expensive.
Besides, that fab is only for 0.35 - 0.25 micron processes.
<strong><a href="http://www.research.ibm.com/journal/rd/462/nair.html"; target="_blank">http://www.research.ibm.com/journal/rd/462/nair.html</a>;
This is an interesting research article detailing the future of chip design and it even gives some history of past designs and why they were implemented. the natural progression shows a cell type processor with many processors running alongside each other.</strong><hr></blockquote>
Thanks, that is a very interesting article. If there are any skeptics about IBM out there, they ought to read that...
<strong>I believe SONY has recently signed a deal, or at least made it publicly known that they're going to port all of their PC games to the Mac.</strong><hr></blockquote>
I haven't found anything online to support this, though it's not beyond the realm of possibility; developers are learning it is possible to make money writing games for the Mac.
<strong>
I don't think so.
This would only really help them if they were actually writing a large part of their games in straight assembly. Even then, the hardware (sound, gfx) and software (OS, libs, APIs) environments are just too different between both architectures.
Bye,
RazzFazz</strong><hr></blockquote>
1) Writing for a console often means using assembly.
2) I would think optimizing for Altivec would be cross-compatible.
3) I was just hypothesizing.
<strong>
1) Writing for a console often means using assembly.
2) I would think optimizing for Altivec would be cross-compatible.
3) I was just hypothesizing.
Porting to the Mac would also force them to be big-endian compatible, which is often one of the biggest stumbling blocks to portability.
<strong>
1) Writing for a console often means using assembly.
</strong><hr></blockquote>
Is that true even today?
Obviously, I'm not a console programmer, but I definitely wouldn't have thought so (aside from performance tweaks in time-critical sections, see below).
[quote]<strong>
2) I would think optimizing for Altivec would be cross-compatible.
</strong><hr></blockquote>
Well, yeah, you could probably share some inner loops and other highly performance-critical sections (if the PS3-PPC will indeed be AltiVec-compatible, that is). But I'd guess that such optimizations are only a rather small part of the whole process of creating a game.
Bye,
RazzFazz
[ 08-02-2002: Message edited by: RazzFazz ]</p>
<strong>
I haven't found anything online to support this, though it's not beyond the realm of possibility; developers are learning it is possible to make money writing games for the Mac.</strong><hr></blockquote>
<a href="http://www.apple.com/games/articles/2002/01/thestation/"; target="_blank">This</a> may be what you're looking for. Not quite all of Sony's games, but I think this is the closest thing to what he's referring to.
The way IBM and Sony marketdroids described it in the press made it sound lake it was a brand new, revolutionary architecture.
<strong>So the "Cell" processor is PowerPC-based, eh? ...The way IBM and Sony marketdroids described it in the press made it sound lake it was a brand new, revolutionary architecture.</strong><hr></blockquote>
Cell is advertised as a broadband architecture. How does it sound to have a product that's PowerPC based (traditionally bandwidth starved) tossed in a mix where it finally has enough to feed it?
<strong>1) Writing for a console often means using assembly.</strong><hr></blockquote>
If the console APIs / compilers suck, that is. In my understanding, for example, the PS1 had nice APIs and thus a lot of people took joy from coding for it and the PS2 requires "too much assembly" which was one of the reasons it only had a few games at it's launch.
Anyhow, since consoles only have 1 configuration scenario (RAM, 3D chip, sound, etc) I think that the APIs can be optimized well enough already and the only assembly, like RazzFazz said, should be used with time-critical parts of the code.
<strong>
1) Writing for a console often means using assembly.
</strong><hr></blockquote>
The vast majority of console game code these days is in C/C++. The time to convert the assembly that is used isn't too significant -- the time to convert between the various graphics engines is much greater. Generally only the vector units and graphics engines get the assembly code, so it really boils down to the same thing anyhow... and even if they had the same processor the conversion would still be required.