Designed for the PlayStation 3, Sony, Toshiba and IBM’s new “Cell processor” promises seemingly obscene computing capabilities for what will rapidly become a very low price. In this 5-part article you can look at what the Cell architecture is, look at the profound implications this new chip has, not for the games market, but for the entire computer industry. Has the PC finally met it’s match?
Cell Architecture Explained
About The Author
Eugenia Loli
Ex-programmer, ex-editor in chief at OSNews.com, now a visual artist/filmmaker.
Follow me on Twitter @EugeniaLoli
Actually, given a 4 fpus per APU at 4.6 GHz (if they can even pull that off [yes, I saw the picture, goals are VERY different from reality]) that would get them to a theoretical peak of 18.4 GFLOPS, which, as we all know, nobody really reaches theoretical peaks, so we’re ever so slightly shy of the 32 GFLOPS stated in the patent.
They are counting 32bit Multiply-Adds as 2 operations which will give 8 operations per cycle – giving 36.8 GFlops at 4.6GHz.
Ok, but then how does it handles operations on very large files (size > 64 Mo) ? Does the programmer has to do all the work of loading/unloading parts of the file in memory from disk ?
The patent indicates an “ideal” of 64MB, this may be hardwired in the PS3 but it does not specify an absolute limit
Workstations will be built with a general purpose cpu and the Cell as a coprocessor for DSP, audio, video, compression etc. This guy makes it sound like IBM developed the POWER5 and is developing the POWER6 for nothing.
POWER5 and POWR6 etc. are for different makets. Also I explicitly suggested Appe use Cells as Co-Processors along side G5/G6s.
Emulation wont work on these systems anyway the APU’s by the sounds of it would be totally useless for emulation. As for the PU its just a modified PPC. Basically the author is way off on his emulation predictions.
There seems to be some confusion over this. Yes I agree the APUs are not likely to be much use for emulation. However, as I said I expect the PC to off load the heavy processing to APUs so a fast x86 should no longer be necessary, the PU would *then* be fine for emulating an x86, especially with a JIT based emulator.
I would have thought that anybody looking at the history of computing for the last 25 years would quickly come to the conclusion that merit – having the technically better product – doesn’t count for much.
I took rather more words but thats what I said and gave specific examples.
—
The memory protections system also seems to be misunderstood. THE PU will likely have a normal memory protection systems with paging as it’ll be dealing with more complex operations (i.e. running an OS).
The APUs will work on specific chunks of RAM set aside specifically. They can either access it or not, nothing more. There is no paging system for the APUs as there is no need for it.
Glad that that’s cleared up.
Anyone want to learn a little about streaming processors can check this link out. Professor Dally urging congress to drop the cheap mainstream intel/ibm chips and go with steaming chips(vectors) but with more memory.
http://radio.weblogs.com/0105910/2003/12/01.html
There is more hype in the video game industry than anywhere else. Go back to 1992 and Nintendo announcing that they were developing a new system that they said would rival the most powerful graphics supercomputers on the market (Project Reality). Trilinearmipmapinterpolation!!!! 150,000 fully textured, shaded polygons per second!!! (yeah right). I don’t think anyone would argue that what eventually became the Ultra 64 and later Nintendo 64, with its 4-polygon trees in 320×240 and incredible 4MB of RAM could really come close to an SGI Onyx Reality Engine. It’s all hype, it’s intended to stir people up and get them interested in the new machine of the month in what is a hyper-competitive market in which every system sale is crucial.
Another example: The 3D0 M2. Dual PowerPC 603e, 1 million polygons per second. A thoroughbred CD-ROM based N64-killer, and pure vaporware, with fudged press release screen shots of graphic effects like warping MPEGs to toruses. In reality it debuted only in the arcade and the graphics were barely, if at all, better than the PlayStation (One!!).
As if these examples aren’t enough, go back to the PS2, circa 1999. The Emotion Engine would revolutionize this and that, and the PS2 would have firewire and maybe a hard drive and it would destroy the desktop computer etc. In reality the graphics were barely better than the Dreamcast which was already out.
Sony would like you to believe that the Cell will be bigger than the second coming of Christ. In reality, the Cell will be just another console CPU with decent performance and a beefed-up SIMD unit and that is that. The PS3 will be cool, as it should be since its predecessor is over 4 years old now, but don’t treat it like the Segway.
This thing looks to be real seXXXy for media and science applications but how well does it do xml. It’s of no imediate use to me if one apu takes four as long to do a simple xsl translation than a Pentium 4. 8 apus running at the same time may make up for it for a single user but if you’ve got 100 people hitting your server at the same time then the cell is of no real use to me at the moment.
maybe the problem is not with the cell chip but with the intel ISA crap we deal with.
To clear up a few things I added a clarifications section:
http://www.blachford.info/computer/Cells/Cell6.html
Simon, my issues is with the complete lack of substance and analysis in the article. This is not an “architecture explained” it is a dreamy, full of factual errors, wild guess.
Check out a rebuttal to one of his earlier articles…
http://www.azillionmonkeys.com/qed/apple.html
or some of the in-depth counterpoints on Slashdot (amongst some of the usual crap)…
http://developers.slashdot.org/article.pl?sid=05/01/21/022226&tid=1…
Should I even go into his articles on PPC/RISC vs. x86/CICS and the others on his site?
You are right that I didn’t provide info, but I’d rather have someone provide NO info then bogus assumptions.
Cell “analysis” a mixed bag (Ars Technica)
http://arstechnica.com/news.ars/post/20050124-4551.html?93593
and discussion…
http://episteme.arstechnica.com/eve/ubb.x/a/tpc/f/174096756/m/78800…
Cell “analysis” a mixed bag (Ars Technica)
And the rebuttal to it:
http://www.blachford.info/computer/Cells/Rebuttal.html
It’s *not* an “analysis”, I never said it was.