IBM, Sony, Toshiba unveil nine-core Cell processor
#3
Posted 07 February 2005 - 02:42 PM
As far as gaming goes, the mini will probably not score as well as X-Box. The simple fact is that even though the X-Box runs at 750MHz, there's a lot less threads running on a graphics console than on a full-fledged computer system. Also, the console systems are specifically tuned and designed to facilitate gaming functions. There's no OS overhead, either. This frees up a ton of resources so that even though a spec-inferior product can turn out superior performance for specialized functions.
#4
Posted 07 February 2005 - 02:50 PM
There's a lot more to gaming or any kind of computation than just what processor is being used. A Mac with a Cell processor would no more be able to run PS 3 games than a Mac now can run GameCube games -- GameCubes use modified PowerPC chips and ATI graphics chips.
What's interesting here is that with GameCube, PS 3 and Xbox "next" (lately rumored to be named Xbox 360), the manufacturers have gone to IBM PowerPC-derived technology. Good for IBM. It'll be interesting to see if Nintendo stays the PowerPC course with its Revolution console, which we'll purportedly get a look at during E3 in Los Angeles this May.
What's interesting here is that with GameCube, PS 3 and Xbox "next" (lately rumored to be named Xbox 360), the manufacturers have gone to IBM PowerPC-derived technology. Good for IBM. It'll be interesting to see if Nintendo stays the PowerPC course with its Revolution console, which we'll purportedly get a look at during E3 in Los Angeles this May.
#6
Posted 07 February 2005 - 04:01 PM
I could see some kind of derivative of this chip being quite helpful to the Mac market--the Mac, with all of its animations, uses floating points quite a bit. And the professional market (esp XSserves), Final Cut Pro, Motion, etc probably do a ton of float point calculations (the main "bragging" spec for supercomputers, the Linpack, I believe is a floating point test). I'm not as familiar with compressions (MPEG), so I don't know if they would gain any boost from extra floating point power, but a lot of the "power hungry" aps (pro and consumer) would likely benefit from more floating point processing power. Of course, any OpenGL-based apps (such as many games) would greatly benefit
#8
Posted 07 February 2005 - 04:22 PM
I think the fact that it will run at 4.5GHz+ and only 30 watts is good news for Apple in 2006. I would expect Apple to use the PowerPC core with Altivec and not include the 8 vector processors of the Cell, but who knows.
There has been more info about this online for a little while here.
There has been more info about this online for a little while here.
#9
Posted 07 February 2005 - 06:24 PM
There is more that we don't know about Cell than what we do know. I think a few people are jumping the gun a bit and assuming this is intended to be used to replace a traditional CPU. I'm not sure that's the case. From every thing I've read so far, the Cell processors can be linked together and do great on floating point processing. Basically, a program that makes use of multiprocessing and perhaps Altivec now, will do well with Cell. I'm not sure most other common programs would benefit much. Let's wait and see before we jump to conclusions. I suspect the application of the Cell processor will be more specialized than we think.
Steve
Steve
#10
Posted 07 February 2005 - 06:32 PM
IBM and its friends are really out to "get" the WinTel near monopoly based on X86 chips. The regular PowerPC chip has only slightly cut into that dominance. What is needed is a category shift, and we have the potential here. This chip is revolutionary, not just evolutionary. And what will the revolution look like?
Imagine a Japanese businessman picking up his Sony phone to call his business partner in America. The Japanese speaks Japanese, and the American hears American English...in real time! Or the Japanese could call his other business associates in China, or India, or wherever. Each party would enjoy instant language translation available only with a chip such as this.
Imagine a Japanese businessman picking up his Sony phone to call his business partner in America. The Japanese speaks Japanese, and the American hears American English...in real time! Or the Japanese could call his other business associates in China, or India, or wherever. Each party would enjoy instant language translation available only with a chip such as this.
#11
Posted 07 February 2005 - 07:10 PM
In reply to:
Imagine a Japanese businessman picking up his Sony phone to call his business partner in America. The Japanese speaks Japanese, and the American hears American English...in real time! Or the Japanese could call his other business associates in China, or India, or wherever. Each party would enjoy instant language translation available only with a chip such as this.
Imagine a Japanese businessman picking up his Sony phone to call his business partner in America. The Japanese speaks Japanese, and the American hears American English...in real time! Or the Japanese could call his other business associates in China, or India, or wherever. Each party would enjoy instant language translation available only with a chip such as this.
While I don't doubt something like that will happen eventually in the future, you are priming yourself for a big disappointment if you think Cell chips will make that possible. We're having enough trouble translating written stuff, but to add voice recognition on top of it?
Besides, linguistic differences between some languages (such as subject-verb-object vs subject-object-verb) practically ensures that fully real-time translation cannot happen.
#12
Posted 07 February 2005 - 07:29 PM
Well, this chip doesn't look "revolutionary" to me. Looks like a special-purpose piece of machinery made by bolting some new architecture SIMD/vector machinery around a lonely (and undoubtedly non-standard) PowerPC core.
The thing to remember when someone trumpets a brand new computer architecture is that the one thing you're guaranteed to have is that it won't run any existing binaries.
The hoopla around Cell suggests that a lot of thought has gone into its definition, but it's no more a 'general purpose processor' than anything from ATI or Nvidia (not to knock any of these - they're fine niche machinery, well-tuned to their applications).
There are very few 'revolutionary' ideas or architectures around. Most new ideas just recapitulate old ideas, and arise because the fashion has changed, the market has changed, or the number of transistors economically available has changed. (Or someone thinks that one of these have changed).
Oh well. Expect the BS to continue for a while.
-- Pete
The thing to remember when someone trumpets a brand new computer architecture is that the one thing you're guaranteed to have is that it won't run any existing binaries.
The hoopla around Cell suggests that a lot of thought has gone into its definition, but it's no more a 'general purpose processor' than anything from ATI or Nvidia (not to knock any of these - they're fine niche machinery, well-tuned to their applications).
There are very few 'revolutionary' ideas or architectures around. Most new ideas just recapitulate old ideas, and arise because the fashion has changed, the market has changed, or the number of transistors economically available has changed. (Or someone thinks that one of these have changed).
Oh well. Expect the BS to continue for a while.
-- Pete
#13
Posted 07 February 2005 - 07:54 PM
I'm sure we'll never see this specific chip in a Mac but doesn't anyone else here remember Sony's president making the cryptic statement about "maybe" working on "other things" with Apple at the recent SteveNote? Maybe this is a hint at the direction those other things might be heading towards.
#14
Posted 07 February 2005 - 08:08 PM
I don't know if Apple will buy into this, but the U.S. Department of Energy sure has.
Bluegene L
IBM is already building the most powerful computer cluster on the planet based on a 64 processor core, this looks like a 1/8th portion spinoff of that technology. Should Apple choose to go this route, you could speculate on one processor being dedicated to translating X86 code to PPC code and other core units running Windows code at speeds that would shame any current Intel processor and still have enough processors left over to engage in any number of other endeavors. Likewise, you could envision Apple going for the mid-range supercomputer cluster market using this core. Inasmuch as operating system technoligies lag behind processor evolution (can you say Longhorn?), I don't think Apple will utilize this in any product line in the forseeable future, but they're sure to try it out in their lab when the cores become readily available.
Bluegene L
IBM is already building the most powerful computer cluster on the planet based on a 64 processor core, this looks like a 1/8th portion spinoff of that technology. Should Apple choose to go this route, you could speculate on one processor being dedicated to translating X86 code to PPC code and other core units running Windows code at speeds that would shame any current Intel processor and still have enough processors left over to engage in any number of other endeavors. Likewise, you could envision Apple going for the mid-range supercomputer cluster market using this core. Inasmuch as operating system technoligies lag behind processor evolution (can you say Longhorn?), I don't think Apple will utilize this in any product line in the forseeable future, but they're sure to try it out in their lab when the cores become readily available.



Sign In
Register
Help


MultiQuote