Cell CPU as the next thing?

[freqmax]

What do you think about using a Cell CPU that has 30x FPU performance to that of a Intel PC as the CPU for a computing platform?

[AJCopland]

Cell is a dead end.

There might be another one developed by Toshiba or IBM (it was a consortium effort) but I doubt it now as I think they've moved on to other designs. At best you're likely to see process shrinkage of the existing design just like the PS3 has used to reduce cost but I dunno if anyone would be able to buy them.

[spirantho]

From where do you get this "30x FPU performance" figure? That sounds a little... unbelievable.
Anyway, Cell is not appropriate for a desktop machine, it's designed for a different task.

[commodorejohn]

I'm not an expert on the matter, but Cell strikes me as one of those things like Itanium where it's just too damn complicated to bother with. Software-directed caching sounds good in theory, and if you're willing to tweak it properly I'm sure it can run very well indeed, but for normal software development, you'd need a pretty freaking smart compiler to make the most of it.

Besides, FPU performance is only one small part of overall system performance.

[danbeaver]

I don't want to get into another argument, but like parallel processing, would not implimenting a really useful compiler be so difficult as to be avoided.  Great at one problem (Photoshop plug-in, playing chess, or acting as a server), but near impossible to apply across the board.

[persia]

Last I read the i7  had caught up to and passed the cell chip in terms of performance.  Cell CPUs are a dead end.

[danbeaver]

Does this topic really belong int the Amiga Hardware Section?

[gertsy]

Does this topic really belong int the Amiga Hardware Section?

Nup.  : http://www.nasa.gov/mission_pages/msl/multimedia/gallery-indexEvents.html

[Hattig]

What do you think about using a Cell CPU that has 30x FPU performance to that of a Intel PC as the CPU for a computing platform?

Cell has around 200 GFLOPS of single-precision floating point power. If you code for them explicitly (you can't just recompile standard FP code). Impressive for its time, but irrelevant today when you can get GPUs 20x faster.

A modern Intel CPU's HD4000 graphics can achieve up to 294 GFLOPS, if you are happy coding in OpenCL. The general purpose AVX isn't a slouch either.

[persia]

I think the OP was trying to compare cell technology to long dead P4 PCs...

[freqmax]

Thought reading - FAIL

I were just thinking of which CPU that would be real neat as a power platform for a low price.

I suspect however CPU:s aren't the technology that will make things possible that ain't currently in the price segment.

The paper: Soft-DVB: A Fully-Software GNURadio-based, ETSI DVB-T Modulator writes about the performance advantage of the Cell CPU:

On the Cell BE platform, the accessible computing power will be approximately 3*10^10 floating point operations per second (flops), which is approximately 30 times the computing power of the machine used to test the current version of Soft-DVB.

As for the current processor:

3.0-GHz Intel Pentium IV processor, draining 83% of its computational power.

[Louis Dias]

Problem is FPU code is only 10% of the code in games and even less so in general applications...

[ferrellsl]

Well, if a cell CPU that performed that well  actually existed, it might be the basis for a decent computing platform, but it doesn't exist and there's no one working on such a project anyway.  Maybe in 2003 a cell processor would have had a chance as a desktop CPU, but today, GPU computing can outperform ANY cell design.

[freqmax]

Any desktop applications that would be possible given a different but cheap platform?

[takemehomegrandma]

Maybe in 2003 a cell processor would have had a chance as a desktop CPU

Cell was never meant (or suitable) as a general desktop CPU, not then, not now.