[Beowulf] Has anyone actually seen/used a cell system?

David Kewley kewley at gps.caltech.edu
Wed Sep 20 13:18:18 PDT 2006


On Wednesday 20 September 2006 10:51, Mark Hahn wrote:
> > Can anyone point me to a url, or tell me what their
> > experience is with this technology?  Is it as fast as
> > it's purported to be?
>
> I haven't come anywhere near a Cell, but then again, I'm not sure
> I'd want to.  14.6 Gflops (64b, and assuming the full 8 SPE's)
> isn't bad, but then again, a 3 GHz Core2 dual-core is 24 Gflops,
> and almost certainly a lot more accessible, shipping now, runs linux,
> supported by compilers and goto-blas, etc.

Yup, that's an appropriate point of skepticism.  We'll have to see what 
works out in practice.

There is a study that says that the Cell has large speedups in real 
double-precision code, compared to Opteron, Itanium2, and Cray X1E:

http://www.cs.berkeley.edu/~samw/projects/cell/EDGE06_abstract.pdf

Related pages:

http://www.google.com/search?hl=en&lr=&q=+site:www.cs.berkeley.edu+cell+double+precision

Also see the following for work on tuning code to use single precision where 
possible, thus attaining an even bigger speedup:

http://icl.cs.utk.edu/iter-ref/

Mercury Computer Systems is designing (and perhaps even shipping 
pre-production units of) a Cell blade system.  IBM is also going to offer 
Cell blade systems.  I am talking to someone who is getting yet an other 
Cell-based company off the ground, who I believe is aiming for HPC.

So there's quite a bit of potential, and I think it's looking pretty likely 
that we'll see real, affordable Cell-based high-GFLOPS/Watt HPC systems in 
the coming year, and that these systems will support high performance on 
double precision code.  Right now though, I'm still in wait-and-see mode.

David



More information about the Beowulf mailing list