The Intel chips are not as good at throughput as Opterons. You want to
optimize throughput if you are processing large datasets, such as fMRI
with SPM. Intel makes up for Opteron's superior memory controller by
throwing cache at the problem. Thus, increasing your cache is the best
way to make large, long jobs complete more quickly. For big long number
crunching jobs, large cache on the Intel chips will be more effective
to fast job completion than a faster CPU. Note that above 4 CPUs, extra
CPUs may bring diminishing returns, since for large jobs, getting the
data to the CPUs is the bottleneck with the Intel chips, not the work
done by the CPUs, and this bottleneck becomes serious above 4 CPUs. If
you can wait a couple months for the new Opteron Barcelona machines,
you may experience much better performance on jobs like this.
Jim
Himachandra Chebrolu wrote:
>We are upgrading our server to a 4 cpu-Xeon - Dell machine and
>I need to make a decision about which cpu's to get. The Xeon's
>now come with L3 cache (4mb, 8mb, or 16mb). The cost difference
>between four 2.6Ghz-4mb L3 and four 3.2Ghz-16mb L3 is 3,0000$. I've
seen
> seen discussions regarding other issues, but not this one. Many
>application will run faster with more L3, but many applications won't.
>
>Anyone have an opinion as how much the L3 will affect spm
>application processing times (we also run fsl, afni, analyze,
>......).
>
>Thanks for any input.
>
>Dave
>
|