Dear Valerio,
The answer will depend on which part of SPM you are using to analyse
your data. Some steps will scale with the number of cores without much
problem, while you have to take into account access to the filesystem
for others (and memory requirements).
As for an Intel vs AMD comparison, I currently only have access to Intel
CPUs so I cannot really comment - if others have benchmarks to share,
this would be useful. There was a "configuration issue" with the MKL on
AMD for some MATLAB versions but it should be a problem of the past:
https://www.extremetech.com/computing/308501-crippled-no-longer-matlab-2020a-runs-amd-cpus-at-full-speed
Best regards,
Guillaume.
On 14/05/2021 13:18, Valerio Rubino wrote:
> Hi everyone.
> I was thinking to parallelize the execution of SPM batches with MATLAB's
> Parallel Computing toolbox.
> However, I would like to know how speed improvement scales up with
> number of CPU cores; is it worth to opt for a CPU with higher number of
> cores (i.e. AMD over intel)?
> Lastly, apart from number of cores, is there any reason to prefer either
> AMD or Intel to SPM12 on a laptop?
>
> Thanks,
> Valerio
--
Guillaume Flandin, PhD
Wellcome Centre for Human Neuroimaging
UCL Queen Square Institute of Neurology
London WC1N 3BG
|