[Eeglablist] Hardware recommendations

[Malte Anders]

Hi to both of you and thanks for the input!

I do also agree with you that I will probably not profit from more cores as
newtimef or EEGLAB in general does not seem to support multiple cores. CPU
utilization climbs to a steady ~40% and stays there. Lowering from 200 to
100 datapoints does however hugely shrinks ram utilization.

Your suggestions are well noted and I will keep them in mind when subject
recruitment is finished and data analyzation starts. I am unsure however if
all of those approaches are applicable but thank you very much anyways! :)

Another approach I am going to try is to only use EEGLAB for preprocessing
(filter, ASR, maybe ICA - not sure yet, EEGLABs study functions are
probably not usable for my study design) and then move to raw MATLAB code.
Maybe if I write my own pwelch-script I can speed things up, but time will
tell.

I have also managed to request access to a virtual machine with 128 Gb and
8 cores with no dollar spent, I'll see how that works.

For now I am glad I am not the only one with that problem.

Kind regards,
Malte

Am Di., 14. Jan. 2020 um 21:27 Uhr schrieb Makoto Miyakoshi <
mmiyakoshi at ucsd.edu>:

> Dear Malte,
>
> If you run Matlab's wavelet toolbox (or spectrogram() from signal
> processing toolbox) you'll be surprised how fast it goes...
> So the main issue of EEGLAB's slow speed is not so much on the hard ware
> but on code. I doubt if the current EEGLAB solution is capable of parallel
> computing using multiple CPUs.
>
> There are few methods to multiplies the processing speed without buying a
> new hardware.
> If you have subj001-subj100, you create let's say 4 EEGLAB STUDYs using
> 001-025, 026-050, 051-075, 076-100. Process them with the same precompute
> parameters. The output files on HDD is valid when you create a single STUDY
> using 001-100--I mean you can skip precompute, but immediately step to
> precluster. This is an example to gain 4X speed. If your CPU has 4 physical
> cores (not counting hyperthreding processes), you can speed up the process
> by launching the same number (4) of MATLAB.
>
> In doing this, make sure you use the same precompute parameters. Otherwise
> you can't build the final single STUDY.
>
> Makoto
>
> On Mon, Jan 13, 2020 at 11:20 PM Malte Anders <malteanders at gmail.com>
> wrote:
>
> > Hi everybody,
> >
> > I have collected huge amounts of EEG data with 512 Hz sampling frequency
> > and 34 electrodes.
> >
> > For one subject (~2-3 hours of data), a simple Time/Frequency
> decomposition
> > between 1 and 40 Hz with standard settings calling the newtimef function
> > takes approx. 8 hours for 1 hour of recorded data on my Computer (4790k,
> 32
> > Gb of DDR3 Ram, Intel HD GPU), so 16-24 hours for one dataset in total.
> > This is simply too long as I am currently working alone on this. CPU
> > utilization is ~40% during the process, RAM utilization is 100% (so 100%
> of
> > the 32 Gb are used).
> >
> > I would like to invest in new hardware (I am also considering used
> > hardware) with a price point around 1500-3000€. Which would make more
> > sense:
> > -A new Ryzen 3950x, 16 cores with 128 Gb of DDR4 RAM (2666 Mhz) ~ 1500€
> > -A used workstation, e.g. with Dual CPU: 2 x 12 core Xeon E52670v3 and
> 256
> > Gb RAM ~ used approx. 2200€
> >
> > The Xeon has more cores, but way lower clock speed. It does have double
> the
> > amount of RAM though.
> >
> > Would you recommend a dedicated GPU (Nvidia?) for this?
> >
> > Any recommendations are greatly appreciated. Thanks!
> > Malte
> > _______________________________________________
> > Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> > To unsubscribe, send an empty email to
> > eeglablist-unsubscribe at sccn.ucsd.edu
> > For digest mode, send an email with the subject "set digest mime" to
> > eeglablist-request at sccn.ucsd.edu
> _______________________________________________
> Eeglablist page: http://sccn.ucsd.edu/eeglab/eeglabmail.html
> To unsubscribe, send an empty email to
> eeglablist-unsubscribe at sccn.ucsd.edu
> For digest mode, send an email with the subject "set digest mime" to
> eeglablist-request at sccn.ucsd.edu



-- 
Mit freundlichen Grüßen,

Malte Anders