<div>Hi Jörg,</div><div>Assuming that you want to run on a single GPU, I think you can fit this benchmark:</div><div><br></div><div>https://github.com/cp2k/cp2k/blob/master/benchmarks/QS_DM_LS/H2O-dft-ls.NREP2.inp<br></div><div><br></div><div>Note that the problem is not the time, rather the memory available.</div><div>Assuming that you can run on a single GPU, then the test should run in a reasonable amount of time (minutes), no need to run 1h.</div><div>Make sure you use libxsmm for the CPU execution so that you can make a fair comparison CPU-GPU executions.</div><div><br></div><div>Time ago I gave some explanations on how to understand what the performance benefit of the GPU is:</div><div><br></div><div>https://github.com/cp2k/cp2k/issues/73<br></div><div><br></div><div>Let me know if you need anything else (or write to me directly, I don't use to read the forum so often)...</div><div><br></div><div>Alfio</div><div><br></div><div><br></div><div><br></div><br><br><div class="gmail_quote"><div dir="auto" class="gmail_attr">Il giorno giovedì 29 ottobre 2020 alle 17:23:13 UTC+1 sassy ha scritto:<br/></div><blockquote class="gmail_quote" style="margin: 0 0 0 0.8ex; border-left: 1px solid rgb(204, 204, 204); padding-left: 1ex;">Hello Alfio,
<br>
<br>Many thanks! I am not after the last bit of performance but more to get an
<br>idea which of the cards I can use would be better suited for us.
<br>If you don't mind: can I run any DFT based job to try that out or are there
<br>some which are better suited? I was thinking of one which is running say 1h or
<br>so on a CPU. Too short and you might get too much noise, too long and I don't
<br>get as much testing done as I would like to.
<br>
<br>I will have a look at the github page.
<br>
<br>All the best from a wet London
<br>
<br>Jörg
<br>
<br>Am Donnerstag, 29. Oktober 2020, 06:19:21 GMT schrieb Alfio Lazzaro:
<br>> Hello Jörg,
<br>> Although CP2K doesn't A100 optimized kernels, you can always use the V100
<br>> ones. I don't expect a big performance impact.
<br>> But if you really want to, please check here on how to optimize for
<br>> A100:
<br>> <a href="https://cp2k.github.io/dbcsr/develop/page/3-developer-guide/3-programming/2" target="_blank" rel="nofollow" data-saferedirecturl="https://www.google.com/url?hl=it&q=https://cp2k.github.io/dbcsr/develop/page/3-developer-guide/3-programming/2&source=gmail&ust=1604392854548000&usg=AFQjCNFvao7aZYLsE8TtykQGEd8_1qlx3g">https://cp2k.github.io/dbcsr/develop/page/3-developer-guide/3-programming/2</a>
<br>> -accelerator-backend/2-libsmm_acc/3-tune.html
<br>>
<br>> Best regards,
<br>>
<br>> Alfio
<br>>
<br>> Il giorno mercoledì 28 ottobre 2020 alle 23:26:28 UTC+1 sassy ha scritto:
<br>> > Hi Fabian, hi all.
<br>> >
<br>> > I am hijacking a bit this thread as I think it is relevant.
<br>> > I will have access to a GPU test machine which got the new Ampere GPU
<br>> > cards
<br>> > installed. If I get the Wiki page correctly, I would need to set the GPU
<br>> > version to A100, i.e.
<br>> > --gpu-ver=A100
<br>> >
<br>> > Are the Ampere cards already supported in CP2K? Given that the latest
<br>> > version
<br>> > was released *before* the Ampere cards were around, I got the feeling that
<br>> > might not be the case.
<br>> >
<br>> > Thanks for your help.
<br>> >
<br>> > All the best from a dark and cold London
<br>> >
<br>> > Jörg
<br>> >
<br>> > Am Dienstag, 18. August 2020, 08:41:34 GMT schrieb <a href data-email-masked rel="nofollow">fa...@gmail.com</a>:
<br>> > > Hi Bidesh,
<br>> > >
<br>> > > this option will select the which architecture to optimize for. You can
<br>> > > simply choose the one which is closest to you card. According to
<br>> > > <a href="https://en.wikipedia.org/wiki/CUDA" target="_blank" rel="nofollow" data-saferedirecturl="https://www.google.com/url?hl=it&q=https://en.wikipedia.org/wiki/CUDA&source=gmail&ust=1604392854548000&usg=AFQjCNGB1dMhE6eISc4k2psv35PtOp5fVg">https://en.wikipedia.org/wiki/CUDA</a> the GTX 1660 Ti has compute
<br>> >
<br>> > capability
<br>> >
<br>> > > 7.5, so you should use --gpu-ver=V100
<br>> > >
<br>> > > Best,
<br>> > > Fabian
<br>> > >
<br>> > > On Tuesday, 18 August 2020 at 09:23:08 UTC+2 <a href data-email-masked rel="nofollow">bide...@gmail.com</a>
<br>> >
<br>> > wrote:
<br>> > > > Hi all,
<br>> > > >
<br>> > > > I don't know if this topic is relevant or not but I am facing a
<br>> > > > problem
<br>> > > > while compiling cp2k with cuda.
<br>> > > > I have a GPU nVIDIA GTX1660ti. I have installed CUDA. Now while
<br>> >
<br>> > compiling
<br>> >
<br>> > > > what should I put in --gpu-ver=. As its showing only K20X, K40, K80,
<br>> >
<br>> > P100,
<br>> >
<br>> > > > V100 are allowed.
<br>> > > > Thanks in advance.
<br>> > > >
<br>> > > > Bidesh Kirtania
<br>> > > > Research Scholar
<br>
<br>
<br>
<br></blockquote></div>