Yehuda, I have tried underlocking a nvidia gpu on my latop longtime ago, I believe I can force a specific power states (P0 to Px) via the nvidia-smi interface. Is this what you used? Also, did you under-clock a specific card among the four? Thanks! On Thu, May 30, 2019 at 1:52 PM Yehuda Halfon <[log in to unmask]> wrote: > We also had this problem. > We fixed it in two ways. One is to run it as single run per card. > The other is to undercloack the GPU, you can do it through the driver > setup under root. > ------------------------------ > *From:* Collaborative Computational Project in Electron cryo-Microscopy < > [log in to unmask]> on behalf of Abhiram Chintangal < > [log in to unmask]> > *Sent:* Thursday, May 30, 2019 11:39:48 PM > *To:* [log in to unmask] > *Subject:* Re: [ccpem] Threadripper Workstation Build Stability Issues > > Clara, > > In my case, the system just powers off. So I am quite puzzled. > > I hooked up a watt-meter to the machine to see if I am stressing the power > supply, but it doesn’t look like it. > > I am yet to check if there is a new bios for the board( Gigabyte X399 > Aorus Extreme) > > Anything specific to PCie settings that I should be watching out for? > > Thanks! > > Abhiram > > > > > On Thu, May 30, 2019 at 1:22 PM Dr. Clara Cai < > [log in to unmask]> wrote: > >> You mean the operation system crashes or RELION is crashing? >> Have you checked if the motherboard BIOS has a new version? Pay attention >> to the memory and PCIe settings especially. Good luck! >> >> Clara >> SingleParticle.com >> >> >> On Thu, May 30, 2019 at 11:48 AM Abhiram Chintangal < >> [log in to unmask]> wrote: >> >>> Hey all, >>> >>> I am troubleshooting one of our new thread ripper based workstations >>> which crashes when using all four GPU's with Relion. The machine is running >>> a thread-ripper 2990x with 4x2080ti's and its powered by a 1500W power >>> supply. >>> >>> Anyone here run into something similar? >>> >>> Initially I was under the impression that its a power-supply problem, >>> but the machine seems quite stable when running gpu/cpu bechmarks together >>> that saturate all the cores and gpu's at the same time. >>> >>> To rule out memory, I also ran memtest86 on the machine. >>> >>> Any ideas are appreciated? >>> >>> Thanks! >>> >>> -- >>> >>> Abhiram Chintangal >>> QB3 Nogales Lab >>> Bioinformatics Specialist @ Howard Hughes Medical Institute >>> University of California Berkeley >>> 708D Stanley Hall, Berkeley, CA 94720 >>> Phone (510)666-3344 >>> >>> >>> ------------------------------ >>> >>> To unsubscribe from the CCPEM list, click the following link: >>> https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1 >>> >> -- > Abhiram Chintangal > QB3 Nogales Lab > Bioinformatics Specialist @ Howard Hughes Medical Institute > University of California at Berkeley > 708D Stanley Hall, Berkeley, CA 94720 > Phone (510) 666-3344 > > ------------------------------ > > To unsubscribe from the CCPEM list, click the following link: > https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1 > -- Abhiram Chintangal QB3 Nogales Lab Bioinformatics Specialist @ Howard Hughes Medical Institute University of California Berkeley 708D Stanley Hall, Berkeley, CA 94720 Phone (510)666-3344 ######################################################################## To unsubscribe from the CCPEM list, click the following link: https://www.jiscmail.ac.uk/cgi-bin/webadmin?SUBED1=CCPEM&A=1