Message boards :
Number crunching :
Problem with SoG Workunits on Nvidia GT710
Message board moderation
Author | Message |
---|---|
Cruncher-American Send message Joined: 25 Mar 02 Posts: 1513 Credit: 370,893,186 RAC: 340 |
Did you get the driver from Nvidia? I don't know about Linux, but from comments here in other threads, Win 10- installed drivers don't include the OpenCL add ons needed for SoG. |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
Yea a bit strange that cuda runs, but not SoG apps. I don't normally run SoG, but I know mine works fine on Beta. My notes for driver reinstall: I'm not sure what version you are using, but works in Mint and Ubuntu. On initial install, install proper NVidia drivers (not default) sudo apt-get purge 'nvidia*' sudo apt-get autoremove reboot Install the latest nvidia driver with the driver manager / software setup reboot |
Brent Norman Send message Joined: 1 Dec 99 Posts: 2786 Credit: 685,657,289 RAC: 835 |
It may be a different problem. <snip> Low-performance GPU detected, default period_iterations_num set to 500 OpenCL-kernels filename : MultiBeam_Kernels_r3584.cl INFO: can't open binary kernel file: /var/lib/boinc-client/projects/setiathome.berkeley.edu/ MultiBeam_Kernels_r3584.cl_GeForceGT710.bin_V7_SoG_37566, continue with recompile... Info : Building Program (binary, clBuildProgram):main kernels: OK code 0 INFO: binary kernel file created WARNING: can't open binary kernel file for oclFFT plan: /var/lib/boinc-client/projects/setiathome.berkeley.edu/ MB_clFFTplan_GeForceGT710_524288_gr64_lr16_wg256_tw0_r3584.bin_37566, continue with recompile... Check that the folder and r3584 files have read/write/execute permissions for everyone. |
TBar Send message Joined: 22 May 99 Posts: 5204 Credit: 840,779,836 RAC: 2,768 |
You could always try the CUDA65 Special App with the 710. If it works it should be about twice as fast on the Arecibo tasks and a bit faster on the BLC tasks. You'd want to use the CUDA 65 App and make sure to change the app_info.xml from <cmdline>-unroll autotune</cmdline> to <cmdline>-unroll 1</cmdline>. If you have existing CUDA 60 tasks, you would want to also change the <plan_class> to cuda60 on the app_info to keep from Ghosting the tasks. You'll still Ghost the OpenCL tasks though, unless you change the client_state.xml to match the app_info or just abort all the OpenCL tasks before making the change. It would be wise to copy the 2 cuda libraries to a different directory just in case they get lost from the setiathome.berkeley.edu folder. The App has been downloaded 22 times, I've heard back from one person, so, it might work and it might not ;-) You'll definitely want to use a different CPU App, I understand the SSSE3 App doesn't work well on AMD CPUs. The App is here, Linux_CUDA65_Special.zip |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.