Message boards :
Number crunching :
Milkyway@Home doesn't uses all GPU
Message board moderation
Author | Message |
---|---|
Send message Joined: 14 Jun 12 Posts: 13 Credit: 2,295,346 RAC: 0 |
For some reason, Milkyway@home only uses around 50% to 70% of my GPU but, why doesn't it use all 100% percent of it? |
Send message Joined: 8 May 09 Posts: 3323 Credit: 520,983,681 RAC: 39,498 |
For some reason, Milkyway@home only uses around 50% to 70% of my GPU but, why doesn't it use all 100% percent of it? Because using 100% would leave your pc unusable for anything else. Some projects do use more than just a little bit and the Nvidia gpu's tend to use alot more of the gpu then the AMD gpu's do, but there are alot more AMD gpu's out there. |
Send message Joined: 19 Jul 10 Posts: 601 Credit: 19,010,618 RAC: 5,131 |
You can try to adjust it with the cmd parameters, either by using an app_info file or by changing the preferences over the webside. All needed information can be found in the Separation updated to 1.00 thread. I'd go with the app_info, that way you can run 2 WUs at once. I run my HD3850 at 98-99% and have no noteworthy performance issues except for if I want to watch HD movies. |
Send message Joined: 14 Jun 12 Posts: 13 Credit: 2,295,346 RAC: 0 |
How would I go about doing 2 at a time with the app_info? |
Send message Joined: 18 Jul 09 Posts: 300 Credit: 303,562,776 RAC: 0 |
This one works well for me: <app_info> <app> <name>milkyway</name> </app> <file_info> <name>milkyway_separation_1.02_windows_x86_64__opencl_amd_ati.exe</name> <executable/> </file_info> <app_version> <app_name>milkyway</app_name> <version_num>102</version_num> <flops>1.0e11</flops> <avg_ncpus>0.05</avg_ncpus> <max_ncpus>1</max_ncpus> <plan_class>ati14ati</plan_class> <coproc> <type>ATI</type> <count>0.5</count> </coproc> <cmdline>--gpu-target-frequency 10 --gpu-disable-checkpointing</cmdline> <file_ref> <file_name>milkyway_separation_1.02_windows_x86_64__opencl_amd_ati.exe</file_name> <main_program/> </file_ref> </app_version> </app_info> Just paste it into your projects folder and restart Boinc |
Send message Joined: 19 Nov 11 Posts: 2 Credit: 155,106 RAC: 0 |
This one works well for me: How about Nvidia GPU? |
Send message Joined: 11 Feb 10 Posts: 2 Credit: 21,276,140 RAC: 1,228 |
<app_info> <app> <name>milkyway</name> </app> <file_info> <name>milkyway_separation_1.02_windows_x86_64__opencl_nvidia.exe</name> <executable/> </file_info> <app_version> <app_name>milkyway</app_name> <version_num>102</version_num> <flops>1.0e11</flops> <avg_ncpus>0.05</avg_ncpus> <max_ncpus>1</max_ncpus> <plan_class>cuda</plan_class> <coproc> <type>NVIDIA</type> <count>0.5</count> </coproc> <cmdline> </cmdline> <file_ref> <file_name>milkyway_separation_1.02_windows_x86_64__opencl_nvidia.exe</file_name> <main_program/> </file_ref> </app_version> </app_info> |
Send message Joined: 8 May 09 Posts: 3323 Credit: 520,983,681 RAC: 39,498 |
<app_info> IT WORKS!!! I am now crunching two units at once, THANKS!!! |
Send message Joined: 11 Feb 10 Posts: 2 Credit: 21,276,140 RAC: 1,228 |
No problem. You can put in 0.25 GPU and do four at a time, or 0.125 and do eight at a time, and play with the number of CPU processors also, but frankly it seems to only slow things down a bit, if anything. I am not even certain running two threads is faster per work unit than one at a time on my Nvidia card. Similarly, increasing the number of CPUs per thread seems to keep the CPUs busier, but without any increase in speed. So just keep cpu avg low ~0.05 and sign up for some other project that will use the cpus (e.g. Einstein). |
Send message Joined: 8 May 09 Posts: 3323 Credit: 520,983,681 RAC: 39,498 |
No problem. Yeah that is what I do, I do NOT crunch with both the cpu and the gpu for the same project, I always separate them. |
Send message Joined: 19 Nov 11 Posts: 2 Credit: 155,106 RAC: 0 |
Thanks, Karl, 4 WU on GTX 680 during ~16min! |
©2024 Astroinformatics Group