Questions and Answers :
GPU applications :
EVGA 2070 Super and Evga 660ti - what sort of performance should I get and recommended configuration?
Message board moderation
Previous · 1 · 2
Author | Message |
---|---|
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
Hey Keith, What are these tasks of this type: 21au19aa.32315.38600.14.41.218_1? My 2070 Super just seems to be bogged down by these and making zero progress on them. Not sure if this is due to the fact that for some reason I can only seem to give my GPU 0.04 of a processor no matter what I do but have a gut feeling that this has something to do with no real processing on them. The blc* task types run at about 3 1/2 mins. https://boincstats.com/signature/-1/user/2360850/sig.png |
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
I finally figured out my bone-headed mistake on the app_config.xml setup. I forgot to show file extensions and it was saved as a .txt file. Removed that and all is good now. Showing 1 CPU and 1 GPU in use for GPU tasks now. Will monitor those 21au19aa.32315.38600.14.41.218_1 type tasks and see if that helps them move at all. https://boincstats.com/signature/-1/user/2360850/sig.png |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Hey Keith, Those are standard AR Arecibo tasks. Or possibly VHAR Arecibo tasks. Notice they end with just a issue number and not .vlar Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
I finally figured out my bone-headed mistake on the app_config.xml setup. I forgot to show file extensions and it was saved as a .txt file. Removed that and all is good now. Showing 1 CPU and 1 GPU in use for GPU tasks now. Will monitor those 21au19aa.32315.38600.14.41.218_1 type tasks and see if that helps them move at all. Common mistake when editing to save as .TXT instead of the native file extension. All tasks that you downloaded with the cpu settings being read from the app_info and before you fixed the app_config will show as the .04 cpu use. Doesn't mean they aren't being crunched that way with the fixed app_config. As long as you reread config file to pick up the new settings from app_config. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
Yeah, especially tweaking my fresh install of Win10. Always hate when I forget the little things. Anyways, still have not noticed any difference in this particular task type with naming structures like: 21au19aa.32315.38600.14.41.218 They literally get on the 2070 and just die on the vine...making no progress. I just forced processing of when of them again through suspending/resuming various tasks to see if there was any improvement. nothing. The last time I looked at a previous task run it was like this for near 45 mins. Just not sure what is going on with these task types. The CPU version of these naming conventions run just fine. Application Local: setiathome_v8 8.20 (opencl_nvidia_SoG) Name 21au19aa.32315.38600.14.41.218 State Running Received 8/23/2019 8:18:08 AM Report deadline 9/12/2019 7:27:50 PM Resources 1 CPU + 1 NVIDIA GPU Estimated computation size 70,727 GFLOPs CPU time 00:00:07 CPU time since checkpoint 00:00:07 Elapsed time 00:04:56 Estimated time remaining 00:21:00 Fraction done 0.005% Virtual memory size 142.26 MB Working set size 105.76 MB Directory slots/1 Process ID 15840 Executable MB8_win_x86_SSE3_OpenCL_NV_SoG_r3557.exe https://boincstats.com/signature/-1/user/2360850/sig.png |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Which points the problem to be located with the gpus and likely the graphics drivers. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Robert Miles is having the same problem with another VHAR species. Same Windows10 and same 436 drivers. All erroring out on time exceeded. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
Interesting. Well, I re-installed Boinc and the Lunatics app but still running 436.02 driver. Picked up a bunch of Cuda32 and Cuda42 GPU tasks. I am going to let this run for a few days and get to the Open_CL SOG tasks again. If I still have issues I am going to back down to the 431.60 driver and see what happens. Mucked around way too much with this today and have my tasks all jacked up. Working slowly through recovery of many of them. https://boincstats.com/signature/-1/user/2360850/sig.png |
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
I had a stuck task this morning when I woke up and checked (however running on cuda42 and not OpenCL). I downgraded to 431.60 driver and rebooted. Upon restart the tasks are again processing! Looks like it is drivers. https://boincstats.com/signature/-1/user/2360850/sig.png |
timminator2 Send message Joined: 12 Sep 99 Posts: 17 Credit: 24,031,980 RAC: 123 |
strange part is that these task type are running just fine with the 436.02 driver on my 660ti to it APPEARS that it may be 20 series drivers. I also see that NVIDIA just relesae 436.15 TODAY. I am going to give that a few days before I try it on the newer rig. https://boincstats.com/signature/-1/user/2360850/sig.png |
Keith Myers Send message Joined: 29 Apr 01 Posts: 13164 Credit: 1,160,866,277 RAC: 1,873 |
Huge difference in architecture between that Kepler 660 Ti and your Turing 2070. The crunching code path through each architecture is massively different. Seti@Home classic workunits:20,676 CPU time:74,226 hours A proud member of the OFA (Old Farts Association) |
©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.