search

JeremyMain / GPUProfiler

GPUProfiler - Understand your application and workflow resource requirements
289 stars 19 forks source link

Failure detecting GPU - NVML API location with DCH drivers #122

Open JeremyMain opened 5 years ago

JeremyMain commented 5 years ago

DCH drivers do not use the standard installation path for the NVML DLL. Fallback test for alternate locations.

Bug fixed being confirmed by reporting user.

JeremyMain commented 5 years ago

Fix complete and to be included in v1.07b

JeremyMain commented 4 years ago

fix backported from v1.07b and added to v1.07a1

andyg336 commented 4 years ago

Still having GPUProfiler v1.07a1 fail to detect my RTX 2070, any information I can provide to help fix?

JeremyMain commented 4 years ago

@andyg336 if you could share the first 25 lines from the output of the following command

C:\Program Files\NVIDIA Corporation\NVSMI>nvidia-smi.exe -q

Ex: ==============NVSMI LOG==============

Timestamp : Thu Aug 13 23:04:26 2020 Driver Version : 442.92 CUDA Version : 10.2

Attached GPUs : 1 GPU 00000000:02:00.0 Product Name : Quadro GP100 Product Brand : Quadro Display Mode : Enabled Display Active : Enabled Persistence Mode : N/A Accounting Mode : Disabled Accounting Mode Buffer Size : 4000 Driver Model Current : WDDM Pending : WDDM Serial Number : 0321617046176 GPU UUID : GPU-ba1918e6-6088-0dc5-609a-32ab00241593 Minor Number : N/A VBIOS Version : 86.00.3A.00.0A MultiGPU Board : No Board ID : 0x200 GPU Part Number : 900-5H400-0000-000 Inforom Version Image Version : H400.0500.01.04 OEM Object : 1.1 ECC Object : 4.1 Power Management Object : N/A

JeremyMain commented 4 years ago

re-opening this issue

andyg336 commented 4 years ago

C:\Program Files\NVIDIA Corporation\NVSMI>nvidia-smi.exe -q Failed to initialize NVML: Unknown Error

Which doesn't bode well for me

andyg336 commented 4 years ago

Reinstalled drivers, several times. Issue intermittently fixed. But I'm fairly confident the issue is my end

JeremyMain commented 4 years ago

It could be that you have a DCH driver that has been installed while the older non-DCH driver remains installed. The only workaround to avert this situation on my end would be to first attempt DCH driver detection and fall back to non-DCH driver detection if that fails. Uninstalling any of the older drivers should help rectify the situation. Thank you for reporting this issue.

andyg336 commented 4 years ago

Appreciate the suggestion, I'll give it a go.

Thanks for a really nice tool as well 👍