CalvinBackup/neuralchen-SimSwap
1
0
Fork 0
Code Issues 288 Pull Requests 16 Actions Packages Projects Releases 2 Wiki Activity

Slow inference performance on 3090 #146

New Issue
Open
opened 2021-11-05 16:49:33 +01:00 by PascalTemel · 7 comments
PascalTemel commented 2021-11-05 16:49:33 +01:00 (Migrated from github.com)
Copy Link
Copy Source

Installed CUDA version is 11.4 and onnxuntime-gpu is 1.9
My GPU is barely being utilized and I'm only getting around 2 IT/S, is there anything I can do to improve that?

Installed CUDA version is 11.4 and onnxuntime-gpu is 1.9 My GPU is barely being utilized and I'm only getting around 2 IT/S, is there anything I can do to improve that?
SaroshZDandoti commented 2021-11-07 08:33:49 +01:00 (Migrated from github.com)
Copy Link
Copy Source

have you solved this problem , im having the same issue, but mine is windows 11 and cuda 11.5 rest is same

have you solved this problem , im having the same issue, but mine is windows 11 and cuda 11.5 rest is same
PascalTemel commented 2021-11-07 17:01:37 +01:00 (Migrated from github.com)
Copy Link
Copy Source

have you solved this problem , im having the same issue, but mine is windows 11 and cuda 11.5 rest is same

No, no changes on my side

> have you solved this problem , im having the same issue, but mine is windows 11 and cuda 11.5 rest is same No, no changes on my side
😕 1
paulescobar commented 2021-11-13 18:45:48 +01:00 (Migrated from github.com)
Copy Link
Copy Source

I'm a total newb at this. But having enabled those recommended GPU options (with some tweaks)...I decided to monitor the "task manager" during a test conversion.

During that conversion, It was utilizing 2.9 GB / 6 GB in "dedicated GPU memory usage" (how task manager displayed it in the "Performance > GPU" section).

It was the only stat of the GPU that moved during the conversion.
And as soon as the conversion was done, the usage in that "dedicated GPU memory usage" dropped to "0.5 GB / 6 GB".

I guess that counts as GPU help?

CPU use was around 44% during the same conversion.

I'm a total newb at this. But having enabled those recommended GPU options (with some tweaks)...I decided to monitor the "task manager" during a test conversion. During that conversion, It was utilizing 2.9 GB / 6 GB in "dedicated GPU memory usage" (how task manager displayed it in the "Performance > GPU" section). It was the only stat of the GPU that moved during the conversion. And as soon as the conversion was done, the usage in that "dedicated GPU memory usage" dropped to "0.5 GB / 6 GB". I guess that counts as GPU help? CPU use was around 44% during the same conversion.
👍 1
whenupdate commented 2021-11-20 17:22:01 +01:00 (Migrated from github.com)
Copy Link
Copy Source

have same issue, only take 2GB GPU maximum working. are u solve this issue ?

have same issue, only take 2GB GPU maximum working. are u solve this issue ?
😕 1
PascalTemel commented 2021-11-20 17:25:18 +01:00 (Migrated from github.com)
Copy Link
Copy Source

I haven't managed to improve the GPU usage unfortunately. I'm using FaceSwap.dev in the meantime and it makes full use of my GPU, but it's a lot more effort to get to the desired result.

I haven't managed to improve the GPU usage unfortunately. I'm using FaceSwap.dev in the meantime and it makes full use of my GPU, but it's a lot more effort to get to the desired result.
paulescobar commented 2021-11-22 17:41:45 +01:00 (Migrated from github.com)
Copy Link
Copy Source

have same issue, only take 2GB GPU maximum working. are u solve this issue ?

I had an interesting experience today. I previously had SimSwap installed on my C drive. At that time, I noticed only 2 IT/S and around 2 GB of GPU dedicated memory usage.

Today I moved my SimSwap to another partition (same drive as C, just different partition on WD Black performance drive). Now I see 5 IT/S and around 2.9 GPU dedicated memory usage.

So no significant increase in dedicated GPU memory usage...but more than double "IT/S" rate.

Maybe you guys could try this and see if it helps?

> have same issue, only take 2GB GPU maximum working. are u solve this issue ? I had an interesting experience today. I previously had SimSwap installed on my C drive. At that time, I noticed only 2 IT/S and around 2 GB of GPU dedicated memory usage. Today I moved my SimSwap to another partition (same drive as C, just different partition on WD Black performance drive). Now I see 5 IT/S and around 2.9 GPU dedicated memory usage. So no significant increase in dedicated GPU memory usage...but more than double "IT/S" rate. Maybe you guys could try this and see if it helps?
👍 1
SaroshZDandoti commented 2021-11-22 18:06:19 +01:00 (Migrated from github.com)
Copy Link
Copy Source

hmm, gonna try this asap. will let you know

> hmm, gonna try this asap. will let you know
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
GIF
GUI
Increase inference time
A solution to increase inference time
 bug
Something isn't working
 documentation
Improvements or additions to documentation
 duplicate
This issue or pull request already exists
 enhancement
New feature or request
 good first issue
Good for newcomers
 help wanted
Extra attention is needed
 invalid
This doesn't seem right
 question
Further information is requested
 wontfix
This will not be worked on
No labels
Milestone
No items
No Milestone
Projects
Clear projects
No projects
Assignees
Clear assignees
CalvinBackup (Calvin Backup)
No Assignees
1 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: CalvinBackup/neuralchen-SimSwap#146