-
Notifications
You must be signed in to change notification settings - Fork 90
[BUG]cant install with torch-rocm #1473
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Comments
We only validate rocm for 6.3.3. Please upgrade to rocm 6.3.3. |
I have switched from 6.2.2 to 6.3.3. its still failing to build. here is just the end: Total number of replaced kernel launches: 4
!! more info about the system its being compiled on: |
@yggdrasil75 Try this. I suspect you are running into bad setuptools or pip bug pip install -U pip setuptools |
https://gist.github.com/yggdrasil75/701ce4a32d69b8e9a7ec627dcf91702c |
@yggdrasil75 Can you make sure you have cuda-toolkit-12.4 or 12.6 installed? The error is due to cuda vs hip compat issue. I honestly don't know why it is throwing those errors. Please make sure you have a relatively new cuda (12.4, 12.6, 12.8) installed. |
I dont have an nvidia card in this system. |
I uninstalled cuda toolkit (dont know why it was there at all) and all other nvidia-cuda packages (dev, gdb, and toolkit-doc were the only other ones installed) and still same error. |
Describe the bug
A clear and concise description of what the bug is.
GPU Info
3 MI60s
Software Info
python 3.12.3
ubuntu 24.04
https://gist.github.com/yggdrasil75/3e593b9fcec1161ad9ed4266f487fd73
I saw another issue that said do it twice, I have done it 4 times and it still hasnt installed.
The text was updated successfully, but these errors were encountered: