Replies: 2 comments
-
|
I have the same GPU and tested it, it seems like it no longer auto offload (-1) correctly. If i manually enter the "GPU layer" to what it was in previous version automaticly it starts using the GPU correctly. |
Beta Was this translation helpful? Give feedback.
0 replies
-
|
On a 6700XT, Windows 11, same issue, It defaults to Vulcan, so I switch manually to hipblas (ROCm) |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Uh oh!
There was an error while loading. Please reload this page.
-
Hi
Just tested latest KoboldCPP-v1.86.2.yr0-ROCm, both koboldcpp_rocm.exe and koboldcpp_rocm_b2.exe files
and it seems that it doesn't use GPU when generating text with default settings. 7900XT
it doesn't allocate vram.
KoboldCPP-v1.85.yr0-ROCm works fine.
Beta Was this translation helpful? Give feedback.
All reactions