Replies: 1 comment 9 replies
-
This model is too old to use with Vulkan. Use a different backend, or get a more up-to-date model. |
Beta Was this translation helpful? Give feedback.
9 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
I
m using: Xeon E5-2690v2 RX 570 8gb 16 RAM And latest version of koboldcpp (no cuda version) with "Wizard-Vicuna-13B-Uncensored.ggmlv3.q4_K_S" and Vulkan NoAVX2 + 10 GPU layers (can
t understand if it`s even useful cuz 0% GPU and VRAM usage) with average speed 2.9 - 3.1 T/s. Can I impove speed somehow,or that the best i can get with my hardware?Beta Was this translation helpful? Give feedback.
All reactions