llama.cpp
Vectorize load instructions in dmmv f16 CUDA kernel
#9816

Merged

Vectorize load instructions in dmmv f16 CUDA kernel #9816

slaren merged 3 commits into ggml-org:master from agray3:ag_vectorize_dmmv_access_instructions

Vectorize load instructions in dmmv f16 CUDA kernel

95c8b9c1

github-actions added Nvidia GPU

JohannesGaessler commented on 2024-10-10

addressed comment

d07dc44c

JohannesGaessler commented on 2024-10-10

Update ggml/src/ggml-cuda/dmmv.cu

d150c7e3

JohannesGaessler approved these changes on 2024-10-10

slaren merged 13dca2a5 into master 1 year ago

Reviewers

JohannesGaessler

Assignees

No one assigned

Labels

Nvidia GPU

Milestone

No milestone