llama.cpp
2x faster (rms) norm cuda kernels (3.7% e2e improvement)
#2985

Merged

2x faster (rms) norm cuda kernels (3.7% e2e improvement) #2985

JohannesGaessler merged 2 commits into ggml-org:master from li-plus:opt-norm

2x faster (rms) norm cuda kernels

54ddacaa

li-plus force pushed to 54ddacaa 2 years ago

JohannesGaessler commented on 2023-09-03

Fix code style

9dc817e5

JohannesGaessler approved these changes on 2023-09-03

JohannesGaessler merged 35195689 into master 2 years ago

li-plus deleted the opt-norm branch 2 years ago

Reviewers

JohannesGaessler

Assignees

No one assigned

Labels

None yet

Milestone

No milestone