llama.cpp
eab4a882
- Using dp4a ptx intrinsics for an improved Mul8MAT perf [By Alcpz]
Go
Login via GitHub
Home
Pricing
FAQ
Install
Login
via GitHub
Commit
View On
GitHub
Commit
1 year ago
Using dp4a ptx intrinsics for an improved Mul8MAT perf [By Alcpz]
References
codeplay/sycl-main
Author
OuadiElfarouki
Parents
439b3fc7
Loading