huggingface/candle
1
0
Fork 0
mirror of https://github.com/huggingface/candle.git synced 2025-06-15 10:26:33 +00:00
Code Issues Packages Projects Releases Wiki Activity

Optimizing decode matmul (Phi at 28tok/s on M3).

Browse Source 
Adding some benchmark in order to help checking out matmul performance.
This commit is contained in:
Nicolas Patry
2023-12-20 09:54:19 +01:00
parent 03641293ee
commit 9b5e4843a6
4 changed files with 66 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

1
Cargo.toml
View File

@ -32,6 +32,7 @@ accelerate-src = { version = "0.3.2" }
anyhow = { version = "1", features = ["backtrace"] }
byteorder = "1.4.3"
clap = { version = "4.2.4", features = ["derive"] }
criterion = { version = "0.5.1", default-features=false }
cudarc = { version = "0.9.14", features = ["f16"] }
gemm = { version = "0.16.6", features = ["wasm-simd128-enable"] }
hf-hub = "0.3.0"