A high-throughput and memory-efficient inference and serving engine for LLMs
Mirrored from https://github.com/tduncan-lum/Nvidia-Optimization-Stack---VLLM.git
Synced Nov 17, 03:00 UTC
punica fix-bgmv-kernel-640 (#4007)
fuchen.ljl committed
Apr 11, 2024 at 23:59 UTC
08ccee1e830d39ecdb3c6cf382c843dbf5ae830e
1 file changed
+1
csrc/punica/bgmv/bgmv_config.h
+1
@@ -14,6 +14,7 @@ void bgmv_kernel(out_T *__restrict__ Y, const in_T *__restrict__ X,
14
f(in_T, out_T, W_T, narrow, 128) \
15
f(in_T, out_T, W_T, narrow, 256) \
16
f(in_T, out_T, W_T, narrow, 512) \
17
+ f(in_T, out_T, W_T, narrow, 640) \
18
f(in_T, out_T, W_T, narrow, 768) \
19
f(in_T, out_T, W_T, narrow, 1024) \
20
f(in_T, out_T, W_T, narrow, 1152) \