tqcq/llama.cpp
0
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-08-18 05:56:00 -04:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: fix peer access logic (#3231)

Browse Source
This commit is contained in:
Johannes Gäßler
2023-09-17 23:35:20 +02:00
committed by GitHub
parent 111163e246
commit ee66942d7e
1 changed files with 8 additions and 6 deletions
Download Patch File Download Diff File Expand all files Collapse all files

14
ggml-cuda.cu
View File

@@ -6286,12 +6286,14 @@ void ggml_cuda_set_peer_access(const int n_tokens) {
continue; continue;
} }
int canAccessPeer; int can_access_peer;
CUDA_CHECK(cudaDeviceCanAccessPeer(&canAccessPeer, id, id_other)); CUDA_CHECK(cudaDeviceCanAccessPeer(&can_access_peer, id, id_other));
if (enable_peer_access) { if (can_access_peer) {
CUDA_CHECK(cudaDeviceEnablePeerAccess(id_other, 0)); if (enable_peer_access) {
} else { CUDA_CHECK(cudaDeviceEnablePeerAccess(id_other, 0));
CUDA_CHECK(cudaDeviceDisablePeerAccess(id_other)); } else {
CUDA_CHECK(cudaDeviceDisablePeerAccess(id_other));
}
} }
} }
} }