CS348Project/llama.cpp
5
0
Fork 0
mirror of https://github.com/ggml-org/llama.cpp.git synced 2025-10-30 08:42:00 +00:00
Code Issues Packages Projects Releases Wiki Activity

CUDA: use min compute capability of GPUs actually used (#2506)

Browse Source
This commit is contained in:
Cebtenzzre
2023-08-04 11:35:22 -04:00
committed by GitHub
parent 02f9d96a86
commit 4329d1acb0
1 changed files with 2 additions and 1 deletions
Download Patch File Download Diff File Expand all files Collapse all files

3
ggml-cuda.cu
View File

@@ -5347,7 +5347,8 @@ void ggml_cuda_mul_mat(const ggml_tensor * src0, const ggml_tensor * src1, ggml_
} else {
int min_compute_capability = INT_MAX;
for (int id = 0; id < g_device_count; ++id) {
if (min_compute_capability > g_compute_capabilities[id]) {
if (min_compute_capability > g_compute_capabilities[id]
&& g_tensor_split[id] < (id + 1 < g_device_count ? g_tensor_split[id + 1] : 1.0f)) {
min_compute_capability = g_compute_capabilities[id];
}
}