We read every piece of feedback, and take your input very seriously.
To see all available qualifiers, see our documentation.
1 parent a53fe0f commit bc1aa41Copy full SHA for bc1aa41
1 file changed
include/caffe/util/device_alternate.hpp
@@ -81,14 +81,8 @@ namespace caffe {
81
const char* cublasGetErrorString(cublasStatus_t error);
82
const char* curandGetErrorString(curandStatus_t error);
83
84
-// CUDA: thread number configuration.
85
-// Use 1024 threads per block, which requires cuda sm_2x or above,
86
-// or fall back to attempt compatibility (best of luck to you).
87
-#if __CUDA_ARCH__ >= 200
88
- const int CAFFE_CUDA_NUM_THREADS = 1024;
89
-#else
90
- const int CAFFE_CUDA_NUM_THREADS = 512;
91
-#endif
+// CUDA: use 512 threads per block
+const int CAFFE_CUDA_NUM_THREADS = 512;
92
93
// CUDA: number of blocks for threads.
94
inline int CAFFE_GET_BLOCKS(const int N) {
0 commit comments