From 44f0114ad2c208f69c0c1baa980a5b3bda37e16b Mon Sep 17 00:00:00 2001 From: aamir Date: Mon, 13 Aug 2018 14:57:41 -0700 Subject: generalized v8 and v16 mode kernel completed --- cuda-kernels/v4p_genericMatrixMultiply.cu | 6 +++--- 1 file changed, 3 insertions(+), 3 deletions(-) (limited to 'cuda-kernels/v4p_genericMatrixMultiply.cu') diff --git a/cuda-kernels/v4p_genericMatrixMultiply.cu b/cuda-kernels/v4p_genericMatrixMultiply.cu index a08903b..1b56eb2 100644 --- a/cuda-kernels/v4p_genericMatrixMultiply.cu +++ b/cuda-kernels/v4p_genericMatrixMultiply.cu @@ -20,9 +20,9 @@ void curandErrCheck_(curandStatus_t stat, const char *file, int line) { using namespace nvcuda; // Must be multiples of 16 for wmma code to work -#define MATRIX_M (32) -#define MATRIX_N (32) -#define MATRIX_K (32) +#define MATRIX_M (1024) +#define MATRIX_N (1024) +#define MATRIX_K (1024) // The only dimensions currently supported by WMMA -- cgit v1.3