Commit f140f4c6 authored by Shucai Xiao's avatar Shucai Xiao
Browse files

fixed a bug in the gpu gemm implementation.

parent 44732d6c
...@@ -342,7 +342,7 @@ argument miopen_gemm::compute(context& ctx, ...@@ -342,7 +342,7 @@ argument miopen_gemm::compute(context& ctx,
bool transb = args[1].get_shape().transposed(); bool transb = args[1].get_shape().transposed();
rocblas_int lda = args[0].get_shape().strides()[transa ? dim_1 : dim_0]; rocblas_int lda = args[0].get_shape().strides()[transa ? dim_1 : dim_0];
rocblas_int ldb = args[1].get_shape().strides()[transb ? dim_1 : dim_0]; rocblas_int ldb = args[1].get_shape().strides()[transb ? dim_1 : dim_0];
rocblas_int ldc = args[3].get_shape().strides()[dim_0]; rocblas_int ldc = args[2].get_shape().strides()[dim_0];
auto out_lens = output_shape.lens(); auto out_lens = output_shape.lens();
rocblas_int m = out_lens[dim_0]; rocblas_int m = out_lens[dim_0];
rocblas_int n = out_lens[dim_1]; rocblas_int n = out_lens[dim_1];
......
Markdown is supported
0% or .
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment