Skip to content

GitLab

"src/vscode:/vscode.git/clone" did not exist on "704d3b1c83575bb8dfa963fa5939181a54a7b3b4"
Commit a8409d7f authored by Khalique Ahmed's avatar Khalique Ahmed
Browse files

hardcode compute type

parent 00d5d880
Show whitespace changes
Inline Side-by-side
Showing with 6 additions and 4 deletions
src/targets/gpu/gemm_impl.cpp
View file @ a8409d7f
...@@ -65,11 +65,13 @@ void gemm_impl(context& ctx, ...@@ -65,11 +65,13 @@ void gemm_impl(context& ctx,
output_type = rocblas_datatype_i32_r; output_type = rocblas_datatype_i32_r;
} }
auto compute_type = output_type; auto compute_type = output_type;
if(ctx.get_stream().get_device_name() == "gfx908")
{
if(args[0].get_shape().type() == shape::half_type) if(args[0].get_shape().type() == shape::half_type)
compute_type = rocblas_datatype_f32_r; compute_type = rocblas_datatype_f32_r;
} // if(ctx.get_stream().get_device_name() == "gfx908")
// {
// if(args[0].get_shape().type() == shape::half_type)
// compute_type = rocblas_datatype_f32_r;
// }
#if ROCBLAS_VERSION_MAJOR >= 2 && ROCBLAS_VERSION_MINOR >= 38 #if ROCBLAS_VERSION_MAJOR >= 2 && ROCBLAS_VERSION_MINOR >= 38
rocblas_gemm_flags flag = rocblas_gemm_flags flag =
......
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