Commit e01224f5 authored by Paul's avatar Paul
Browse files

Dont divide by vec.size

parent 0a5e9b99
...@@ -80,7 +80,7 @@ struct softmax_compiler : compiler<softmax_compiler> ...@@ -80,7 +80,7 @@ struct softmax_compiler : compiler<softmax_compiler>
vec = vectorize::elements(faxis, inputs); vec = vectorize::elements(faxis, inputs);
} }
auto relements = inputs[0].lens()[axis] / vec.size; auto relements = inputs[0].lens()[axis] / vec.size;
auto nelements = (inputs.back().elements() / inputs[0].lens()[axis]) / vec.size; auto nelements = (inputs.back().elements() / inputs[0].lens()[axis]);
auto block_size = compute_block_size(relements, 256); auto block_size = compute_block_size(relements, 256);
hip_compile_options options; hip_compile_options options;
options.set_launch_params( options.set_launch_params(
......
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