Home / Class/ is_matmul Class — pytorch Architecture

is_matmul Class — pytorch Architecture

Architecture documentation for the is_matmul class in FusionUtils.h from the pytorch codebase.

Entity Profile

Source Code

aten/src/ATen/native/mkldnn/xpu/FusionUtils.h lines 28–51

template <bool is_matmul = false>
onednn::Attr& construct_binary_attr(
    onednn::Attr& attr,
    std::string_view binary,
    const Tensor& other) {
  if (binary == "mul") {
    attr.append_post_binary<is_matmul>(attr.kind_with_binary_mul, other);
  } else if (binary == "sub") {
    attr.append_post_binary<is_matmul>(attr.kind_with_binary_sub, other);
  } else if (binary == "div") {
    attr.append_post_binary<is_matmul>(attr.kind_with_binary_div, other);
  } else if (binary == "add") {
    attr.append_post_binary<is_matmul>(attr.kind_with_binary_add, other);
  } else if (binary == "sum") {
    attr.append_post_sum(1.f, 1.f, 0);
  } else {
    TORCH_CHECK(
        binary == "none",
        "Binary attr ",
        binary,
        "is not supported for conv/linear post binary fusion");
  }
  return attr;
}

Analyze Your Own Codebase

Get architecture documentation, dependency graphs, and domain analysis for your codebase in minutes.

Try Supermodel Free