Skip to content

[ExecuTorch] Precompute multiplicative inverse when possible in op_div #5209

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Closed
wants to merge 2 commits into from
Closed

[ExecuTorch] Precompute multiplicative inverse when possible in op_div #5209

Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
92e5a7e
[ExecuTorch] Precompute multiplicative inverse when possible in op_div
swolchok Sep 9, 2024
dc9b241
Update on "[ExecuTorch] Precompute multiplicative inverse when possib…
swolchok Sep 12, 2024
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
12 changes: 9 additions & 3 deletions kernels/optimized/cpu/op_div.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -84,8 +84,11 @@ Tensor& opt_div_out(
tensor->const_data_ptr<CTYPE>(),
out.numel());
} else {
Vec inv_scalar_casted_vec(CTYPE(1) / scalar_casted);
executorch::vec::map<CTYPE>(
[scalar_casted](Vec x) { return x / Vec(scalar_casted); },
[inv_scalar_casted_vec](Vec x) {
return x * inv_scalar_casted_vec;
},
out.mutable_data_ptr<CTYPE>(),
tensor->const_data_ptr<CTYPE>(),
out.numel());
Expand Down Expand Up @@ -220,8 +223,9 @@ Tensor& opt_div_scalar_out(
CTYPE b_casted = static_cast<CTYPE>(b_val);

using Vec = executorch::vec::Vectorized<CTYPE>;
Vec inv_b_casted_vec(CTYPE(1) / b_casted);
executorch::vec::map<CTYPE>(
[b_casted](Vec x) { return x / Vec(b_casted); },
[inv_b_casted_vec](Vec x) { return x * inv_b_casted_vec; },
out.mutable_data_ptr<CTYPE>(),
a.const_data_ptr<CTYPE>(),
out.numel());
Expand All @@ -239,14 +243,16 @@ Tensor& opt_div_scalar_out(
CTYPE_B b_val;
ET_EXTRACT_SCALAR(b, b_val);
CTYPE_IN b_casted = static_cast<CTYPE_IN>(b_val);
CTYPE_IN inv_b_casted = CTYPE_IN(1) / b_casted;

const size_t n = a.numel();
const CTYPE_A* a_data = a.const_data_ptr<CTYPE_A>();
CTYPE_OUT* out_data =
out.mutable_data_ptr<CTYPE_OUT>();
for (auto i = 0; i < n; ++i) {
out_data[i] = static_cast<CTYPE_OUT>(
static_cast<CTYPE_IN>(a_data[i]) / b_casted);
static_cast<CTYPE_IN>(a_data[i]) *
inv_b_casted);
}
});
});
Expand Down
Loading