Skip to content

Implement batched matmul for large 1D dot products #3580

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

Open
Ved235 wants to merge 3 commits into
base: main
Choose a base branch
from
Open

Implement batched matmul for large 1D dot products #3580

Changes from all commits
Commits
Show all changes
3 commits
Select commit Hold shift + click to select a range
fb488bc
Route large 1D dot products through batched matmul
Ved235 May 22, 2026
24d639d
Change chunk size
Ved235 May 22, 2026
0d2e620
Remove operand level checking
Ved235 May 24, 2026
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
29 changes: 29 additions & 0 deletions mlx/ops.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -58,6 +58,28 @@ Dtype at_least_float(const Dtype& d) {
return issubdtype(d, inexact) ? d : promote_types(d, float32);
}

array inner_1d_gpu_chunked(const array& a, const array& b, Stream s) {
constexpr int kChunkSize = 4096;

int main_size = (a.size() / kChunkSize) * kChunkSize;
int batches = main_size / kChunkSize;

auto a_main =
reshape(slice(a, {0}, {main_size}, s), {batches, 1, kChunkSize}, s);
auto b_main =
reshape(slice(b, {0}, {main_size}, s), {batches, kChunkSize, 1}, s);

// Route large 1D dot products through batched matmul so gemv parallelizes
array total = sum(reshape(matmul(a_main, b_main, s), {batches}, s), false, s);
if (main_size == a.size()) {
return total;
}

auto a_tail = slice(a, {main_size}, {static_cast<int>(a.size())}, s);
auto b_tail = slice(b, {main_size}, {static_cast<int>(b.size())}, s);
return add(total, sum(multiply(a_tail, b_tail, s), false, s), s);
}

array indices_or_default(
std::optional<array> indices,
const array& x,
Expand Down Expand Up @@ -5406,6 +5428,13 @@ array tensordot(
}
}

auto stream = to_stream(s);
if (a.ndim() == 1 && b.ndim() == 1 && axes_a.size() == 1 &&
axes_b.size() == 1 && stream.device == Device::gpu &&
csize >= 32 * 4096) {
return inner_1d_gpu_chunked(a, b, stream);
}

std::vector<bool> cdims1(x.ndim(), false);
std::vector<bool> cdims2(y.ndim(), false);
for (const auto n : axes_a) {
Expand Down
Loading