zhangyiss/mlx
1
0
Fork 0
mirror of https://github.com/ml-explore/mlx.git synced 2025-12-16 01:49:05 +08:00
Code Issues Actions 1 Packages Projects Releases Wiki Activity

WIP (gpu)

Browse Source
This commit is contained in:
Ronan Collobert
2025-10-31 13:19:54 -07:00
parent b0d985416a
commit 8277e71ea9
2 changed files with 3 additions and 3 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
mlx/backend/gpu/primitives.cpp
View File

@@ -51,7 +51,7 @@ void Contiguous::eval_gpu(const std::vector<array>& inputs, array& out) {
MLX_PROFILER_RANGE("Contiguous::eval_gpu");
assert(inputs.size() == 1);
auto& in = inputs[0];
constexpr size_t extra_bytes = 16384;
constexpr int64_t extra_bytes = 16384;
if (in.buffer_size() <= out.nbytes() + extra_bytes &&
(in.flags().row_contiguous ||
(allow_col_major_ && in.flags().col_contiguous))) {

4
mlx/backend/gpu/slicing.cpp
View File

@@ -11,7 +11,7 @@ void slice_gpu(
array& out,
const Shape& start_indices,
const Shape& strides,
const Stream& s) {
const Stream& /* s */) {
slice(in, out, start_indices, strides);
}
@@ -27,7 +27,7 @@ void pad_gpu(
// Find offset for start of input values
size_t data_offset = 0;
for (int i = 0; i < axes.size(); i++) {
for (int i = 0; i < std::ssize(axes); i++) {
auto ax = axes[i] < 0 ? out.ndim() + axes[i] : axes[i];
data_offset += out.strides()[ax] * low_pad_size[i];
}