Skip to content

Use 16-byte alignment for CMSIS-NN scratch buffers #17765

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
rascani wants to merge 1 commit into
base: main
Choose a base branch
from
+7 −4
Open

Use 16-byte alignment for CMSIS-NN scratch buffers #17765

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
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
3 changes: 3 additions & 0 deletions backends/cortex_m/ops/cortex_m_ops_common.h
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -36,6 +36,9 @@ using KernelRuntimeContext = torch::executor::KernelRuntimeContext;
#define ARM_NN_Q31_MAX ((int32_t)(0x7FFFFFFFL))
#define ARM_NN_Q31_MIN ((int32_t)(0x80000000L))

// 16-byte alignment for MVE vector operations.
constexpr size_t kCortexMMveAlignment = 16;

// Basic tensor type / layout validation and dimension order checking
inline void validate_cmsis_nn_tensor_requirements(
const Tensor& input1,
Expand Down
2 changes: 1 addition & 1 deletion backends/cortex_m/ops/op_quantized_conv2d.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,7 +192,7 @@ Tensor& quantized_conv2d_out(
arm_convolve_s8_get_buffer_size(&input_dims, &filter_dims));
if (buffer_bytes > 0) {
auto buffer_or_error =
context.allocate_temp(buffer_bytes, alignof(int16_t));
context.allocate_temp(buffer_bytes, kCortexMMveAlignment);
if (!buffer_or_error.ok()) {
if (buffer_or_error.error() != Error::NotFound) {
ET_LOG(
Expand Down
2 changes: 1 addition & 1 deletion backends/cortex_m/ops/op_quantized_depthwise_conv2d.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -237,7 +237,7 @@ Tensor& quantized_depthwise_conv2d_out(
}

auto buffer_or_error = context.allocate_temp(
static_cast<size_t>(buffer_bytes), alignof(int16_t));
static_cast<size_t>(buffer_bytes), kCortexMMveAlignment);
if (!buffer_or_error.ok()) {
ET_LOG(
Error,
Expand Down
4 changes: 2 additions & 2 deletions backends/cortex_m/ops/op_quantized_transpose_conv2d.cpp
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -192,7 +192,7 @@ Tensor& quantized_transpose_conv2d_out(
const int32_t buffer_bytes = arm_transpose_conv_s8_get_buffer_size(
&transpose_conv_params, &input_dims, &filter_dims, &output_dims);
auto buffer_or_error = context.allocate_temp(
static_cast<size_t>(buffer_bytes), alignof(int16_t));
static_cast<size_t>(buffer_bytes), kCortexMMveAlignment);
if (!buffer_or_error.ok()) {
ET_LOG(
Error,
Expand All @@ -209,7 +209,7 @@ Tensor& quantized_transpose_conv2d_out(
arm_transpose_conv_s8_get_reverse_conv_buffer_size(
&transpose_conv_params, &input_dims, &filter_dims);
auto output_buffer_or_error = context.allocate_temp(
static_cast<size_t>(output_buffer_bytes), alignof(int16_t));
static_cast<size_t>(output_buffer_bytes), kCortexMMveAlignment);
if (!output_buffer_or_error.ok()) {
ET_LOG(
Error,
Expand Down
Loading