Skip to content

[ET-VK] Minor improvement to permute op. #10117

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
wants to merge 6 commits into
base: gh/trivedivivek/77/base
Choose a base branch
from
Open

[ET-VK] Minor improvement to permute op. #10117

Changes from all commits
Commits
Show all changes
6 commits
Select commit Hold shift + click to select a range
b687df0
[ET-VK] Minor improvement to permute op.
trivedivivek Apr 11, 2025
4731247
Update on "[ET-VK] Minor improvement to permute op."
trivedivivek Apr 14, 2025
3457b0b
Update on "[ET-VK] Minor improvement to permute op."
trivedivivek Apr 14, 2025
42d0570
Update on "[ET-VK] Minor improvement to permute op."
trivedivivek Apr 16, 2025
45d27ee
Update on "[ET-VK] Minor improvement to permute op."
trivedivivek Apr 16, 2025
35faf60
Update on "[ET-VK] Minor improvement to permute op."
trivedivivek Apr 16, 2025
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
14 changes: 9 additions & 5 deletions backends/vulkan/runtime/graph/ops/glsl/permute.glsl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -31,6 +31,8 @@ layout(push_constant) uniform PRECISION restrict Block {
layout(local_size_x_id = 0, local_size_y_id = 1, local_size_z_id = 2) in;
layout(constant_id = 3) const int packed_dim = C_DIM;

#extension GL_EXT_control_flow_attributes : require

void main() {
ivec3 pos = ivec3(gl_GlobalInvocationID);

Expand All @@ -54,11 +56,16 @@ void main() {
in_bchw_pos[out_ndims[2]] = pos.y;
in_bchw_pos[out_ndims[3]] = pos.x;

for (int j = 0; j < 4; ++j) {
const int in_packed_dim_size = in_sizes[3 - out_ndims[in_packed_dim_bchw_index]];

[[unroll]] for (int j = 0, bchw_index = in_bchw_pos[out_ndims[in_packed_dim_bchw_index]]; j < 4; ++j, ++bchw_index) {
// terminate the loop if trying to access input texture out of bounds
if (any(greaterThanEqual(in_bchw_pos.wzyx, in_sizes.xyzw))) {
if (bchw_index >= in_packed_dim_size) {
break;
}
// go to position in the input, that is mapped to the packed dim in the output
in_bchw_pos[out_ndims[in_packed_dim_bchw_index]] = bchw_index;

ivec3 fetch_pos;

fetch_pos.xy = in_bchw_pos.wz;
Expand All @@ -74,9 +81,6 @@ void main() {
// fetch input texel
VEC4_T inval = VEC4_T(load_texel(t_in, fetch_pos));
outval[j] = inval[in_packed_dim_lane_index];

// go to next position in the input, that is mapped to the packed dim in the output
in_bchw_pos[out_ndims[in_packed_dim_bchw_index]]++;
}

pos[packed_dim] = int(gl_GlobalInvocationID[packed_dim]);
Expand Down
Loading