Skip to content

[ET-VK] Minor improvement to q_linear op shader. #7728

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

Merged
merged 8 commits into from
Jan 24, 2025
Merged

[ET-VK] Minor improvement to q_linear op shader. #7728

Changes from all commits
Commits
Show all changes
8 commits
Select commit Hold shift + click to select a range
51a582a
[ET-VK] Minor improvement to q_linear op shader.
trivedivivek Jan 17, 2025
0e4be6e
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 21, 2025
2fad849
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 22, 2025
ce14377
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 22, 2025
3d12797
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 22, 2025
58fd17c
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 23, 2025
c425b9b
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 24, 2025
c72ff0f
Update on "[ET-VK] Minor improvement to q_linear op shader."
trivedivivek Jan 24, 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
11 changes: 5 additions & 6 deletions backends/vulkan/runtime/graph/ops/glsl/q_8w_linear.glsl
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -92,7 +92,7 @@ void main() {

#extension GL_EXT_shader_explicit_arithmetic_types_int16 : require

VEC4_T q_8w_linear(const u16vec3 out_pos, const uint16_t K) {
VEC4_T q_8w_linear(const u16vec2 out_pos, const uint16_t K) {
const uint16_t qmat2_pos_y = out_pos.x * uint16_t(4);

VEC4_T outtex = VEC4_T(0);
Expand All @@ -101,7 +101,7 @@ VEC4_T q_8w_linear(const u16vec3 out_pos, const uint16_t K) {
const VEC4_T scales = load_texel(t_scales, scales_pos);

for (uint16_t i = uint16_t(0), x = uint16_t(0); i < K; i += uint16_t(4), x++) {
const VEC4_T mat1_tex = load_texel(t_mat1, u16vec3(x, out_pos.yz));
const VEC4_T mat1_tex = load_texel(t_mat1, u16vec3(x, out_pos.y, 0));
const VEC4_T sums = VEC4_T(
dot(mat1_tex, load_texel(t_qmat2, u16vec3(x, qmat2_pos_y, 0))),
dot(mat1_tex, load_texel(t_qmat2, u16vec3(x, qmat2_pos_y + uint16_t(1), 0))),
Expand All @@ -117,16 +117,15 @@ VEC4_T q_8w_linear(const u16vec3 out_pos, const uint16_t K) {
}

void main() {
const u16vec3 out_pos = u16vec3(
const u16vec2 out_pos = u16vec2(
gl_GlobalInvocationID.x / out_limits.y,
gl_GlobalInvocationID.x % out_limits.y,
0);
gl_GlobalInvocationID.x % out_limits.y);
if (out_pos.x >= out_limits.x) {
return;
}

VEC4_T outtex = q_8w_linear(out_pos, uint16_t(mat1_sizes.x));
write_texel(t_out, out_pos, outtex);
write_texel(t_out, u16vec3(out_pos, 0), outtex);
}

#endif
Loading