Skip to content

Reformulating matrix multiplication scale equation to reduce math ops and improve power and performance. #6437

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 1 commit into from
Oct 23, 2024
Merged

Reformulating matrix multiplication scale equation to reduce math ops and improve power and performance. #6437

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
14 changes: 6 additions & 8 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 @@ -102,22 +102,20 @@ VEC4_T q_8w_linear(const ivec3 out_pos, const int K) {

for (int i = 0; i < K; i += 4) {
const VEC4_T mat1_tex = load_texel(t_mat1, mat1_pos);

const VEC4_T sums = VEC4_T(
dot(mat1_tex, load_texel(t_qmat2, qmat2_pos) * scales.x),
dot(mat1_tex,
load_texel(t_qmat2, qmat2_pos + u16vec3(0, 1, 0)) * scales.y),
dot(mat1_tex,
load_texel(t_qmat2, qmat2_pos + u16vec3(0, 2, 0)) * scales.z),
dot(mat1_tex,
load_texel(t_qmat2, qmat2_pos + u16vec3(0, 3, 0)) * scales.w));
dot(mat1_tex, load_texel(t_qmat2, qmat2_pos)),
dot(mat1_tex, load_texel(t_qmat2, qmat2_pos + u16vec3(0, 1, 0))),
dot(mat1_tex, load_texel(t_qmat2, qmat2_pos + u16vec3(0, 2, 0))),
dot(mat1_tex, load_texel(t_qmat2, qmat2_pos + u16vec3(0, 3, 0))));

outtex += sums;

mat1_pos.x++;
qmat2_pos.x++;
}

outtex *= scales;

return outtex;
}

Expand Down
Loading