onnxruntime
[webgpu] Apply dp4a for generation shader
#24064
Merged

[webgpu] Apply dp4a for generation shader #24064

guschmue merged 12 commits into main from matmulnbist_dp4a_gen
qjia7
qjia7 [webgpu] Apply dp4a for generation shader
d9430217
qjia7 support any block_size % 32 = 0
356410a5
qjia7 qjia7 requested a review from sushraja-msft sushraja-msft 1 year ago
guschmue guschmue added ep:WebGPU
qjia7 apply it only for float type
4cd3a3a1
qjia7 qjia7 requested a review from guschmue guschmue 1 year ago
guschmue
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft requested changes on 2025-03-19
qjia7 Merge branch 'main' into matmulnbist_dp4a_gen
e116df85
qjia7 use 1D dispatch group size
4631638e
qjia7 Adjust the code to make it more flexible
5074d164
qjia7 Use workgroup size = 128
d96de51c
qjia7 Add more annotations
36db69d4
qjia7 qjia7 marked this pull request as draft 1 year ago
qjia7 fix error in scale_a
701acbd3
qjia7 Extract common functions for code reuse
e538dd57
qjia7 qjia7 marked this pull request as ready for review 1 year ago
qjia7 qjia7 requested a review from sushraja-msft sushraja-msft 1 year ago
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft commented on 2025-03-19
sushraja-msft
sushraja-msft commented on 2025-03-19
qjia7 address comments
f3a93e74
qjia7 qjia7 requested a review from sushraja-msft sushraja-msft 1 year ago
sushraja-msft
sushraja-msft commented on 2025-03-20
sushraja-msft
sushraja-msft commented on 2025-03-20
sushraja-msft
sushraja-msft commented on 2025-03-20
sushraja-msft
sushraja-msft commented on 2025-03-20
sushraja-msft
sushraja-msft dismissed these changes on 2025-03-20
qjia7 address comments
f9ac9ab1
qjia7 qjia7 dismissed their stale review via f9ac9ab1 1 year ago
qjia7 qjia7 requested a review from sushraja-msft sushraja-msft 1 year ago
guschmue
guschmue approved these changes on 2025-03-20
guschmue guschmue merged 127c8503 into main 1 year ago
guschmue guschmue deleted the matmulnbist_dp4a_gen branch 1 year ago

Login to write a write a comment.

Login via GitHub

Assignees
No one assigned
Labels
Milestone