[js/webgpu] Provide a naive vectorized matmul algorithm #18758
[js/webgpu] Optimize the naive nhwc conv
5245593d
output multiple col data
772321e2
support maxInComponent
9676e23c
[js/webgpu] Optimize matmul with large batch size
7c4a148b
add components support for output
84e0b54d
add A components
4f8c09d7
output multiple results per thread
2bb39744
fix rebase errors
26376073
go to naive matmul path for special shape
88078ecd
nits
ea9144f1
remove the unused file
82b16b3e
Use uniforms
dff4153f
use snake case instead of camel case
f8161ab4
guschmue
approved these changes
on 2023-12-12
fs-eire
approved these changes
on 2023-12-13
guschmue
merged
b30e721d
into main 2 years ago
qjia7
deleted the opt_matmul_vec branch 2 years ago
Assignees
No one assigned
Login to write a write a comment.
Login via GitHub