[CUDA] fp16 intB gemm #24854
tianleiwu
merged
26 commits into
main
from
tlwu/fp16_intB_gemm
Add fp16 int4/8 gemm
baa6022a
clean up moe
b275f1ab
Merge branch 'main' into tlwu/fp16_intB_gemm
acf8d020
Revert "clean up moe"
ea3c1fe2
update generate_kernels.py
c77f9bf6
rename nv_infer_runtime.h to nv_infer_datatype.h
fe8aea92
Add tests
731301db
fix 4 bit zero point adaptor
43b6a849
Add scale only kernels
8916212d
github-advanced-security
commented
on 2025-05-25
format
0c3461b9
Revert "Add scale only kernels"
13e9be14
remove scaleonly cu files
f48ddfbb
Update cuda compiler settings
1fdf33f8
remove ENABLE_BF16
d0dbfacc
fix unused parameters warnings
528e261c
Add bf16 gemv
358c832c
keep last virtual
75841280
move min cuda version to cuda_configuraiton.cmake
31a41945
increase test tolerance
213f0968
static cast
cae37c97
fix unused paramter warnings
7207bc4c
prepacking
7b23374d
copilot-pull-request-reviewer
commented
on 2025-05-29
format
39a5d99d
Merge branch 'main' into tlwu/fp16_intB_gemm
bd27273b
kunal-vaishnavi
commented
on 2025-05-29
kunal-vaishnavi
commented
on 2025-05-29
disable by default
b3f49b00
kunal-vaishnavi
commented
on 2025-05-29
rewrite transpose b kernel
00f36f4d
nenad1002
approved these changes
on 2025-05-30
Assignees
No one assigned
Labels
None yet
Milestone
No milestone
Login to write a write a comment.
Login via GitHub