onnxruntime
[CUDA] Speed up 8-bit MatMulNBits dequantization with byte permutes
#31350

Merged

[CUDA] Speed up 8-bit MatMulNBits dequantization with byte permutes #31350

tianleiwu merged 2 commits into main from tlwu/20260801/matmul_8bits_fast_dequant

[CUDA] Speed up 8-bit MatMulNBits dequantization with byte permutes

972f27c1

tianleiwu requested a review from

copilot-pull-request-reviewer 7 days ago

tianleiwu requested a review from

jambayk 7 days ago

copilot-pull-request-reviewer commented on 2026-08-01

[CUDA] Use safe bit casts in 8-bit dequantization

0405b748

tianleiwu enabled auto-merge (squash) 7 days ago

tianleiwu requested a review from

apsonawane 6 days ago

tianleiwu requested a review from

nenad1002 6 days ago

tianleiwu requested a review from

titaiwangms 6 days ago

titaiwangms approved these changes on 2026-08-03

tianleiwu merged 5b249f65 into main 4 days ago

tianleiwu deleted the tlwu/20260801/matmul_8bits_fast_dequant branch 4 days ago

Reviewers

titaiwangms

jambayk

apsonawane

nenad1002

Assignees

No one assigned

Labels

None yet

Milestone

No milestone