PR #17990 HIP: Refactor mma for RDNA and CDNA

HIP: Refactor mma for RDNA and CDNA #17990

JohannesGaessler merged 10 commits into ggml-org:master from refactor_mma_for_rdna

mma.cuh for rdna4

318cb5b8

mma for rdna3

074b9314

mmq for rdna4

98846cb9

mmq for rdna3

62e4954d

align i-major and j-major

8b26bc38

cdna

afb0e3d5

zhanghuiyulo requested a review from

JohannesGaessler 296 days ago

zhanghuiyulo requested a review from

am17an 296 days ago

fix cuda error

6b8ed41f

zhanghuiyulo marked this pull request as draft 296 days ago

add missing tile of mfma

6acad9c7

JohannesGaessler commented on 2025-12-13

github-actions added Nvidia GPU

github-actions added ggml

fix j-major wrong ne on CDNA

cffa070b

zhanghuiyulo marked this pull request as ready for review 293 days ago

JohannesGaessler approved these changes on 2025-12-16

fix gramma and empty spaces

cad07fa4

JohannesGaessler merged acec774e into master 292 days ago

zhanghuiyulo deleted the refactor_mma_for_rdna branch 291 days ago

Reviewers

JohannesGaessler

am17an

Assignees

No one assigned

Labels

Nvidia GPU ggml

Milestone

No milestone