Jack HuandDouglas Lehr ef38e8d006 [Model][Hardware][AMD][Kernel] Part 2/2 -> Enable e2e QK Norm + RoPE + KV Cache runtime fusion for Qwen3-30B-A3B on ROCM_ATTN (#42755) ...

Signed-off-by: Jack Hu <Jack.Hu@amd.com>
Signed-off-by: Jack Hu <jingxujaster@hotmail.com>
Signed-off-by: Jack Hu <jack.hu@amd.com>
Co-authored-by: Douglas Lehr <91553416+dllehr-amd@users.noreply.github.com>

2026-10-15 17:10:14 -07:00

..

quickreduce

[Misc] Consolidate csrc shared headers and prune dead code (#60721)

2026-10-12 09:46:22 -07:00

attention.cu

[Model][Hardware][AMD][Kernel] Part 2/2 -> Enable e2e QK Norm + RoPE + KV Cache runtime fusion for Qwen3-30B-A3B on ROCM_ATTN (#42755)

2026-10-15 17:10:14 -07:00

dispatch_utils.h

[Misc] Consolidate csrc shared headers and prune dead code (#60721)

2026-10-12 09:46:22 -07:00

moe_q_gemm_rdna3.cu

[ROCm][RDNA3] Fix W4A16 split-K accuracy and determinism (#54706)

2026-10-03 23:44:17 +08:00

ops.h

[Model][Hardware][AMD][Kernel] Part 2/2 -> Enable e2e QK Norm + RoPE + KV Cache runtime fusion for Qwen3-30B-A3B on ROCM_ATTN (#42755)

2026-10-15 17:10:14 -07:00

q_gemm_rdna3_wmma.cu

[ROCm][RDNA3] Fix W4A16 split-K accuracy and determinism (#54706)

2026-10-03 23:44:17 +08:00

q_gemm_rdna3.cu

[Misc] Consolidate csrc shared headers and prune dead code (#60721)

2026-10-12 09:46:22 -07:00

qdq_4_rdna3.cuh

[ROCm][RDNA3] Fix W4A16 split-K accuracy and determinism (#54706)

2026-10-03 23:44:17 +08:00

skinny_gemms_int4.cu

[ROCm][Perf] W4A16: speed up skinny GEMM decode, including speculative batches (#57199)

2026-10-13 17:21:30 +00:00

skinny_gemms.cu

[Misc] Consolidate csrc shared headers and prune dead code (#60721)

2026-10-12 09:46:22 -07:00

torch_bindings.cpp

[Model][Hardware][AMD][Kernel] Part 2/2 -> Enable e2e QK Norm + RoPE + KV Cache runtime fusion for Qwen3-30B-A3B on ROCM_ATTN (#42755)

2026-10-15 17:10:14 -07:00