Matthias Gehre dc2e8f1157 [ROCm][Perf] W4A16: keep skinny GEMM zero-points packed 4-bit (#54965) ...

Signed-off-by: Matthias Gehre <matthias.gehre@amd.com>

2026-09-14 17:40:25 +08:00

..

attention.cu

[Bugfix][ROCm] Mask paged attention V cache padding (#53856)

2026-09-08 14:39:23 +08:00

moe_q_gemm_rdna3.cu

[ROCm][Perf] Fused MoE W4A16 HIP kernel for AMD RDNA3 (gfx1100) (#44075)

2026-06-06 15:30:39 -05:00

ops.h

[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)

2026-09-08 12:45:12 -07:00

q_gemm_rdna3_wmma.cu

[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)

2026-09-08 12:45:12 -07:00

q_gemm_rdna3.cu

[Quant][Kernel] Remove GPTQ Group/Dynamic Activation Ordering (#54809)

2026-09-08 12:45:12 -07:00

qdq_4_rdna3.cuh

[Kernel][ROCm] Native W4A16 kernel for AMD RDNA3 (gfx1100) — fp16 + bf16 (#41394)

2026-05-29 11:04:40 +00:00

skinny_gemms_int4.cu

[ROCm][Perf] W4A16: keep skinny GEMM zero-points packed 4-bit (#54965)

2026-09-14 17:40:25 +08:00

skinny_gemms.cu

[ROCm][Bugfix][Perf] Tune multi-stream shared experts use; wvSplitKrc fixes (#56098)

2026-09-10 15:14:22 -07:00

torch_bindings.cpp

[ROCm][Perf] W4A16: keep skinny GEMM zero-points packed 4-bit (#54965)

2026-09-14 17:40:25 +08:00