fix(GFX1100-TG200): skinny_gemm OOB write + MoE silu gate rounding - #2894

Open

ghazni101 wants to merge 15 commits into

mudler:mainfrom

ghazni101:row/GFX1100-TG200-BUGFIX

Commits on Sep 6, 2026

feat(KERNEL-QUANT-CIQ-GEMM-ROCM): land the W1 keep-quant providers on kROCM
ghazni101
committed
8ab75ca
fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider

mudler
authored and
ghazni101
committed
c42eb83
Revert "fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider"

mudler
authored and
ghazni101
committed
b63c7ea
fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider

mudler
authored and
ghazni101
committed
867e977
spec(BACKEND-ROCM): define Qwen3.5 gfx1100 numerical gate

VikashLoomba
authored and
ghazni101
committed
c305ead
spec(BACKEND-ROCM): repair Qwen3.5 numerical plan

VikashLoomba
authored and
ghazni101
committed
71f8995
spec(BACKEND-ROCM): correct Qwen3.5 capture contracts

VikashLoomba
authored and
ghazni101
committed
712c97a
perf(GFX1100-TG200): T4a MMVQ K-quant decode GEMV arm
ghazni101
committed
5df3368
feat(GFX1100-TG200): lever-C fuses Q8_K activation quant into the RmsNorm epilogue
ghazni101
committed
9e7f889
T27: warp-cooperative QuantizeQ8KK for decode (+2.06%, byte-identical)
ghazni101
committed
b11dc53
feat(GFX1100-TG200): T24 LDS-buffered quant epilogue in RmsNormRowCoopKernel
ghazni101
committed
067713f
feat(GFX1100-TG200): T21 keep-quant for V-head row-permuted GDN projections
ghazni101
committed
da3d269
fix(rocm): gate the coop rmsnorm vector body to bf16 gamma, seed quant amax at 0
ghazni101
committed
e66b8d1
T25: keep ssm_out as Q5_K with runtime input permutation
ghazni101
committed
a3c6108
fix(GFX1100-TG200): guard the skinny-GEMM tail store against the partial M tile
ghazni101
committed
361886c

Navigation Menu

fix(GFX1100-TG200): skinny_gemm OOB write + MoE silu gate rounding - #2894

fix(GFX1100-TG200): skinny_gemm OOB write + MoE silu gate rounding#2894
ghazni101 wants to merge 15 commits into
mudler:mainfrom
ghazni101:row/GFX1100-TG200-BUGFIX

Commits on Sep 6, 2026

feat(KERNEL-QUANT-CIQ-GEMM-ROCM): land the W1 keep-quant providers on kROCM

fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider

Revert "fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider"

fix(KERNEL-QUANT-CIQ-GEMM-ROCM): keep grouped Q8_0 on the kROCM provider

spec(BACKEND-ROCM): define Qwen3.5 gfx1100 numerical gate

spec(BACKEND-ROCM): repair Qwen3.5 numerical plan

spec(BACKEND-ROCM): correct Qwen3.5 capture contracts

perf(GFX1100-TG200): T4a MMVQ K-quant decode GEMV arm

feat(GFX1100-TG200): lever-C fuses Q8_K activation quant into the RmsNorm epilogue

T27: warp-cooperative QuantizeQ8KK for decode (+2.06%, byte-identical)

feat(GFX1100-TG200): T24 LDS-buffered quant epilogue in RmsNormRowCoopKernel

feat(GFX1100-TG200): T21 keep-quant for V-head row-permuted GDN projections

fix(rocm): gate the coop rmsnorm vector body to bf16 gamma, seed quant amax at 0

T25: keep ssm_out as Q5_K with runtime input permutation

fix(GFX1100-TG200): guard the skinny-GEMM tail store against the partial M tile