Search results

[Issue]: ck_tile::type_convert<float> not working for half to float

Status: Open.

#1901 In ROCm/aiter;

· michael604work opened on Jan 24, 2026

[Issue]: Not support conv3d triton kernel. the APP ComfyUI will call them.

Status: Open.

#1886 In ROCm/aiter;

· kxyk99 opened on Jan 22, 2026

[Documentation]: fused_fp8_quant docstrings incorrect

Status: Open.

#1819 In ROCm/aiter;

· tpopp opened on Jan 12, 2026

[KERNEL] AMDGPU@ROCM ramp up doc/link for kernel contributors

Status: Open.

#1802 In ROCm/aiter;

· carlushuang opened on Jan 9, 2026

[Issue]: Aiter setup.py causing arbitrary code execution

Status: Open.

#1785 In ROCm/aiter;

· andre-motta opened on Jan 7, 2026

[Feature]: Per tensor act + Per channel weight int8 gemm

Status: Open.

#1766 In ROCm/aiter;

· ZhiweiYan-96 opened on Jan 5, 2026

[Issue]: Does gemm_a8w8 support per-tensor quant?

Status: Open.

#1765 In ROCm/aiter;

· byjiang1996 opened on Jan 3, 2026

Add support for non-causal (encoder-only) attention

Status: Open.

#1702 In ROCm/aiter;

· westers opened on Dec 20, 2025

fp4 gemm aiter.gemm_a4w4 API to handle missing heuristics

Status: Open.

#1689 In ROCm/aiter;

· hongxiayang opened on Dec 19, 2025

[Feature]: Support pre-build multi arches for ck_gemm_moe_2stages_codegen

Status: Open.

#1641 In ROCm/aiter;

· yuguo68 opened on Dec 13, 2025

[Issue]: Deploy GLM-4.6 on MI300X failed. Error msg "aiter : wrapper_mha_batch_prefill() Expected a value of type 'int' for argument 'max_seqlen_k‘ but instead found type 'NoneType'

Status: Open.

#1549 In ROCm/aiter;

· naiveshen opened on Dec 3, 2025

[Issue]: AITER all reduce kernel segmentation fault in MI300x

Status: Open.

#1542 In ROCm/aiter;

· graceleeis opened on Dec 3, 2025