Search results

Ascend950环境编译测试

Status: Open.

#631 In sgl-project/sgl-kernel-npu;

· IceShip423 opened on Jul 28, 2026

Decouple DeepEP and attention kernel compilation.

Status: Open.

#625 In sgl-project/sgl-kernel-npu;

· kaniel-outis opened on Jul 27, 2026

文档写的不够详细

Status: Open.

#622 In sgl-project/sgl-kernel-npu;

· xingLong-xl opened on Jul 27, 2026

[compatibility] the int8 quantization retains the original community mode.

Status: Open.

#620 In sgl-project/sgl-kernel-npu;

· kaniel-outis opened on Jul 25, 2026

The A5 fusion operator Cutlass utilizes a submodule.

Status: Open.

#612 In sgl-project/sgl-kernel-npu;

· kaniel-outis opened on Jul 22, 2026

Why is there magic code “1536” in the code? Does it have any special meaning?

Status: Open.

#606 In sgl-project/sgl-kernel-npu;

· Abbottmo opened on Jul 22, 2026

Performance regression in causal_conv1d

Status: Open.

#596 In sgl-project/sgl-kernel-npu;

· ChristosMatzoros opened on Jul 13, 2026

[Bug] mega_chunk_gdn unsupported head config for Qwen3.6-27B (28 heads GQA)

Status: Open.

#569 In sgl-project/sgl-kernel-npu;

· JosephNew opened on Jun 24, 2026

[Bug] mega_chunk_gdn unsupported head config for Qwen3.6-27B GDN attention

Status: Open.

#568 In sgl-project/sgl-kernel-npu;

· JosephNew opened on Jun 24, 2026

[SGL-KERNEL-NPU] support of A5 is required

Status: Open.

#544 In sgl-project/sgl-kernel-npu;

· VDV1985 opened on Jun 4, 2026

Current state of kernel tests

Status: Open.

#527 In sgl-project/sgl-kernel-npu;

· fiskrt opened on May 29, 2026

[NPU/Ascend] RuntimeError: unsupported mega_chunk_gdn (NumValueHeads, NumKeyHeads) pair on Qwen3-6-27B with GDN backend

Status: Open.

#522 In sgl-project/sgl-kernel-npu;

· AndyLi429 opened on May 27, 2026