Search results

[Bug] DGX Spark (sm_121): CUTLASS can_implement() rejects sm_120f binaries

bugSomething isn't workingSomething isn't working

Status: Open.

#36835 In vllm-project/vllm;

· LironKesem opened on Mar 12, 2026

[Bug]: GLM-4.7-Flash does not return tool_calls field in vLLM 0.16.0 even with --tool-call-parser glm47

bugSomething isn't workingSomething isn't working

Status: Open.

#36833 In vllm-project/vllm;

· CodeofGame opened on Mar 12, 2026

[Bug]: delta_text and delta_token_ids get out of sync when stop sequences are used.

bugSomething isn't workingSomething isn't working

Status: Open.

#36830 In vllm-project/vllm;

· maxdebayser opened on Mar 11, 2026

[Bug]: vLLm 0.17.1 (docker) crash with Qwen 3.5 27B-FP8 in BatchPrefillWithPagedKVCache

bugSomething isn't workingSomething isn't working

Status: Open.

#36828 In vllm-project/vllm;

· matatonic opened on Mar 11, 2026

[Bug]: Streaming stalls and can crash when concurrent requests hit the same vLLM server

bugSomething isn't workingSomething isn't working

Status: Open.

#36826 In vllm-project/vllm;

· imi187 opened on Mar 11, 2026

[Bug]: No sm_121 (Blackwell) support on aarch64 — NVIDIA DGX Spark / Acer GN100

bugSomething isn't workingSomething isn't working

Status: Open.

#36821 In vllm-project/vllm;

· blogtheristo opened on Mar 11, 2026

[Bug]: CUDA illegal memory access on GPTQ Marlin

bugSomething isn't workingSomething isn't working

Status: Open.

#36811 In vllm-project/vllm;

· tonibofarull opened on Mar 11, 2026

[Bug]: Misleading error message for FP8 quantization: refers to CUDA version instead of GPU compute capability

bugSomething isn't workingSomething isn't working

rocmRelated to AMD ROCmRelated to AMD ROCm

Status: Open.

#36805 In vllm-project/vllm;

· jranaraki opened on Mar 11, 2026

[Feature]: FA4 Attention Sinks

feature requestNew feature or requestNew feature or request

Status: Open.

#36804 In vllm-project/vllm;

· pbarker-synth opened on Mar 11, 2026

[Bug]: Tesla T4 GPU - triton.runtime.errors.OutOfResources: out of resource: shared memory, Required: 81920, Hardware limit: 65536. Reducing block sizes or num_stages

bugSomething isn't workingSomething isn't working

Status: Open.

#36802 In vllm-project/vllm;

· ksecurity45 opened on Mar 11, 2026

[Bug]: CPU offload errors on nightly with NVIDIA GH200 Unified Memory (UMA)

bugSomething isn't workingSomething isn't working

torch.compile

Status: Open.

#36796 In vllm-project/vllm;

· ehfd opened on Mar 11, 2026

[Feature]: Improve Error Message When Max Length Reached With Tool Call=Required

feature requestNew feature or requestNew feature or request

Status: Open.

#36794 In vllm-project/vllm;

· SvenLorenz opened on Mar 11, 2026