Search results

GC segfault with tensors and free threading

Status: Open.

#169136�?In pytorch/pytorch;

· kevmo314 opened on Nov 26, 2025

vLLM tlparse sometimes missing compilation_metrics

Status: Open.

#169133�?In pytorch/pytorch;

· zou3519 opened on Nov 26, 2025

Higher-order ops like cond, scan should work with func torch (torch.func.functional_call)

Status: Open.

#169120�?In pytorch/pytorch;

· BillHuang2001 opened on Nov 26, 2025

<code>torch.compile(backend="cudagraphs", fullgraph=True, cuda)</code> fails for sparse CSR → COO (<code>torch.compile does not support sparse Tensors</code>)

Status: Open.

#169119�?In pytorch/pytorch;

· tinywisdom opened on Nov 26, 2025

AOTI codegen flips the order of buffer mutation and torch.cond

module: aotinductoraot inductoraot inductor

oncall: export

oncall: pt2

Status: Open.

#169118�?In pytorch/pytorch;

· larryliu0820 opened on Nov 26, 2025

Bottleneck in create_read_items_for_chunk_list when resharding checkpoints at scale

Status: Open.

#169113�?In pytorch/pytorch;

· mansiag05 opened on Nov 26, 2025

<code>torch.compile(fullgraph=True, dynamic=True)</code> on CUDA fails when using <code>torch.utils.dlpack.to_dlpack</code> / <code>from_dlpack</code> (<code>torch._C._to_dlpack</code> skipped by Dynamo)

Status: Open.

#169112�?In pytorch/pytorch;

· tinywisdom opened on Nov 26, 2025

<code>torch.compile(backend="tvm", fullgraph=True, dynamic=True)</code> fails on PyTorch 2.9.0 with <code>RuntimeError: Tracer cannot infer type ... Only tensors ... supported, but got SymInt</code>

Status: Open.

#169111�?In pytorch/pytorch;

· tinywisdom opened on Nov 26, 2025

Why is fusion restricted here in dynamic mode?

module: inductor

oncall: pt2

Status: Open.

#169106�?In pytorch/pytorch;

· Jin-TaoZhang opened on Nov 26, 2025

[RFC] Implement StoreExchange.broadcast

module: symm_memIssues and PRs of Symmetric MemoryIssues and PRs of Symmetric Memory

oncall: distributedAdd this issue/PR to distributed oncall triage queueAdd this issue/PR to distributed oncall triage queue

Status: Open.

#169099�?In pytorch/pytorch;

· kwen2501 opened on Nov 26, 2025

[CI] driver outdated issue on AWS A100 runner

module: ciRelated to continuous integrationRelated to continuous integration

triagedThis issue has been looked at a team member, and triaged and prioritized into an appropriate moduleThis issue has been looked at a team member, and triaged and prioritized into an appropriate module

Status: Open.

#169093�?In pytorch/pytorch;

· tinglvv opened on Nov 26, 2025

cuDNN version incompatibility in CUDA 13 inductor build

module: cudaRelated to torch.cuda, and CUDA support in generalRelated to torch.cuda, and CUDA support in general

module: inductor

oncall: pt2

Status: Open.

#169092�?In pytorch/pytorch;

· tinglvv opened on Nov 26, 2025