List view

Make GraphicsResource inherit from Buffer

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

P0High priority - Must do!High priority - Must do!

Status: Open (in progress).

NVIDIA/cuda-pythonnumber 1701

#1701 In NVIDIA/cuda-python;

· rparolin opened on Feb 27, 2026

[BUG]: cuda.core.experimental._stream.IsStreamT and cuda.core.experimental._memory.DevicePointerT missing in v0.5.0

bugSomething isn't workingSomething isn't working

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

P0High priority - Must do!High priority - Must do!

Status: Open.

#1419 In NVIDIA/cuda-python;

· xiakun-lu opened on Dec 22, 2025

[EPIC] Support CUDA graphs

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

P0High priority - Must do!High priority - Must do!

Status: Open.

Epic

#111 In NVIDIA/cuda-python;

· leofang opened on Sep 19, 2024

EPIC: Support TMA descriptor

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

P1Medium priority - Should doMedium priority - Should do

Status: Open.

Epic

#199 In NVIDIA/cuda-python;

· leofang opened on Oct 30, 2024

[EPIC] Support green contexts

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

EPICSoul of a releaseSoul of a release

featureNew feature or requestNew feature or request

P1Medium priority - Should doMedium priority - Should do

Status: Open.

Epic

#112 In NVIDIA/cuda-python;

· leofang opened on Sep 19, 2024

GreenContext: Support allocating SMs

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

Status: Open.

#1563 In NVIDIA/cuda-python;

· leofang opened on Feb 2, 2026

Add TMA TensorMapDescriptor support

Status: Open (in progress).

NVIDIA/cuda-pythonnumber 1687

#1687 In NVIDIA/cuda-python;

· rparolin opened on Feb 24, 2026

Design the TensorMap object

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

P1Medium priority - Should doMedium priority - Should do

Status: Open.

#200 In NVIDIA/cuda-python;

· leofang opened on Oct 30, 2024

[BUG]: numba-cuda SEGV with cuda.core when <code>cuLibraryUnload</code> is Enabled

bugSomething isn't workingSomething isn't working

Status: Open.

Bug

#1670 In NVIDIA/cuda-python;

· Andy-Jost opened on Feb 20, 2026

RFC: Cythonize <code>cuda.core</code> while keeping it CUDA-agnostic

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

P0High priority - Must do!High priority - Must do!

packagingAnything related to wheels or Conda packagesAnything related to wheels or Conda packages

RFCPlans and announcementsPlans and announcements

Status: Open.

#866 In NVIDIA/cuda-python;

· leofang opened on Aug 20, 2025

ci: remove build-ctk-ver from doc builds

CI/CDCI/CD infrastructureCI/CD infrastructure

enhancementAny code-related improvementsAny code-related improvements

P0High priority - Must do!High priority - Must do!

Status: Open (in progress).

NVIDIA/cuda-pythonnumber 1503

#1503 In NVIDIA/cuda-python;

· cpcloud opened on Jan 15, 2026

Cythonize <code>_graph.py</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

P1Medium priority - Should doMedium priority - Should do

Status: Open.

#1076 In NVIDIA/cuda-python;

· leofang opened on Oct 3, 2025

Cythonize <code>_linker.py</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

Status: Open.

#1079 In NVIDIA/cuda-python;

· leofang opened on Oct 3, 2025

Add tests for <code>_decide_nvjitlink_or_driver()</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

P2Low priority - Nice to haveLow priority - Nice to have

testImprovements or additions to testsImprovements or additions to tests

Status: Open.

#1095 In NVIDIA/cuda-python;

· leofang opened on Oct 7, 2025

Support managed memory <code>advise</code>, <code>prefetch</code>, and <code>discard-prefetch</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

P1Medium priority - Should doMedium priority - Should do

Status: Open.

#1332 In NVIDIA/cuda-python;

· leofang opened on Dec 6, 2025

[FEA]: Tests for <code>ltoir</code>, <code>fatbin</code>, <code>object</code>, and <code>library</code> constructors

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

P1Medium priority - Should doMedium priority - Should do

testImprovements or additions to testsImprovements or additions to tests

Status: Open.

#663 In NVIDIA/cuda-python;

· brandon-b-miller opened on May 29, 2025

Python latency testing &amp; benchmarking

triageNeeds the team's attentionNeeds the team's attention

Status: Open.

#1580 In NVIDIA/cuda-python;

· leofang opened on Feb 5, 2026

[BUG]: IPC-enabled pinned pool uses a fixed host NUMA node

bugSomething isn't workingSomething isn't working

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

Status: Open.

Bug

#1603 In NVIDIA/cuda-python;

· rwgk opened on Feb 11, 2026

Provide a fast path for constructing a <code>StridedMemoryView</code> from a <code>cupy.ndarray</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

enhancementAny code-related improvementsAny code-related improvements

P1Medium priority - Should doMedium priority - Should do

Status: Open.

#748 In NVIDIA/cuda-python;

· leofang opened on Jul 8, 2025

Implement Context.__init__() in cuda.core.experimental._context

blockedThis task is currently blocked by other tasksThis task is currently blocked by other tasks

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

Status: Open.

Enhancement

#189 In NVIDIA/cuda-python;

· keenan-simpson opened on Oct 25, 2024

ManagedMemoryResource() without options fails when the default pool does not support managed allocations

bugSomething isn't workingSomething isn't working

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

P1Medium priority - Should doMedium priority - Should do

Status: Open.

Bug

#1617 In NVIDIA/cuda-python;

· Andy-Jost opened on Feb 12, 2026

Update <code>pyproject.toml</code> for <code>uv</code>

cuda.bindingsEverything related to the cuda.bindings moduleEverything related to the cuda.bindings module

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

supportAll things related to the project that can't be categorizedAll things related to the project that can't be categorized

Status: Open.

#1247 In NVIDIA/cuda-python;

· leofang opened on Nov 17, 2025

Feature Request: Device context manager for temporary device switching

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

triageNeeds the team's attentionNeeds the team's attention

Status: Open.

#1586 In NVIDIA/cuda-python;

· bdice opened on Feb 6, 2026

Add Device context manager for temporary device switching

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

Status: Draft (not ready).

NVIDIA/cuda-pythonnumber 1597

#1597 In NVIDIA/cuda-python;

· Andy-Jost opened on Feb 11, 2026

WIP: Support passing SMV as a kernel argument of type <code>cuda::std::mdspan</code>

cuda.coreEverything related to the cuda.core moduleEverything related to the cuda.core module

featureNew feature or requestNew feature or request

Status: Draft (not ready).

NVIDIA/cuda-pythonnumber 1387

#1387 In NVIDIA/cuda-python;

· leofang opened on Dec 16, 2025