Skip to content

Commit 53b4374

Browse files
leofangmdboom
andauthored
Prepare cuda.core v1.2.0 release (#2757)
* Prepare cuda.core v1.2.0 release Fill remaining user-visible entries in the 1.2.0 release notes and add 1.1.1 and 1.2.0 to the docs version switcher. Version is derived from the git tag via setuptools-scm, so no pyproject.toml bump is needed. New features documented: - ProgramOptions.use_bundled_headers for NVRTC 13.3+ (#2753 closes #2363) - cuda-gdb source display for JIT-compiled kernels (#2678, #2679) Fixes and enhancements documented: - PinnedMemoryResource host-pool validation at construction (#2487) - is_host_accessible for NUMA-located VMM (#2503) - Graph predecessor/successor query truncation on large graphs (#2587) - Per-domain NVML clock queries treated as independently optional (#2651) - Temperature threshold checks forward-compat with unknown archs (#2488) - import cuda.core non-integer opt-out crash (#2535) - Frozen CUresult explanation table refreshed for CUDA 13.3 (#2383) -- Leo's bot * Update cuda_core/docs/source/release/1.2.0-notes.rst Co-authored-by: Michael Droettboom <[email protected]> --------- Co-authored-by: Michael Droettboom <[email protected]>
1 parent 3df6115 commit 53b4374

2 files changed

Lines changed: 57 additions & 0 deletions

File tree

‎cuda_core/docs/nv-versions.json‎

Lines changed: 8 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -3,6 +3,14 @@
33
"version": "latest",
44
"url": "https://nvidia.github.io/cuda-python/cuda-core/latest/"
55
},
6+
{
7+
"version": "1.2.0",
8+
"url": "https://nvidia.github.io/cuda-python/cuda-core/1.2.0/"
9+
},
10+
{
11+
"version": "1.1.1",
12+
"url": "https://nvidia.github.io/cuda-python/cuda-core/1.1.1/"
13+
},
614
{
715
"version": "1.1.0",
816
"url": "https://nvidia.github.io/cuda-python/cuda-core/1.1.0/"

‎cuda_core/docs/source/release/1.2.0-notes.rst‎

Lines changed: 49 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -37,6 +37,22 @@ New features
3737
(`#2456 <https://github.com/NVIDIA/cuda-python/pull/2456>`__,
3838
`#1334 <https://github.com/NVIDIA/cuda-python/issues/1334>`__)
3939

40+
- Added :attr:`ProgramOptions.use_bundled_headers`, which lets NVRTC
41+
resolve the CUDA and CCCL headers from the toolkit bundled with NVRTC
42+
itself, installed into a per-user cache directory, instead of requiring
43+
a full CUDA Toolkit installation on the compile host. NVRTC backend
44+
only; requires NVRTC 13.3 or newer.
45+
(`#2753 <https://github.com/NVIDIA/cuda-python/pull/2753>`__,
46+
closes `#2363 <https://github.com/NVIDIA/cuda-python/issues/2363>`__)
47+
48+
- When a :class:`Program` is compiled with ``debug`` or ``lineinfo``,
49+
the NVRTC input source is now materialized as a temporary ``.cu`` file
50+
so ``cuda-gdb`` can list the original source while stepping through
51+
JIT-compiled kernels. ``#include "..."`` search still resolves against
52+
the original source directory.
53+
(`#2678 <https://github.com/NVIDIA/cuda-python/pull/2678>`__,
54+
`#2679 <https://github.com/NVIDIA/cuda-python/pull/2679>`__)
55+
4056
Fixes and enhancements
4157
----------------------
4258

@@ -196,6 +212,39 @@ Fixes and enhancements
196212
from being treated as CUDA devices by :meth:`Device.get_all_devices`, examples,
197213
and tests.
198214

215+
- :class:`PinnedMemoryResource` now rejects unsupported host memory pools
216+
at construction with ``RuntimeError``, instead of letting a later
217+
allocation or copy fail with ``CUDA_ERROR_INVALID_VALUE``.
218+
(`#2487 <https://github.com/NVIDIA/cuda-python/pull/2487>`__)
219+
220+
- :attr:`VirtualMemoryResource.is_host_accessible`, and by extension
221+
:attr:`Buffer.is_host_accessible`, now correctly return ``True`` for a
222+
resource configured with ``location_type="host_numa"`` or
223+
``"host_numa_current"``. Previously both properties reported ``False``
224+
on those NUMA-located variants.
225+
(`#2503 <https://github.com/NVIDIA/cuda-python/pull/2503>`__)
226+
227+
- Graph predecessor and successor queries no longer truncate their results
228+
on large graphs.
229+
(`#2587 <https://github.com/NVIDIA/cuda-python/pull/2587>`__)
230+
231+
- Per-domain clock queries in :mod:`cuda.core.system` treat each domain
232+
(minimum, maximum, and current) as independently optional, so an
233+
unsupported domain no longer fails the whole clock query for a device.
234+
(`#2651 <https://github.com/NVIDIA/cuda-python/pull/2651>`__)
235+
236+
- Temperature threshold checks in :mod:`cuda.core.system` are now
237+
forward-compatible with GPU architectures newer than the generated
238+
``DeviceArch`` enum. An unrecognized architecture no longer raises
239+
``ValueError`` before the query runs.
240+
(`#2488 <https://github.com/NVIDIA/cuda-python/pull/2488>`__)
241+
242+
- The frozen fallback ``CUresult`` explanation table, used when the
243+
driver's ``cuGetErrorName`` / ``cuGetErrorString`` are unavailable, is
244+
refreshed for CUDA 13.3 and now recognizes
245+
``CUDA_ERROR_GRAPH_RECAPTURE_FAILURE``.
246+
(`#2383 <https://github.com/NVIDIA/cuda-python/pull/2383>`__)
247+
199248
Deprecation Notices
200249
-------------------
201250

0 commit comments

Comments
 (0)