1.25.0 rc - #300

Merged

Anerudhan merged 36 commits into

mainfrom

Jun 10, 2026

Commits on May 21, 2026

test/python: cap peak GPU memory via PYTORCH_CUDA_ALLOC_CONF (#247)
vedaanta
authored
1c3d490

Commits on May 23, 2026

Fix DSA link in README.md
Anerudhan
committed
7ee0491

Commits on May 26, 2026

Remove stale H200 benchmark artifacts (#252)
hwanseoc
authored
b4fb358
Change profile_pass from 'fwd' to 'both'

Vinnie6167
authored and
Anerudhan
committed
cb74feb
Bump the develop to 1.25.0
Anerudhan
committed
8ec19b9
Fix varpack-template lifecycle bugs + add defensive checks

Anerudhan
and
claude
committed
3f72bd4
Clear deserialize-owned containers on re-deserialize

Anerudhan
and
claude
committed
5bff9bd

Commits on May 28, 2026

Add row-scale support to grouped GEMM quant

zianglih
authored and
Anerudhan
committed
02d2471
Tighten row-scale grouped GEMM quant tests

zianglih
authored and
Anerudhan
committed
1f8cde3

Commits on May 29, 2026

feat(python): add get_engine_and_knobs_at_index for structured plan pinning (#259)

YangXu1990uiuc
and
claude
authored
7cd11d7

Commits on Jun 1, 2026

Update SDPA Benchmarking Artifacts (#265)
brandonfzhang
authored
1ad529d

Commits on Jun 2, 2026

Adding coderabbit review guide (initial template)
Anerudhan
committed
e851cc8
fix: allow overriding libcudart selection via CUDNN_FRONTEND_CUDART_LIB_NAME

Anerudhan
and
claude
committed
f06a076

Commits on Jun 3, 2026

Clean up guardword-flagged comments (xmma path, gitlab URL, P4 label, Perfsim, HACK/Ugly, STS/CGA SASS terms) (#273)

liujane-dev
and
claude
authored
28462c3
remove_9.99_version_tag

yanqinz2
authored and
Anerudhan
committed
4d65508
add_protection_flags

yanqinz2
authored and
Anerudhan
committed
69e5b42
fix(windows): consolidate getenv access and fix C4996/C4005 on MSVC

Anerudhan
and
claude
committed
c85a306
fix(shim): warn instead of throwing when multiple libcudart libraries are found

Anerudhan
and
claude
committed
1a2799b

Commits on Jun 4, 2026

Unblock SDPA tests and promote FP8 ragged backward to L0 (#275)
hwanseoc
authored
f90a2ec
Add per-expert reduction (group_offset) for MoE grouped GEMM

authored and
Anerudhan
committed
1770caf
Fix the 9.99 bound
Anerudhan
committed
035b520

Commits on Jun 5, 2026

Skip flexible-graph SDPA bwd sample on SM120 and above (#284)

authored
d214a11
1

yanqinz2
authored and
Anerudhan
committed
3542eb0
Add pre-commit hooks (#286)
Anerudhan
authored
28b5837
Skip TensorIR MemBound / compile-time-const samples on consumer Blackwell (SM12x) (#285)

YangXu1990uiuc
and
claude
authored
dbcc4a8
Support cu_seqlens in unified SDPA (#266)
egilliam-nv
authored
2965e7a

Commits on Jun 8, 2026

Commits on Jun 9, 2026

Fix formatting issues from #263 (#294)
Anerudhan
authored
1f837f9
Support static linking of libcudnn (#182)

take-cheeze
and
coderabbitai[bot]
authored
436643d
make dgeglu config values compile time constants instead of runtime values (#293)
saltyminty
authored
601b84d
bench: add autoregressive video DiT SDPA config + GB200/GB300 results (#277) (#295)

authored
3d4aef4
- Update the Black version. (#296)
Anerudhan
authored
d6e3edc

Commits on Jun 10, 2026

Add ragged offset multiplier support (#290)
egilliam-nv
authored
65f40b9
Merge develop into 1.25.0-rc
Anerudhan
committed
259e895
Fix unused ragged offset version error variable (#299)

shraiysh
authored and
Anerudhan
committed
e8e219d