Feat/cp nvshmem enhanced#2737

Open

Knight-of-Thunder wants to merge 24 commits intoNVIDIA:mainfrom

ETOgaosion:feat/cp_nvshmem_enhanced

Commits on Aug 18, 2025

Changed VERSION to 2.7.0
ptrendx
committed

Commits on Aug 26, 2025

[JAX] Error checking for mesh resource and update GemmPrimitive to use global_mesh_resource().fsdp_resource (#2088)

jberchtold-nvidia
authored and
KshitijLakhani
committed
[PyTorch] Avoid garbage collection when capturing a CUDA Graph (#2092)

timmoon10
authored and
KshitijLakhani
committed
Fix incorrect version checks for atomic GEMM (#2095)

authored and
KshitijLakhani
committed
[ TE-JAX ] Expose cp_strategy argument to DPA api (#2090)

kocchop
authored and
KshitijLakhani
committed
[PyTorch] Debug Mcore wgrad fusion with te.ops (#2097)

authored and
KshitijLakhani
committed
[JAX] Add Shardy warning in GEMM custom call (#2101)

phu0ngng
authored and
KshitijLakhani
committed
Revert "[Common] PDL for Quantization Kernels" (#2114)

jberchtold-nvidia
authored and
KshitijLakhani
committed
Bump cuDNN FE to 1.14.0 (#2072)

vcherepanov-nv
authored and
KshitijLakhani
committed
Revert "[Common] PDL for Blockwise Quantization" (#2115)

jberchtold-nvidia
authored and
KshitijLakhani
committed

Commits on Aug 27, 2025

Commits on Aug 28, 2025

Temporarily remove comm_gemm tests (#2133)

vcherepanov-nv
authored and
KshitijLakhani
committed
[PyTorch] Disable determinism for sm100 (#2130)

authored and
KshitijLakhani
committed

Commits on Jan 5, 2026

dev-base initialize: with version change and log verify
miceforrat
committed

Commits on Jan 6, 2026

Merge pull request #1 from ETOgaosion/dev-base
miceforrat
authored

Commits on Jan 21, 2026

Commits on Jan 26, 2026

feat: store fa result in nvshmem
Knight-of-Thunder
committed

Commits on Mar 5, 2026

Commits on Mar 10, 2026

feat: context_parallel_nvshmem_enhanced
Knight-of-Thunder
committed