mscclpp

mirror of https://github.com/microsoft/mscclpp.git synced 2026-07-18 09:47:34 +00:00

Files

Binyang Li 02c65cfbcd Support hidden size 6656 in low-latency EP (#835 )

Instantiate low-latency dispatch and combine kernels for hidden size
6656 and expose the shape through the functional and benchmark entry
points. Reuse syncNamedBarrier for scheduler named barriers instead of
inline PTX.

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

2026-07-14 16:33:25 -07:00

CMakeLists.txt

Optimized LL implementation (#834 )

2026-07-14 16:09:02 -07:00

cupti_kernel_timer.cpp

Optimized LL implementation (#834 )

2026-07-14 16:09:02 -07:00

ep_bench_ll.py

Support hidden size 6656 in low-latency EP (#835 )

2026-07-14 16:33:25 -07:00

mscclpp_ep_bench.cu

Support hidden size 6656 in low-latency EP (#835 )