Fix Qwen3 decode all-reduce perf validation#270

Open

chhwang wants to merge 14 commits into

mainfrom

qwen3-allreduce-bench

Commits on Jun 17, 2026

Clean branch: packet all-reduce prereqs + torch-free bench
chhwang
committed
Ship Q7 decode all-reduce as a clean PR off main on branch qwen3-allreduce-bench: codegen OFFSET fix + ops copy-into-registered-buffer + 3 C++ regression tests + torch-free bench_allreduce.py + d2h…
chhwang
committed
Ship Q7 decode all-reduce as a clean PR off main on branch qwen3-allreduce-bench: codegen OFFSET fix + ops copy-into-registered-buffer + 3 C++ regression tests + torch-free bench_allreduce.py + d2h…
chhwang
committed
Q7 all-reduce: fix codegen_test.cpp include-order (clang-format), finalize test_allreduce.py d2h-safe equivalence tests, add bench caveats, and open the PR off main.
chhwang
committed

Commits on Jun 18, 2026

Commits on Jun 19, 2026

Commits on Jun 25, 2026

Report Qwen3 all-reduce internal input latency (#272)
chhwang
authored