[Kernel][SM70] Integrate concurrent QPN and verifier paths (#476) - #554

Merged

yangzhuxinyzx merged 12 commits into

mainfrom

codex/v100-sep07-concurrency-audit-20260907-100627

Sep 7, 2026

Commits on Sep 3, 2026

[Kernel][SM70] Extend NVFP4 QPN2 verifier batches
yangzhuxinyzx
committed
65f988c
[Kernel][SM70] Batch exact DFlash2 grouped verifier
yangzhuxinyzx
committed
028bca7
[Core][SM70] Admit batched compact DFlash2 rejection
yangzhuxinyzx
committed
bb487af
[Doc] Record Qwen3.8 DFlash2 concurrency audit
yangzhuxinyzx
committed
5154106
[Doc] Record DFlash2 endpoint scaling
yangzhuxinyzx
committed
4507907
[Kernel][SM70] Optimize DFlash2 concurrency projections
yangzhuxinyzx
committed
086cdee
[Kernel][SM70] Reuse NVFP4 weights across B2 verifier rows
yangzhuxinyzx
committed
d444517

Commits on Sep 4, 2026

Commits on Sep 7, 2026

[Kernel][SM70] Integrate concurrent QPN projection and verifier improvements

yangzhuxinyzx
and
codex
committed
1504c0c
[Core][SM70] Compose concurrency paths with current E4M3 precision main

yangzhuxinyzx
and
codex
committed
64ecce6