Chm02 CUDA: mainline correctness fixes, tests, and timing surface - #84

Merged

tpn merged 15 commits into

mainfrom

issue-79-chm02-cuda-mainline

Apr 9, 2026

Commits on Mar 31, 2026

Chm02 CUDA: advance single-graph bring-up
tpn
committed
7c06b9c
Chm02 CUDA: fix no-file-io completion path
tpn
committed
e5f2696
Chm02 CUDA: add GPU assignment and verify path
tpn
committed
1425eaa
Chm02 CUDA: fix Linux compat file-work path
tpn
committed
0d09976
Chm02 CUDA: validate peel order against CPU oracle
tpn
committed
9daa957
Chm02 CUDA: add peel-order validation and GPU verify
tpn
committed
3613919
Chm02 CUDA: add first-class regression tests
tpn
committed
c627a56
Chm02 CUDA: add generated non-Assigned16 regression
tpn
committed
dd72151
Chm02 CUDA: add explicit timing surface
tpn
committed
03f0b10
Chm02 CUDA: address review findings
tpn
committed
ccdcd5e
Chm02 CUDA: expand mainline design note
tpn
committed
c119c5b
Chm02 CUDA: tighten review contracts
tpn
committed
7507e4b
Chm02 CUDA: fix managed bitmap and test contract
tpn
committed
302f13b

Commits on Apr 8, 2026

Chm02 CUDA: add non-debug regression coverage
tpn
committed
67f957a

Commits on Apr 9, 2026

Chm02 CUDA: free verify scratch on error
tpn
committed
4f65fc3