test(pt_expt): speed up AOTInductor UT compilation and fix hardcoded device - #5394

Closed

wanghan-iapcm wants to merge 4 commits into

deepmodeling:masterfrom

wanghan-iapcm:fix-pt2-default-device

Commits on Apr 12, 2026

fix(pt_expt): centralize default-device guard in AOTInductor compilation
Han Wang
committed
d69632e
test(pt_expt): reduce AOTInductor compile time with fast configs
Han Wang
committed
016ea5b
fix(pt_expt): revert centralized device guard, keep per-test workarounds
Han Wang
committed
2a687ce
fix(test): replace hardcoded cuda:9999999 with torch.get_default_device()
Han Wang
committed
8050d47