perf(test): reduce redundant .pt2 compilations in GPU CI#5379

Merged

wanghan-iapcm merged 4 commits into

deepmodeling:masterfrom

wanghan-iapcm:fix-gpu-ci-efficiency

Apr 7, 2026

Commits on Apr 6, 2026

perf(test): reduce redundant .pt2 AOTInductor compilations in GPU CI
Han Wang
committed
perf(test): remove stale .pt2/.pte before regeneration in CI
Han Wang
committed
fix: address reviewer comments on BUILD_PT_EXPT guards and wait propagation
Han Wang
committed
fix: add TearDownTestSuite to release CUDA resources before static destruction
Han Wang
committed