Reapply #3621

[wujingyue]

No description provided.

[wujingyue]

!test

[github-actions]

PR Reviewer Guide 🔍

(Review updated until commit 6b23637)

Here are some key observations to aid the review process:

⏱️ Estimated effort to review: 3 🔵🔵🔵⚪⚪

🧪 No relevant tests

⚡ Recommended focus areas for review Logic Change The cacheInputs function signature has been modified to include a new parameter propagate_allocation. This change may affect the behavior of the function and its usage in other parts of the code. std::vector<TensorView*> cacheInputs( Fusion* fusion, bool unroll, bool propagate_allocation) { Function Call Update The cacheInputs function is now called with an additional argument propagate_allocation=true. This change may affect the behavior of the cacheInputsAndOutputs function. scheduler_utils::cacheInputs( fusion_, /*unroll=*/true, /*propagate_allocation=*/true); Function Call Update The cacheInputs function is now called with an additional argument propagate_allocation=true. This change may affect the behavior of the cacheInputsAndOutputs function. scheduler_utils::cacheInputs( fusion_, /*unroll=*/true, /*propagate_allocation=*/true);

[wujingyue]

!test