Skip to content

Conversation

@bangtianliu
Copy link
Contributor

@bangtianliu bangtianliu commented Nov 26, 2025

Per comment: https://github.com/nod-ai/amd-shark-ai/pull/2692/files#r2560936898, This PR brings two changes:

  • Remove promotion of operand 2 since codegen supports handling padded outputs without promotion.
  • Added promote_operands and res_type parameters to calculate_shared_memory_usage_in_bytes.

@bangtianliu bangtianliu requested review from Max191 and kuhar November 26, 2025 23:43
@bangtianliu bangtianliu force-pushed the smem_usage_tuner branch 2 times, most recently from 9bcbf1a to 24ffead Compare November 26, 2025 23:45
@bangtianliu
Copy link
Contributor Author

The failed cases (where compilation with the generated tuning spec was unsuccessful) revealed bugs in IREE. I’ve created an issue to track them: iree-org/iree#22777 (self-assigned).

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants