Skip to content

Conversation

@copybara-service
Copy link

Optimize remap ops in IFRT IR by performing more work at compile time

With two-stage program interpreter, we can now calculate input_devices_for_output_map and verify the remap plan at compile time without having to worry about their overheads. The former may improve the performance for runtime implementations that leverage the additional information.

With two-stage program interpreter, we can now calculate `input_devices_for_output_map` and verify the remap plan at compile time without having to worry about their overheads. The former may improve the performance for runtime implementations that leverage the additional information.

PiperOrigin-RevId: 841180291
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.

1 participant