Skip to content

Add HiFi-optimized reduce operations for Xtensa - #3663

Open
unmeshna017 wants to merge 6 commits into
tensorflow:mainfrom
unmeshna017:xtensa-reduce-hifi
Open

Add HiFi-optimized reduce operations for Xtensa#3663
unmeshna017 wants to merge 6 commits into
tensorflow:mainfrom
unmeshna017:xtensa-reduce-hifi

Conversation

@unmeshna017

Copy link
Copy Markdown
Contributor
  1. Add HiFi3/4/5-optimized REDUCE_MEAN and REDUCE_MAX kernels using xa_nn_reduce_* NNLib APIs.
  2. Supports int8, int16, and float32 types.
  3. Removes anonymous namespace in reduce_common.cc to expose PrepareSimple/ResolveAxis helpers to reduce_hifi.cc

@cad-audio @joshih-cad

@unmeshna017
unmeshna017 requested a review from a team as a code owner August 13, 2026 07:04
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