Skip to content

Performance improvements

Alejandro Homs Puron requested to merge performance into master

The branch implements several performance improvements, notably 4-to-8 bit expansion with a SIMD algorithm. The structure of auxiliary threads was completely redesigned: from Recv::Port-based to Eiger::Recv-based. The Recv::Port callbacks now block until the Eiger::Recv::Threads finish frame processing.

This branch should be merged once 92-memutils-softbufferallocmgr-fix-issues-when-allocating-more-hw-buffers is merged.

Edited by Alejandro Homs Puron

Merge request reports