Skip to content

v0.5.2

Choose a tag to compare

@shwestrick shwestrick released this 12 Jan 15:43
· 205 commits to main since this release

MPL-v0.5.2 Release Notes (Jan 12, 2025)

This release consists of two updates:

  • More robust support for Arm-based Macs. #214 fixed a weak memory bug in the scheduler. MPL now successfully runs all of the parallel-ml-bench without error (tested on a MacBook Air M2, 1-8 cores).
  • Major performance fix: promotion stacks (#211). Benchmarks with excessively deep call-stacks now perform as expected; for example, the trees/test-unbalanced example (which walks a highly unbalanced tall tree) now is orders of magnitude faster. Previously, deep call-stacks incurred a non-constant performance penalty at each PCall promotion. This fix extends the PCall calling convention with an auxiliary deque of promotable frames, ensuring that every promotion is constant-time.