Скачать или смотреть [ICFP'22] Reference Counting with Frame Limited Reuse

Reference Counting with Frame Limited Reuse (Video, ICFP 2022)
Anton Lorenzen and Daan Leijen
(University of Bonn, Germany; Microsoft Research, USA)

Abstract: The recently introduced Perceus algorithm can automatically insert
reference count instructions such that the resulting (cycle-free) program is
_garbage free_: objects are freed at the very moment they can no longer be
referenced. An important extension is reuse analysis. This optimization pairs
objects of known size with fresh allocations of the same size and tries to
reuse the object in-place at runtime if it happens to be unique. Unfortunately,
current implementations of reuse analysis are fragile with respect to small
program transformations, or can cause an arbitrary increase in the peak heap
usage. We present a novel drop-guided reuse algorithm that is simpler and
more robust than previous approaches. Moreover, we generalize the linear
resource calculus to precisely characterize garbage-free and frame-limited
evaluations. On each function call, a frame-limited evaluation may hold on to
memory longer if the size is bounded by a constant factor. Using this framework
we show that our drop-guided reuse is frame-limited and find that an
implementation of our new reuse approach in Koka can provide significant
speedups.

Article: https://doi.org/10.1145/3547634

Supplementary archive: https://doi.org/10.5281/zenodo.6946310 (Badges: Artifacts Available, Artifacts Evaluated , Reusable)

Supplementary web page: https://www.microsoft.com/en-us/resea...

ORCID: https://orcid.org/0000-0003-3538-9688, https://orcid.org/0000-0003-1027-5430

Video Tags: Reference Counting, Reuse, Frame Limited, Koka, icfp22main-p41-p, doi:10.1145/3547634, doi:10.5281/zenodo.6946310, orcid:0000-0003-3538-9688, orcid:0000-0003-1027-5430, Artifacts Available, Artifacts Evaluated , Reusable

Presentation at the ICFP 2022 conference, September 11-16, 2022, https://icfp22.sigplan.org/
Sponsored by ACM, ACM SIGPLAN, https://www.sigplan.org/