Changeset 8eeca77


Ignore:
Timestamp:
Apr 27, 2022, 2:34:25 PM (20 months ago)
Author:
Peter A. Buhr <pabuhr@…>
Branches:
ADT, ast-experimental, master, pthread-emulation, qualifiedEnum
Children:
4994d67
Parents:
37ef5e41
Message:

small changes to conclusion chapter

File:
1 edited

Legend:

Unmodified
Added
Removed
  • doc/theses/mubeen_zulfiqar_MMath/conclusion.tex

    r37ef5e41 r8eeca77  
    3232Finally, contention-free statistics gathering and debugging have a low enough cost to be used in production code.
    3333
    34 The ability to compile llheap static/dynamic linking and optional statistics/debugging provides programers with multiple mechanisms to balance performance and safety.
    35 These allocator versions are easy to use because can be linked to an application without recompilation.
     34The ability to compile llheap with static/dynamic linking and optional statistics/debugging provides programers with multiple mechanisms to balance performance and safety.
     35These allocator versions are easy to use because they can be linked to an application without recompilation.
    3636
    3737Starting a micro-benchmark test-suite for comparing allocators, rather than relying on a suite of arbitrary programs, has been an interesting challenge.
    3838The current micro-benchmark allows some understand of allocator implementation properties without actually looking at the implementation.
    39 For example, dmalloc and ptmalloc3 were quickly identified has having course-grain locking in the Speed micro-benchmark.
     39For example, the memory micro-benchmark quickly identified how several of the allocators work at the global level.
    4040It was not possible to show how the micro-benchmarks adjustment knobs were used to tune to an interesting test point.
    4141Many graphs were created and discarded until a few were selected for the thesis.
     
    4545
    4646A careful walk-though of the allocator fastpath should yield additional optimizations for a slight performance gain.
     47In particular, looking at the implementation of rpmalloc, which is often the fastest allocator,
    4748
    4849The micro-benchmarks project requires more testing and analysis.
Note: See TracChangeset for help on using the changeset viewer.