7
[Lemmy] Lemmy Release v0.19.20 - Reduced Memory Usage (join-lemmy.org)
submitted 3 months ago by [B] to c/fediverse@lemmy.zip
1 comment fedilink hide all child comments

1 comment

sorted by: hot top controversial new old
[–] [M] 2 points 3 months ago* (last edited 3 months ago)

The same news post got reposted under a new URL. Older post: https://lemmy.zip/post/68284209

  • permalink
  • fedilink
  • source
    •
    11 comments in c/announcements@lemmy.ml view that post
    [–] 28 points 3 months ago

    Since memory is so expensive this is a huge boon to those who host Lemmy!

  • permalink
  • fedilink
  • source
    •
    [–] 14 points 3 months ago

    oh my, good work all involved

  • permalink
  • fedilink
  • source
    •
    [–] 11 points 3 months ago

    Nice!

  • permalink
  • fedilink
  • source
    •
    [–] 10 points 3 months ago (1 child)

    This whole community is running in <8 GB of RAM? 🤯 love the work y'all do, keep it up!

  • permalink
  • fedilink
  • source
  • hideshow 1 child comment
    •
    • [–] 7 points 3 months ago* (last edited 3 months ago)

    Holy fucking shit those graphs* are insane.

  • permalink
  • fedilink
  • source
    •
    [–] 12 points 3 months ago

    Oh, nice! I don't run an instance, but efficiency upgrades are always cool and useful!

  • permalink
  • fedilink
  • source
    •
    [–] 5 points 3 months ago

    Noice

  • permalink
  • fedilink
  • source
    •
    [–] 1 point 3 months ago (2 children)

    Weird they were not using jemalloc to begin with. It's basically the default go to allocator for everything I do unless there is some specific use case that requires a different one.

  • permalink
  • fedilink
  • source
  • hideshow 2 child comments
    •
  • [–] 1 point 3 months ago (1 child)

    jemalloc is a like a speed limit, but for ram usage. i'm guessing that they're doing more active memory management now.

  • permalink
  • fedilink
  • source
  • parent
  • hideshow 1 child comment
    •
  • [–] 3 points 3 months ago

    As much as people around here rage at AI usage one thing it is very very good at is speeding up efficiency work like this. Building test harnesses and synthesizing the output from a number of profiling tools is made much much faster with the use of large language models.

  • permalink
  • fedilink
  • source
  • parent