Samsung's Processing-in-Memory (PIM)

(chipsandcheese.com)

57 points | by ingve 2 hours ago

8 comments

  • consp 54 minutes ago
    So you basically dispose of cache for the memory region used? I wonder what the offsets of the cache misses is going to be in practice (the article addresses it but there is no solution/impact given by samsung).
    • manmal 27 minutes ago
      I guess you also get very high bandwidth that way? I‘m not sure that would come for free though.
    • PunchyHamster 3 minutes ago
      if your working set fits in cache PIM is irrelevant
  • reliabilityguy 1 hour ago
    Interesting that Samsung still pursues PIM. IIRC they had a paper in ISCA21 or 22 where they showed HBM2 module with PIM, which back then impressed me quite a lot.

    That being said, I am not sure what’s the killer application for this technology, and without such application adoption is unlikely.

    • xyzzy123 1 hour ago
      As I understand it, the killer app is llms. You could run MACs directly in RAM, offloading a lot of work from CPU and cutting down on insane (external) memory bandwidth required.

      Imagine (this is a fantasy pitch but potentially achievable for some use cases) wanting to run a larger llm and all you have to do is buy more RAM so it fits.

      • embedding-shape 40 minutes ago
        > Imagine (this is a fantasy pitch but potentially achievable for some use cases) wanting to run a larger llm and all you have to do is buy more RAM so it fits.

        Isn't this how it works today already? Granted you wanted to run it on RAM rather than VRAM.

    • netfortius 1 hour ago
      > That being said, I am not sure what's the killer application for this technology...

      Build it, and they will come ;)

  • userbinator 38 minutes ago
    In-memory computation was already possible with regular DRAM: https://news.ycombinator.com/item?id=22712811

    Add a new set of CPU instructions like “rep macb”

    ...and it's been long enough now, that I can say there was an effort to implement this on standard x86 memory controllers and have the existing string instructions do so, back in the days of SDR SDRAM, but the tradeoffs weren't (yet) in favour.

  • saejox 29 minutes ago
    If we could buy a 64gb stick and run a 32b model with 30tps on it. This could sell
  • Torkel 12 minutes ago
    [dead]
  • LogTrim 1 hour ago
    [flagged]
  • sehw 1 hour ago
    [dead]
  • johnnyApplePRNG 52 minutes ago
    [flagged]