PoX is a new class of ultra‑fast, ultra‑green memories that meet the swelling appetite of large‑language‑model accelerators.
  • CouncilOfFriends@slrpnk.netEnglish
    209·
    14 days ago

    By tuning the “Gaussian length” of the channel, the team achieved two‑dimensional super‑injection, which is an effectively limitless charge surge into the storage layer that bypasses the classical injection bottleneck.

  • boonhet@lemm.eeEnglish
    89·
    14 days ago

    AI AI AI AI

    Yawn

    Wake me up if they figure out how to make this cheap enough to put in a normal person’s server.

    • Zip2@feddit.ukEnglish
      134·
      14 days ago

      normal person’s server.

      I’m pretty sure I speak for the majority of normal people, but we don’t have servers.

      • Rose@slrpnk.netEnglish
        45·
        14 days ago

        Yeah, when you’re a technology enthusiast, it’s easy to forget that your average user doesn’t have a home server - perhaps they just have a NAS or two.

        (Kidding aside, I wish more people had NAS boxes. It’s pretty disheartening to help someone find old media and they show a giant box of USB sticks and hard drives. In a good day. I do have a USB floppy drive and a DVD drive just in case.)

        • KnightontheSun@lemmy.worldEnglish
          15·
          14 days ago

          Hello fellow home labber! I have a home built xpenology box, proxmox server with a dozen vm’s, a hackentosh, and a workstation with 44 cores running linux. Oh, and a usb floppy drive. We are out here.

          I also like long walks in Oblivion.

          • MrPistachios@lemmy.todayEnglish
            7·
            14 days ago

            Man oblivion walks are the best until a crazy woman comes at you trying to steal your soul with a fancy sword

        • pivot_root@lemmy.worldEnglish
          4·
          14 days ago

          It’s pretty disheartening to help someone find old media and they show a giant box of USB sticks and hard drives.

          Equally disheartening is knowing that both of those have a shelf-life. Old USB flash drives are more durable than the TLC/QLC cells we use today, but 15 years sitting unpowered in a box doesn’t have very good prospects.

      • notabot@lemm.eeEnglish
        18·
        14 days ago

        You… you don’t? Surely there’s some mistake, have you checked down the back of your cupboard? Sometimes they fall down there. Where else do you keep your internet?

        Appologies, I’m tired and that made more sense in my head.

      • fmstrat@lemmy.nowsci.comEnglish
        4·
        13 days ago

        “Normal person” is a modifier of server. It does not state any expectation of every normal person having a server. Instead, it sets expectation that they are talking about servers owned by normal people. I have a server. I am norm… crap.

    • gravitas_deficiency@sh.itjust.worksEnglish
      7·
      14 days ago

      You can get a Coral TPU for 40 bucks or so.

      You can get an AMD APU with a NN-inference-optimized tile for under 200.

      Training can be done with any relatively modern GPU, with varying efficiency and capacity depending on how much you want to spend.

      What price point are you trying to hit?

      • boonhet@lemm.eeEnglish
        8·
        14 days ago

        What price point are you trying to hit?

        With regards to AI?. None tbh.

        With this super fast storage I have other cool ideas but I don’t think I can get enough bandwidth to saturate it.

        • barsoap@lemm.eeEnglish
          1·
          14 days ago

          With regards to AI?. None tbh.

          TBH, that might be enough. Stuff like SDXL runs on 4G cards (the trick is using ComfyUI, like 5-10s/it), smaller LLMs reportedly too (haven’t tried, not interested). And the reason I’m eyeing a 9070 XT isn’t AI it’s finally upgrading my GPU, still would be a massive fucking boost for AI workloads.

        • gravitas_deficiency@sh.itjust.worksEnglish
          1·
          14 days ago

          You’re willing to pay $none to have hardware ML support for local training and inference?

          Well, I’ll just say that you’re gonna get what you pay for.

          • bassomitron@lemmy.worldEnglish
            9·
            14 days ago

            No, I think they’re saying they’re not interested in ML/AI. They want this super fast memory available for regular servers for other use cases.

                • boonhet@lemm.eeEnglish
                  3·
                  13 days ago

                  I mean the image generators can be cool and LLMs are great for bouncing ideas off them at 4 AM when everyone else is sleeping. But I can’t imagine paying for AI, don’t want it integrated into most products, or put a lot of effort into hosting a low parameter model that performs way worse than ChatGPT without a paid plan. So you’re exactly right, it’s not being sold to me in a way that I would want to pay for it, or invest in hardware resources to host better models.

      • WorldsDumbestMan@lemmy.todayEnglish
        1·
        14 days ago

        I just use pre-made AI’s and write some detailed instructions for them, and then watch them churn out basic documents over hours…I need a better Laptop

  • conditional_soup@lemm.eeEnglish
    67·
    14 days ago

    This article appeared in my feed just above another article about how China has the world’s first operational thorium reactor. Meanwhile, the US is about to fight a civil war over whether vaccination causes measles and stripping away the last of our social programs in order to get our wealthiest people another 2% subsidy.

    • pycorax@lemmy.worldEnglish
      10·
      14 days ago

      Probably because is an ethnicity and nationality. There are ethnic Chinese people all over the world and a few countries and regions are made of a majority of ethnic Chinese but are not related to China. Calling them the same thing is playing into the PRC’s “all ethnic Chinese pledge their allegiance to China” nonsense.

      • Netux@lemmy.worldEnglish
        6·
        14 days ago

        Isreal like that game of pretend. They believe anti zionist Jews are traitors.

        • pycorax@lemmy.worldEnglish
          3·
          13 days ago

          Perhaps but I haven’t encountered that myself. I’m ethnic Chinese that’s a citizen of another ethnic Chinese majority nation so I’ve encountered this specific type a lot more.

        • pycorax@lemmy.worldEnglish
          6·
          14 days ago

          The reverse, however, isn’t true. It may be somewhat understandable but not entirely reasonable to assume someone who is Chinese is from China which is what I’m trying to say.

    • liquidparasyte@pawb.socialEnglish
      9·
      14 days ago

      Real talk, why is discussion around people and subjects in China so fucking weird?

      If it’s not referring to the entire population when it only applies to the government or a subset of them as a global “the Chinese” or doing silly shit like “China scientists” everyone’s grammatical skills suddenly tank when even broaching a topic even tangential to the PRC.

    • DasSkelett@discuss.tchncs.deEnglish
      8·
      14 days ago

      Seriously, for me a “China scientist” is someone doing research on China, like a space scientist would do research on astronomy and similar. But I’m not a native English speaker, so, idk

      • Jesus@lemmy.worldEnglish
        1·
        14 days ago

        Someone doing research on China is a chiologist.

        Same as someone doing research on biology is a biologist.

        • hazelEnglish
          2·
          14 days ago

          Biology -> biolog -> biologist

          China -> chin -> chinist?

  • AItoothbrush@lemmy.zipEnglish
    36·
    14 days ago

    Brother, have you heard of buses? Even INSIDE cpus/socs bus speeds are a limitation. Also i fucking hate how the first thing people mention now is how ai could benefit from a jump in computing power.

    Edit: I havent dabbled that much in high speed stuff yet but isnt the picosecond range so fast that the capacitance of simple traces and connectors between chips influence the rising and falling edge of chips?

    • xthexder@l.sw0.comEnglish
      8·
      14 days ago

      That’s pretty much my understanding. Most of the advancements happened in memory speeds are related to the physical proximity of the memory and more efficient transmission/decoding.

      GDDR7 chips for example are packed as close as physically possible to the GPU die, and have insane read speeds of 28 Gbps/pin (and a 5090 has a 512-bit bus). Most of the limitation is the connection between GPU and RAM, so speeding up the chips internally 1000x won’t have a noticeable impact without also improving the memory bus.

  • 800XL@lemmy.worldEnglish
    19·
    14 days ago

    You just fucking wait. Trump is bringing manufacturing to the US. And when that plant opens someday you’ll be so sorry you doubted.

    • BobSentMe@lemm.eeEnglish
      8·
      13 days ago

      I’m sure the foxconn plant in Wisconsin will fire up ANY DAY NOW! drums fingers

      • 800XL@lemmy.worldEnglish
        6·
        13 days ago

        I talked to like 50 people today and all of the people said they were starting manufacturing plants tomorrow and they’ll be fully functional Tuesday around 3:15.

        I started mine earlier and I’ve already done manufacturing 3 times today. It’s really easy. By this time tomorrow I’ll have a couple more and they’ll all be winning manufacturing.

        Tariffs gave me the ability to finally believe in myself. Tariffs have increased my stamina in bed, given me a full head of hair again, and since I started manufacturing plant yesterday I’ve dropped 50 pounds.

  • WorldsDumbestMan@lemmy.todayEnglish
    18·
    14 days ago

    Whenever they say X whatever times, I doubt it right away, because they always interpret the statistics in the dumbest ways possible. You have a solar panel that is 28% efficient. There is no way it can be 20x times as efficient, that’s just clickbait.

  • tetris11@lemmy.mlEnglish
    11·
    14 days ago

    Wow, finally graphene has been cracked. Exciting times for portable low-energy computing

    • CosmoNova@lemmy.worldEnglish
      25·
      14 days ago

      It‘s likely BS anyway. Maybe it’s just me but reading about another crazy breakthrough from China every single day during this trade war smells fishy. Because I‘ve seen the exact same propaganda strategy during the pandemic when relations between China and the rest of the world weren‘t exactly the best. A lot of those headlines coming from there are just claims about flashy topics with very little substance or second guessing. And the papers releasing the stories aren‘t exactly the most renowned either.

      • LadyAutumnEnglish
        6·
        14 days ago

        It’s definitely possible they’re amplifying these developments to maintain confidence in the Chinese market, but I doubt they’re outright lying about the discoveries. I think it’s also likely that some of what they’ve been talking about has been in development for a while and that China is choosing now to make big reveals about them.