Bro here's your size 12 shoe, sorry I mean your GPU.

Bro here's your size 12 shoe, sorry I mean your GPU.

  1. 3 weeks ago
    Anonymous

    This just proves Moore's law is dying as we are getting diminishing returns from shrinking nodes so we have to use more transistors and more power instead.

    • 3 weeks ago
      Anonymous

      Fine by me. 1000W PSU will survive until 2030.

      • 3 weeks ago
        Anonymous

        Only if they were made recently because even 1000w seasonic focus had problem with 3090 due to transient spike.
        They released new version which can survive those 2X spike.

        • 3 weeks ago
          Anonymous

          Bought it in early 2020.
          I think the 3090 having power delivery problems due to spikes says more about the GPU than the PSU in question. My Radeon VII consumes 350W under maximum load and I have never noticed any instabilities. It has an overkill power-delivery system though, which probably helps a lot.

          • 3 weeks ago
            Anonymous

            Both Nvidia and PSU makers blame each other but in my opinion it is the fault of PSU companies.
            Those transient spike are for anywhere from 1/10 to 1/100 of a second and PSU makers are using low quality Over power protection.

            • 3 weeks ago
              Anonymous

              The reason I blame Nvidia is because they've had power problems in the past because of companies wanting to cheap out on things like capacitors. The design of something like a Radeon VII

              Bought it in early 2020.
              I think the 3090 having power delivery problems due to spikes says more about the GPU than the PSU in question. My Radeon VII consumes 350W under maximum load and I have never noticed any instabilities. It has an overkill power-delivery system though, which probably helps a lot.

              is going to be superior to the design of a 3080 because the VRMs are in a ring around the SoC, although in the case of the Radeon VII this is really only possible because of the HBM2 memory. Nvidia could get better stability with higher quality components, a higher quality PCB, and better designing. It will be more expensive to produce the cards, but seeing as they're already thousands it really shouldn't even be a question.
              If I remember correctly, capacitors (high quality caps and a lot of them) are how you deal with massive power spikes. The instabilities of early 3080s was because of the capacitors used. This is absolutely within the power of Nvidia to fix.

              • 3 weeks ago
                Anonymous

                >The reason I blame Nvidia is because they've had power problems in the past because of companies wanting to cheap out on things like capacitors.
                That's more the damn AIBs fault
                The hardware was probably fine, it's just the AIBs overcooking the VBIOS because even on stock clocked cards thay can't leave very well alone.

              • 3 weeks ago
                Anonymous

                can't wait for "factory udnervolted, lowest on the market wattage" marketing next year
                untapped potential market niche, I think it's huge too

              • 3 weeks ago
                Anonymous

                I don't think we'll see that until the 5000 series hits at least if this trend of brute forcing the wattage continues

              • 3 weeks ago
                Anonymous

                Thats why i love the new "Quadro" cards like the a4000 and a6000. They are much more efficient than the RTX cards. But ofcourse Nvidia israelites us with their prices

          • 3 weeks ago
            Anonymous

            The reason I blame Nvidia is because they've had power problems in the past because of companies wanting to cheap out on things like capacitors. The design of something like a Radeon VII [...] is going to be superior to the design of a 3080 because the VRMs are in a ring around the SoC, although in the case of the Radeon VII this is really only possible because of the HBM2 memory. Nvidia could get better stability with higher quality components, a higher quality PCB, and better designing. It will be more expensive to produce the cards, but seeing as they're already thousands it really shouldn't even be a question.
            If I remember correctly, capacitors (high quality caps and a lot of them) are how you deal with massive power spikes. The instabilities of early 3080s was because of the capacitors used. This is absolutely within the power of Nvidia to fix.

            I bought an EVGA 1000 G6 PSU to power my RTX 3090 and never had a single issue. It's as if it was smart to buy my PSU and GPU witht he same brand because they actually test them together.

          • 3 weeks ago
            Anonymous

            >My Radeon VII consumes 350W under maximum load
            Power transients happen well under <1ms and cannot be measured by software lol
            But yeah, nvidia's 3080/90/ti has insane transients because they're so fucking leaky

    • 3 weeks ago
      Anonymous

      It’s been dead for a long time. They’ve reached the physical limit with known materials and lithography techniques. All process improvements are just tiny tweaks.

      • 3 weeks ago
        Anonymous

        Should've trade more humans to cave alien for tech

      • 3 weeks ago
        Anonymous

        Even AMD has 350W card and RDNA 3 will stretch it to 400W.
        Both companies are trying to make fastest GPU in the world so top end will consume insane amount of energy.

        >400 W GPU
        Please don't. You're destroying the environment. Is the planet really worth less to you than your world of warcraft or bitcoin mining?

        • 3 weeks ago
          Anonymous

          You individually, even your whole countries population literally doesn't make as much as a splash in the grand scheme of things even if your GPU used 1kW compared to companies never mind the military.

    • 3 weeks ago
      Anonymous

      moops*

    • 3 weeks ago
      Anonymous

      It’s been dead for a long time. They’ve reached the physical limit with known materials and lithography techniques. All process improvements are just tiny tweaks.

      Moore's law is about transistor count, not speed or density.

    • 3 weeks ago
      Anonymous

      It’s been dead for a long time. They’ve reached the physical limit with known materials and lithography techniques. All process improvements are just tiny tweaks.

      I guess we can finally start squeezing performance gains by optimising softwa- AHAHAHAHAHAHAHAHA

      • 3 weeks ago
        Anonymous

        >optimising softwa- AHAHAHAHAHAHAHAHA
        It's all so tiresome

      • 3 weeks ago
        Anonymous

        That's what I generally do for a living, good to see I'm about to become even more relevant.

      • 3 weeks ago
        Anonymous

        Even AMD has 350W card and RDNA 3 will stretch it to 400W.
        Both companies are trying to make fastest GPU in the world so top end will consume insane amount of energy.

      • 3 weeks ago
        Anonymous

        Let me join you in laughter. Software since 2012 has been working like garbage no matter how powerful your hardware is.

    • 3 weeks ago
      Anonymous

      no, that's just means maxwell was a fluke and nvidia is back in house burning business

    • 3 weeks ago
      Anonymous

      >Moore's law

      • 3 weeks ago
        Anonymous

        >Cucker Tarlson

  2. 3 weeks ago
    Anonymous

    That rounded design is going to look so outdated after just a short while.
    You know how you are always disgusted by your old hardware when you replace it, how it looks so outdated and teribly chink-shit cheap?
    Rounded designs multiply this by a hundred.
    They will look like a cheap plastic toy and you will feel absolutely retarded for spending a thousand dollars on it.

    Only rectangles are timeless.

  3. 3 weeks ago
    Anonymous

    peak chink design

  4. 3 weeks ago
    Anonymous

    >here is your 10 kilo gpu anon

  5. 3 weeks ago
    Anonymous

    There are rumor 4090 could do 24000 timespy extreme graphics score. Given that top chip at launch usually 50% faster than xx70 24000/1.5 = 16000. 4070 could be faster than 3090 ti cooling with ln2

  6. 3 weeks ago
    Anonymous

    >4-slot, 350mm monster card
    AMD sisters... it's over

    • 3 weeks ago
      Anonymous

      Big Bulging GPU vs virgin Pencil AMD card

  7. 3 weeks ago
    Anonymous

    >too poor to afford a yacht
    >too poor to afford a land-yacht
    >just rich enough to afford a gpu-yacht
    Finally I can flex my wealth on those damn kids...

    • 3 weeks ago
      Anonymous

      Top tier gpu was already pretty big flex the past 2 years.

  8. 3 weeks ago
    Anonymous

    Does it come with its own nuclear power plant?

    • 3 weeks ago
      Anonymous

      I'm more worried if it comes with it's own cooling tower at this point.
      I've modded my 3090 with heatsinks and two noctuas on the back and if I really push it I can still hit above 100°C on the RAM modules.

      • 3 weeks ago
        Anonymous

        Gddr6x can safely run till 110°C.
        I know it is too much but its not like you are running your GPU at 100% 24/7.
        That is why I always say don't buy miner GPU which has gddr6x, those things run really hot and miners ran it continuous for two years.

  9. 3 weeks ago
    Anonymous

    I'm okay with my 1080p display and 2070 GPU. Runs cool, rarely even toggles the fans. Runs everything I could ever want and more. I often wonder who buys these behemoths. Professionals doing graphics shit? I can't imagine a normal person ever needing this kind of performance

  10. 3 weeks ago
    Anonymous

    Caselets btfo

  11. 3 weeks ago
    Anonymous

    At first I thought each of those was a stack of two cards on top of each other

    good luck using anything else in any of your PCIe slots if you're using one of these fuckers.

  12. 3 weeks ago
    Anonymous

    It doubles as a hoverboard.

Your email address will not be published.