• [object Object]@lemmy.ca
    81·
    2 days ago

    Is that the gpu wire that keeps catching on fire?

    I don’t think you should besmirch the good name of FireWire with such comparisons.

        • SeductiveTortoise@piefed.socialEnglish
          8·
          19 hours ago

          Yes, but if I remember correctly the issue is that they don’t make sure to limit the current per pin, but to let electricity decide which path to choose. So for some it might be 2A and for others 16A for example.

          • glibg10b@lemmy.zip
            3·
            18 hours ago

            It depends on the board. Some boards have balancing circuitry, some board can only detect overcurrent on individual wires, and some can only detect total overcurrent (through the shunt)

          • jjagaimo@sh.itjust.worksEnglish
            2·
            18 hours ago

            It is a combination of multiple issues. This is one of them

            Bussing the connection on the card side means no balancing of currents between pins, like you said. Older cards dont do it per pin but do divide it into several input rails which are individually monitored for current / voltage differences and will disable the card if it exceeds a certain threshold.

            Usually cards would use larger pins on the older 8 pin pcie power cables, which can handle more current and will make better contact in addition to being easier to seat fully without jamming and keep retained.

            Additionally they tend to use multiple connectors (e.g. 2x8 pin connectors = 8 pins for +12V), with lower power allowed per pin for more headroom. This means higher current capability and less potential for failures

            For example, a standard 9070xt with a 304W TDP may use 2x8 pin connectors. Larger pins, more pins and lower tdp, with current balancing/monitoring on board. It comes out to 3A per pin at full load when the pins are rated for something like 9A.

        • boonhet@sopuli.xyz
          3·
          1 day ago

          Your calculation is correct, but just in case anyone not versed in electric circuits is reading: It’s 47.92/6 not 47.92/12 because the other 6 pins are ground (the “return” path for the electricity, it’s a circuit after all).

    • Y|yukichigai@kbin.earth
      16·
      1 day ago

      It was great in design but too few peripherals made use of it. The lack of input devices (e.g. mice, joystick, keyboards) meant that for most people it was far less useful than USB.

      Still pretty cool that you could (and still can) use it for networking in a pinch, not to mention the decades worth of video devices that will happily dump their contents to DV video as long as you have a FireWire port. Finding a VCR with DV output is one of the best ways to digitize old VHS recordings if you can’t set up VHS-decode.

      • Zink@programming.dev
        7·
        17 hours ago

        I remember reading that one reason firewire was so great was that every device had a controller chip or something like that, whereas USB used the system processor for a lot of stuff.

        So the cheaper option won the mainstream market over the better option. Now where have I heard that before other than all of the times?

        • Bluewing@lemmy.world
          1·
          17 hours ago

          Eh, it’s less about which is the “better choice” and more about “how good does good enough need to be”. With a dash of the 80/20 rule.

          • Zink@programming.dev
            2·
            13 hours ago

            Oh I’m an engineer that works on physical products (embedded software) and you will get zero disagreement from me there, lol. That is the case even with expensive stuff that’s sold to other businesses rather than cheap consumer tech.

            And the way I meant “better” was more like “superior specs / technically better.” I mean, as far as “good enough” and “meeting user needs at lowest cost > exceeding needs at higher cost” you cannot argue with the success of USB. It has succeeded beyond the scope of its own name.