ASRock Launches Legacy M.2 Graphics Card

ASRock Rack
(Image credit: ASRock Rack)
Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • JfromNucleon
    Ummmm, interesting but who actually use it? I mean, it's gotta be worse than Intel integrated graphics even on their non-tigerlake processors
    Reply
  • Giroro
    JfromNucleon said:
    Ummmm, interesting but who actually use it? I mean, it's gotta be worse than Intel integrated graphics even on their non-tigerlake processors
    It's not really intended for a "real" desktop computer using a typical x86 cpu. It looks like it could be used to add removable graphics for embedded systems that use ARM... or much more likely it's probably meant to add a GUI to a headless server.
    Reply
  • SpudmanWP
    This would be good for WorkStation/Server boxes where you don't often access them or need the pcie slots for other things like folding cards, mining cards, etc.
    Reply
  • mikewinddale
    SpudmanWP said:
    This would be good for WorkStation/Server boxes . . .

    But in such a case, why not use a USB graphics adapter? Just quickly searching on Amazon, I found a StarTech USB 3.0 - dual HDMI adapter that supports 4K @ 60 Hz, for $111.41. And I found an Ablewe USB 3.0 - single HDMI adapter that supports 1080P @ 60 Hz for $15.99.

    For a headless server, a USB graphics adapter is something a technician could carry in their toolbox and use as-needed. Or maybe the server room would have a single monitor with a USB graphics adapter, and the technician would cart the monitor+adapter over to whichever server needed servicing at that time. That seems easier and cheaper than installing an M.2 card in every single server.

    And for permanent installations, a USB graphics adapter is just fine as well, if all you need is 2D graphics.

    Maybe for some people, M.2 PCIe 2D VGA makes sense. But for me personally, if I had a computer that only needed occasional weak graphics, I'd just use a USB adapter.
    Reply
  • Trident1983
    Even better is bc for older games and apps on windows and even Andriod/ios/macOS made in the 90s and 80s here. proper MHz speed timings are an issue as i9 like fast but not that fast for playing sim Isle and SimTower and simcity 2000 and civ 2 MGE!
    Reply
  • hotaru.hino
    mikewinddale said:
    But in such a case, why not use a USB graphics adapter? Just quickly searching on Amazon, I found a StarTech USB 3.0 - dual HDMI adapter that supports 4K @ 60 Hz, for $111.41. And I found an Ablewe USB 3.0 - single HDMI adapter that supports 1080P @ 60 Hz for $15.99.

    For a headless server, a USB graphics adapter is something a technician could carry in their toolbox and use as-needed. Or maybe the server room would have a single monitor with a USB graphics adapter, and the technician would cart the monitor+adapter over to whichever server needed servicing at that time. That seems easier and cheaper than installing an M.2 card in every single server.

    And for permanent installations, a USB graphics adapter is just fine as well, if all you need is 2D graphics.

    Maybe for some people, M.2 PCIe 2D VGA makes sense. But for me personally, if I had a computer that only needed occasional weak graphics, I'd just use a USB adapter.
    According to StarTech themselves, their USB to whatever video adapters still require a GPU to drive the display output.
    Reply
  • escksu
    mikewinddale said:
    But in such a case, why not use a USB graphics adapter? Just quickly searching on Amazon, I found a StarTech USB 3.0 - dual HDMI adapter that supports 4K @ 60 Hz, for $111.41. And I found an Ablewe USB 3.0 - single HDMI adapter that supports 1080P @ 60 Hz for $15.99.

    For a headless server, a USB graphics adapter is something a technician could carry in their toolbox and use as-needed. Or maybe the server room would have a single monitor with a USB graphics adapter, and the technician would cart the monitor+adapter over to whichever server needed servicing at that time. That seems easier and cheaper than installing an M.2 card in every single server.

    And for permanent installations, a USB graphics adapter is just fine as well, if all you need is 2D graphics.

    Maybe for some people, M.2 PCIe 2D VGA makes sense. But for me personally, if I had a computer that only needed occasional weak graphics, I'd just use a USB adapter.

    Hmm.... I am not sure if you can boot from a USB graphics adapter. Server boards mostly have server management module by aspeed, thats why it can boot without any graphics card.

    But for those without, I don't think you can plug in the USB adapter and boot. The system will likely detect no GPU and throw error beep/code. Not all CPUs come with integrated GPU. So I guess this is where this module is needed.

    There are small form factors that does not come with any PCIE slots for GPU. And the size is also too small for any GPU to be installed. And then, the CPU does not come with integrated graphics (eg. Ryzen CPUs), then this is a good solution.

    Another scenario is that there is lack of available PCIE slots. (Eg. they are occupied by Tesla cards), then this comes handy.
    Reply
  • nofanneeded
    I was expecting at least Nvidia MX 330/450 12 watts variant on this . I am disappointed. it is easy to cool 12 watts on M2 card.
    Reply
  • Rdslw
    SpudmanWP said:
    This would be good for WorkStation/Server boxes where you don't often access them or need the pcie slots for other things like folding cards, mining cards, etc.
    I think IOT is the target audience.
    The Second option would be raspberry pi's class boards for stuff like your washing machine or vacuum cleaner or maybe even cars.
    If they manage to make this connector some kind of standard for random items that need minimal computing power, those things can save half of the cost while not having a gpu on board and sometimes you just need to connect to get a service message.
    imagine micro cpu, 1 core 1 pcie slot, 1-5W of computing.
    pull wifi card out, push gpu -> debug until no more errors on screen, put wifi back.
    Reply
  • hotaru.hino
    Rdslw said:
    I think IOT is the target audience.
    The Second option would be raspberry pi's class boards for stuff like your washing machine or vacuum cleaner or maybe even cars.
    If they manage to make this connector some kind of standard for random items that need minimal computing power, those things can save half of the cost while not having a gpu on board and sometimes you just need to connect to get a service message.
    imagine micro cpu, 1 core 1 pcie slot, 1-5W of computing.
    pull wifi card out, push gpu -> debug until no more errors on screen, put wifi back.
    I don't think it's IOT because there's no need for such devices to have a user-serviceable PCIe stack. And if the device doesn't have a need for a display, the software isn't going to be built to drive one. Any output otherwise is likely through some UART based debug port.
    Reply