7 ms·

Wonder how good it is compared to the latest NVIDIA GPUs in laptops for gaming and ML Tasks

• lysecret 4y ago
  Ml (Using Pytorch or Tensorflow) still much much better on cuda from what I have heard.
  • mrtranscendence 4y ago
    Oh it's no contest. You may as well be comparing chihuahua to a mastiff. Still, it's better than nothing and useful to have around for development and small-scale experimentation.
    • rawoke083600 4y ago
      True... unless you factor in all those blackscreens nvidia has given you on linux(well me at least). Ive wasted days and weeks on stupid nvidia drivers.
• jpgvm 4y ago
  Probably still way slower than advertised. The M1's were otherwise awesome but the GPU performance was much over-stated (claims like faster than 3090 when barely on-par with 3050 which is 1/4th as fast). You have to remember that these are just scaled mobile GPUs. They can't hold a candle to proper dedicated GPU hardware and never will until Apple gets around to designing desktop specific GPU SoCs.
• wmf 4y ago
  Laptops this size cannot fit NVIDIA GPUs.
  • neogodless 4y ago
    Largely true, though... https://www.asus.com/us/site/gaming/rog/gaming-laptops/flow-x13.html https://www.asus.com/us/site/gaming/rog/gaming-laptops/flow-... It's a very small laptop with the option to utilize a current generation Nvidia GPU.
    • f311a 4y ago
      Mobile GPUs are much slower, they only share the same name
      • neogodless 4y ago
        I don't think you read the full details. The Asus Flow convertible is a tablet/laptop that supports a full desktop external GPU, in addition the discrete mobile GPU that is built into the device.
  • mrtranscendence 4y ago
    They can totally fit Nvidia GPUs, just not particularly high performance ones. It's not that hard to find a thin-and-light laptop with a low powered discrete Nvidia GPU. Heck, even the Switch technically has an Nvidia GPU.
  • spfzero 4y ago
    Well, even if they could, battery life…