Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

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


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.


Ml (Using Pytorch or Tensorflow) still much much better on cuda from what I have heard.


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.


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.


Laptops this size cannot fit NVIDIA GPUs.


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.


Largely true, though...

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.


Mobile GPUs are much slower, they only share the same name


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.


Well, even if they could, battery life…




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: