A little bit deceiving the information. It is 16.6% loss to an external monitor and less than 10% for the 4070 TI Super. The almost a quarter is performance loss getting back to the laptop monitor. Which honest, if you are looking to have better GPU in your laptop it is better to get a dGPU for it.
Sadly (or almost comically) everyone keeps re-posting what anyone interested in a eGPU setup would already know:
- the limited bandwidth would cripple top-of-the-line GPU's in graphics applications - not using an external monitor eats up bandwidth - using the external enclosure for other connections eats up bandwidth
Systems that don't have TB, USB4 or Oculink ports but DO have a M.2 port can sometimes be fitted with a DIY TB or Oculink port.
Considering how useless or dismal many iGPU's are the idea of adding even a LOW to Medium end Desktop GPU card in a eGPU setup can be much more cost effective then buying into a 'gaming laptop' if gaming in the wild is not a use requirement.
Besides cost, HEAT is another consideration where an eGPU offloads the considerable power requirements and heat generated by the GPU to the external enclosure.
How about test it on some AMD cards and not cards that cost more than most people's entire build. I have a few mini pc's with oculink, and am building a homebrew egpu. Would love to see real numbers for a 7900XT/X or a 7800XT.
It turns out that, despite OCulink's higher bandwidth, it still chokes high-end GPUs like the Nvidia GeForce RTX 4090 by up to 23% when used in an external GPU enclosure, but Thunderbolt's limitations are far more obvious. As with the older standard, the eGPU setup suffered more when the graphics output was sent back to the laptop.
Deutsch
English
Español
Français
Italiano
Nederlands
Polski
Português
Русский
Türkçe
Svenska
Chinese
Magyar