Re:Hetzner appears to impose unlisted “fair use” on traffic
by NavinF on Tuesday July 26, 2022 @05:53PM
(#32244572)
> AWS makes a ton of sense if […] you need GPU power to train your AI model,
Breakeven time between paying for GPUs, servers, and colo vs GCP was 3 months when I last checked. AWS was even worse. For training, cloud GPUs only make sense if you don’t have anyone on your team that’s familiar with physical infra or if you have a ton of free credits which is fairly common for startups with VC funding.
Breakeven time between paying for GPUs, servers, and colo vs GCP was 3 months when I last checked. AWS was even worse. For training, cloud GPUs only make sense if you don’t have anyone on your team that’s familiar with physical infra or if you have a ton of free credits which is fairly common for startups with VC funding.