3 ms·
I've been following this company for a while now, and others like it, and I'm really wondering how the economics of these rent-a-big-GPU services shake out. The
by nyx 5y ago
I've been following this company for a while now, and others like it, and I'm really wondering how the economics of these rent-a-big-GPU services shake out. The specs say you get a GeForce GTX 1080 (or equivalent, so maybe they're licensing NVIDIA vGPUs or something)... do they actually have one big GPU per subscription, or do I just have access to a big GPU whenever I want? You can't slice up a GPU as effectively as you can slice up CPU cores... Filling up a datacenter with big GPUs seems like a very expensive proposition.
- riekus 5y agoWonder if you could leverage Bitcoin farms for this.
- vgeek 5y agoI wondered about using it for ML type stuff, but figured they have application or time limit restrictions.
- kotaKat 5y agoWhen I originally used it at the beginning, if I recall, it was essentially a QEMU/KVM server with, indeed, a physical GPU passed to the VM, and each server had 2-4 GPUs assigned to it. When you logged in, you had to wait for one of the available servers to have an available GPU, in which point your virtual machine was mounted to that server and booted up, and their custom software on the VM would send the graphics back to you.
- c0balt 5y agoFYI, Linus Tech Tips showcased one of their servers a year ago: https://www.youtube.com/watch?v=DS1FcT93Qy8 https://www.youtube.com/watch?v=DS1FcT93Qy8 Tl;dr 2-3 NVIDIA Quadros split up over hypervisor