HN Explorer

Show HN: Couldn't afford GPUs as a student. Built my own cloud GPU provider

2 points by Samin100 ↗ HN
Last semester while working on deep learning side projects (most notably a TensorFlow CharRNN chatbot trained on ~500k of my own text messages) I'd always come across the same problem: I couldn't afford GPUs as a broke college student.

After being fed up spending more money on cloud resources than I did on food, I decided to build my own cloud GPU provider with prices ~3-5x less than AWS, GCP, and Azure.

This was done by aggregated underutilized servers from different hosting providers across the world and developing a layer of abstraction over their heterogenous hardware, then letting people use it for far cheaper than AWS.

It's live at https://vectordash.com - you can create a Linux VM within seconds.

Any feedback or feature requests would be much appreciated and happy training!

3 comments

[ 3.6 ms ] story [ 22.3 ms ] thread
coolspot
NVidia hates this one simple trick.
anonfunction
You have ~500k personal text messages?
Samin100
Apple stores all of your iMessages in a nifty MySQL database hidden away deep in your applications folder. Since I got an iPhone ~4 years ago, it was about ~300 messages back and forth per day. (yes, I may have a texting problem!). I actually wanted to prank my friends by wrapping the model in a Twilio API, telling people I got a new phone number, and waiting to see how long it would take them to realize they weren't actually texting me.