A New Peer to Peer GPU Computing Platform for Machine Learning, Big Data Analytics and 3D Design

Hello Hackernoon!

I along with my team recently built Q Blocks, a peer to peer GPU computing platform that lets you rent out idle GPU computing in a remote machine at 10X less costs compared to other cloud networks.

You can easily spin up a virtual GPU instance pre-equipped with AI frameworks such as Tensorflow, PyTorch, Caffe etc and ready to launch Jupyter notebooks for compute intensive workloads.

We are soon adding support for Big data analytics and 3D design rendering as well.

The purpose is to offer affordable GPU supercomputing using our peer to peer computing network to teams and individual scientists, designers and researchers.

We are inviting limited number of early adopters to test out our platform (Free 30 GPU Computing hours for Hackernoon users only).

Sign up for early access via this link: Qblocks hackernoon early access.

If you have any feedback for us - I’d love to know about it.

Cheers!
Gaurav

2 Likes

Hey @gaurav-vij, one of these days someone will nail the holy grail of distributing computing. And that person may have face some life threats, reminds me that scene in Upload:
Screen Shot 2020-06-26 at 10.50.25 AM
Initial impressions. I like the sass of calling AWS and the other incumbents “primitive cloud companies” lol (but their should be a space between “Up” and “to”).


The gaming use case if my favorite of your big three:

Can the CPU value of an unused Playstation exceed the cost of the electricity to run the machine? This increased usage would probably also have a negative effect on the life expectancy of the machine. How far along are you in step 2? Gamers are always looking for more ways to generate payments, so I think if the tech generates revenue, marketing it should be relatively easy and fun.

Thanks for giving hosting credit to the Hacker Noon community. What’s the simplest way to test it out? Could it handle this use case?

1 Like

Haha hope we are able to nail the holy grail of distributed computing without going to grave for it :smiley:

Tbh they are pretty much primitive in the kind of experience and infrastructure. Some of the GPUs and machines are like 3-4 generations old on these providers. But customers still have to pay a premium for that.

Totally. Playstations and Xbox type gaming consoles mostly have a system like APU instead of a dedicated GPU (D-GPU) and CPU. APUs are like a single die structure for accelerated performance. They are less powerful than D-GPUs but less power hungry at the same time. So they suffice for some of the computational workloads.

And based on the initial feedback we received from the tests that we performed on the gaming consoles, there is definitely economic viability in it. We will be bringing a scalable solution over the next few months for the gaming consoles too.

And you are right about gamers. We already have several PC gamers as compute providers on our network.

Thanks David for going through the website in-depth. Really appreciate that. You would be receiving an invite email with credits asap to access the platform.

Please feel free to use the credits and launch GPU instances. Would be happy to help in case there’s any hurdle.

I think that use case is around forming meaningful insights around the collected data from webpages. The GPU instances on our network can definitely come in handy for that kind of big data processing.

If there’s any developer here who’s working on that use case then we’d happy to explore how our resources could be used for building something like that.