Machine Learning service

Hi,

I dabble with many tech projects as hobbies, and one thing I’ve been playing around with is renting out my computing power, emphasis on GPU, for machine learning, deep learning, AI etc. One thing all services have in common, is that the experience is terrible for the providers.

The STORJ community and way of working is amazing, and I would like to suggest that the higher-ups of STORJ have a cup of coffee and take 10 minutes on Slack to discuss the possibility of venturing into this field in the future.

Thanks,
alfananuq

1 Like

There used to be a thing called Seti at Home or something like that… It was approximately what you are describing, but without the micropayment system. It was one of the first distributed computing projects I ran on my old PCs.


It looks like the project shutdown this past March:

https://www.seti.org/setihome-going-hibernation

But you can still download the source code:

https://setiathome.berkeley.edu/sah_source_code.php

If I remember, the project mostly computed Fourier Transforms looking for anomalous signals in radio telescope data.

@alfananuq there is actually the project Placeholders that has such goals, check this
https://www.placeh.io/

i think fourier transform can be done by analog optronic coprocessors today, which makes any attempt at distributed computation on such tasks a waste of time, as it’s like 1000 times faster… or something like that… still very limited market for that tho…

and it’s analog… tsk tsk

anyways it would explain why seti@home was shutdown if it didn’t make sense to cloud compute their data.

and i use the term co processor loosely… its a pcie card that has some laser stuff in it and then uses basically lenses, mirrors, filters and what not…

then it simply uses a digital sensor to take an image of the result which then is returned to digital form.
i forget the name of the company that made the cards tho… was like 3-4 + years since read about it