Farm all your compressed plots from one central GPU!
Does this work with Flexfarmer?
Not yet it’s been less than 24 hours.
But planned? No rush you guys have been great!
Yes although it is low priority.
That is nice. Thank you for contributing, wish I had seen it two days ago.
What’s the ETA on Flexpool renting GPUs to Chia farmers ?
Using GPUs in the cloud would not be cost effective by a large margin. So someone has to build their own DC in the garage…
I’m thinking about providing a broker service, to broker compute between users and providers.
Yeah we’d have to charge a large % fee that would double after the halving, not to mention 24/7 monitoring and support, it’s very unlikely we’d have enough business to make it worthwhile.
What would be the main drawback ? I know some people who run remote harvesters to a DC farmer. How does the compressed plots situation change the cost ?
Renting GPUs in the cloud is very expensive, it would not make sense.
We are talking about compute as a service.
That was my idea too, not a generic solution but a “plot decompression” service tailored to Chia farming. Would you say it’s a viable idea ?
Sure there will be a pool in the future that will not need gpu in every farmer. This was my thought when plot compression first came. And this is the best regarding total power consumption of the blockchain. The pool will increase or decrease gpu power depending from it’s netspace. Plot decompression as stated produces very low net traffic . Also ASICS programmed to only decompress plots would be an exciting idea … If Chia gets its respect and price accordingly everything will happen . At this price very few people get involved either for farming or programming or chip design…
P.S. After 2 years involved with Chia farming , I feel like John Locke in Lost having to press the numbers every 6 hours in the crypt. Hope that it makes sense some time
This is something that nobody seems to be talking about. The next logical step is FPGA/ASICs. Curious to know what @madMAx43v3r thinks about it in the current conversation.
Farming compute seems to be memory bandwidth limited, so not sure if that makes sense.
In any case, Chia netspace is way too small to develop ASICs for. Need another 10x or 100x.