Open adewinter opened 11 years ago
At the moment I am testing the water, making sure there is not something I am missing.
It really is a general idea, I am asking as many people as I know on how each part would best be implemented.
But with the core idea, miners who currently have the rig set up with GPUs could install this code. It would by default run generic hashing. Researchers and Medical Physics personnel (and many other people who would have application for Monte Carlo simulations) would submit to the network a piece of code that would be shared in a torrent fashion. They would attach to this code a declared payment amount per iteration. GPU miners could pick up this code, test to see if it is more worth while than their current work, and if so they would produce a result, encrypt in a public key fashion along with their bitcoin address. The payer would then receive the code and pay the address to which supplied the result.
Random numbers for the use within the Monte Carlo would be supplied by a trusted node. In that way, when an untrusted node runs a declared amount of iterations (say 10 000), 1% of these will be run with random numbers that are duplicated within nodes that are trusted. That way the rest of the data can be verified as truthful.
Have to go, if there are issues, please bring them up, and see if we can brainstorm solutions.
Once each part is addressed we can begin :) I will focus intially on the OpenCL implementation of the Monte Carlo (more the researcher side initially ) On 21/05/2013 2:31 AM, "Anton de Winter" notifications@github.com wrote:
I'd like to try get involved : )
— Reply to this email directly or view it on GitHubhttps://github.com/MeshachBlue/compushare/issues/1 .
Btw, this has become a lot more simple. And has been moved over to: https://github.com/MeshachBlue/DistributedGPU
I'd like to try get involved : )