Ask HN: Potential for pay-as-you-go serverless scientific computing?
Scientific computing are conducted by researchers and engineers. A headache of scientific computing is each full scale iteration before the researcher can view the result takes long CPU hours, making testing and debugging only feasible in smaller surrogate scales.
Thus, more and more researchers are using serverless clouds to dynamically summon a high surge of computational power and release them when the full-scale iteration finish running.
For computation tasks without sensitive data or code, such as protein folding and public domain data crunching, there is BOINC and Gridcoin to facilitate distributed scientific computing and allow anyone to donate or sell computation power for Gridcoin, but BOINC's programming interface is not as easy to use as Amazon and Google's. In addition, BOINC requires manual approval of projects, because BOINC agent is not containerized/virtualized, computing code could present malware threat to host.
I wish to create an alternative to BOINC/Gridcoin, using container/VM images as computation tasks, and support unmonitored pay-as-you-go serverless computing.
I want to know if there is a large enough market for it.
Think this way, the world mines 6 bitcoins/day (US$42K/day, or US$15.3 million/year). Do all "embarrassingly parallel" scientific computing projects combined have comparable annual revenue with BTC or a lesser cryptocurrency such as ETH or DOGE?