Hi,
Here are some suggestions:
1) Come up with an app that basically mirrors/simulates what performance any potential contractors should achieve with your rendering service. The goal for you is to check the particulars of the "horsepower" of the candidate farm under test. There are several configuration factors that may or may not be important such as employment of:
a) SSD's
b) high end CPU's
c) internet connection
d) MB idiosyncrasies
As a crude, at hand mining example using whattomine, 41 1080ti GPUS, have the equivalent ETH hashing power of about 100 1060 cards. Therefore GPU count can't be the sole criteria.
To try to avoid gaming the system you may want to spec also cclock, TDP, and Mclock of the FUT.
Its a latency issue, not a computing issue. 100 gtx 1060s on the same network will greatly outperform 41 1080tis spread across the world.
Ah, good hint. I think you are right. Also the work has to be splitted, synced, updated etc. Theres a lot of overhead.
will there be a demand for such services ? I know people in the freelance market who do rendering, but they often say that there are very few orders.
This service would be at a high scale and maybe a lower price as in the general rendering market. So, I would say yes.