You're trying to compare mining farms running 24/7 with people using a single card for a probably an hour a day on average. Without actual data on the number of users, this is meaningless.
Bitcoin mining compares with PC gaming at 85TWh/a vs 75TWh/a
You're not citing your source, but it's likely "Taming the energy use of gaming computers" which is a problem: it's from 2014, it's a rough estimate, assumes that you turn off your computer only for 8h sleep, includes all the other work done on a PC if it's at all used for gaming, is based on hardware before we got massive frequency scaling, double graphics (like optimus) and before everyone moved away from CRTs... and finally before a massive exodus to laptops. (their follow up report discusses moving away from CRTs as a way to limit power usage) It's really not a good source today, it was a disputable one at the time, and doesn't even match the question - it measures gaming computers usage, not gaming usage.