ASIC mining took over GPU mining in efficiency ages ago and wouldn’t make sense.
Efficiency wouldn’t matter if customers pay for the power while you pocket the Bitcoin.
When everyone else is using ASIC at terahashes/s and you are only doing mega or gigahashs per second the chances of your pool getting the next coin is still low.
Every video card that’s plugged into an AI data center starts losing money the second it’s powered on. It might actually be more profitable to mine crypto.
Profitable is a stretch. Arguably them running with no workload will mean the card lives longer overall compared to mining which has known to cause hardware burnout in GPUs.
I didn’t say profitable. I said more profitable. As in losing less money.
Well that’s not exactly true, most GPU cards in mining operations are voltage underclocked and run at a stable rate. It’s much harder on a card to run in a typical desktop scenario with constant voltage changes and power on / off cycles.
That being said, there is a difference between a large scale mining operation that are maximizing profits to jimmy bob wanna be techbro trying to start one up
(And honestly, fuck both of those)
https://www.pcworld.com/article/395149/should-you-buy-a-used-mining-gpu.html
Is Bitcoin still a thing?
I’m not up on ROI for non-asic mining anymore, but I’d be interested to know if this would be economical for these centres. When I used to mine ethereum on 27 Rx 580 8GB, I had to flash them with different bios to get them to be profitable.
Then the heat. I had them in a separate building on 60A panel and turning on the lights was a 50/50 that the breaker would trip.
Would running crypto on GPUs geared for AI be worth the extra power draw?
I haven’t done the math, but you are forgetting the massive tax breaks for “job creation” and subsidizing the energy cost across the population around you. That’s gotta help the profitability
Why a quarter
The Bitcoin network adjusts its difficulty based on the amount of computing power available so that the discover rate is relatively constant, around every ten minutes.
Adding more compute power without an increase in regular currency value per Bitcoin will just make it less likely to break even.
Not saying it couldn’t happen, but the odds of it making enough money at scale and without asic mining gear seems unlikely.
I don’t know about other cryptocurrencies though - they might have different designs that are more amenable to clandestine shenanigans.
