I don’t fully understand the ins and outs of Bitcoin. What I do just about understand (correct me if I’m wrong!) is that the amount of processing it’s worth doing per day is a product of the price, and vice versa.
As Bitcoin mining is effectively an arms race to producing the most processing power per unit of electricity/money. To do this there is now custom hardware. Graphics cards and “normal” supercomputers have their theoretical power measured in FLOPS (there’s other better measures, I think FLOPS is the BMI of supercomputing). The important point here is that it stands for “Floating point operations per second”. A floating point number is basically the same as scientific notation. A number and then the number of times to shift the decimal point left or right. Bitcoin mining uses hardware optimised specifically for integer operations, rather than floating point.
Why does this matter?
Well, it seems reasonably likely that Bitcoin might crash and make the vast amount of bespoke hardware deployed to mine it suddenly unprofitable. The owners will have the choice of scrapping/mothballing it or to try and monetise it in other ways which would mean a sudden glut of availability of processing power for integer tasks (unlike the more common FLOPS). Maybe that’s something we should be preparing for? Could it have a significant impact on commodity data processing? Maybe there’s going to be a window where teams with tasks that can run on such hardware can get very good deals… until everyone else catches up.
I’m not experienced with supercomputing, so this is just a back-of-the-envelope theory.
Is this a realistic scenario? Have I overestimated the significance of floating point vs integer hardware?
0 Responses
Stay in touch with the conversation, subscribe to the RSS feed for comments on this post.