Author: Ron Elliott 2014-07-04 18:39:07
Published on: 2014-07-04T18:39:07+00:00
In an email dated July 4th, 2014, kjj shared some general thoughts on the topic of algorithms and application-specific devices versus general-purpose computers. The author believes that there is no algorithm that cannot be performed better in an application-specific device than in a general-purpose computer. If there is, then it must necessarily perform best on one specific platform, which would make that platform the de facto application-specific device. The author also notes that IO-bound is the same as memory bound for devices that have enough memory. They believe that 20 GB is already trivial today, and you don't really get into “ask-the-wife-for-permission” money until you cross 128 GB. The exception would be if the IO was to an oracle outside of the device's control and artificially limited in throughput. The author concludes by saying that keeping the algorithm simple and ASIC-easy has the advantage of allowing just about anyone to sit down and design an ASIC for SHA, thus leading to diversity in the marketplace. A harder algorithm can still be made into an ASIC (or more generally into an ASD), but will require more skilled designers, more expensive fabrication, and concentrate the ASIC advantage into the hands of fewer people, which is contrary to the stated goals.
Updated on: 2023-06-09T00:42:01.457831+00:00