Originally Posted by Atomizer
The only reason I could see a use for 128bit is more precision, not really as a speedup, and mainly for floating point stuff, which is now better done on the GPU anyway.
I mean, 63 bits can store values upto 9,223,372,036,854,775,807, thats 7 exabytes, or, 8191 petabytes, or, 8,388,607terabytes, etc, in terms of RAM/HDD space, I cant imagine anything for a long time needing that much, especially considering thats using only 63 bits, not the full 64bits, for 64bit unsigned values it would be pretty much double that amount.
As far as im aware, the largest single HDD is 2TB right now, lets say in 5 years we have 10TB drives, youd still need over 100 drives to make 1pb, and the only way youd actually need to address upto 1pb would be having them all in a raid, otherwise if they arent in a raid you only need to address 10tb, and then the drive partition that you're accessing.
So honestly the only practical use, right now, for 128bit, is higher floating point precision for scientific stuff