...
I was under the impression that the sieving part of the algorithm was taking up much more than 1% of the computing time. For example, currently my "Sieve/Test ratio" is at 75%/25%... or am I confusing things?
Redacted that statement, but still, even if you made that 25% sieving 1000 times faster, you would still need those other 75% of the time and wouldn't mine much faster.
It's 75% sieving, not 25%. If you make the sieve 10000 times faster you can eliminate 99.9999% of candidates that need testing by filtering more prime divisors. The purpose of the sieve is to avoid having to do the slow tests, so improving the sieve does actually make more sense.
Ah, sorry, need some sleep... But you'd still have dimishing returns for sieving, as filtering by 2x as many primes won't filter out 2x more candidates.