20150309, 07:01  #12  
"Curtis"
Feb 2005
Riverside, CA
4,219 Posts 
Quote:
At smaller sizes, those steps are so fast that optimizing for fastest sieve is generally best. Starting with Mr Womack's data above, 29LP can be used as small as 125 and be faster than any 28LP alternative. It was nearly 10% faster, so the 28/29 cutoff is likely even lower. If anyone is willing to try this experiment on a GNFS 123 or 124, run 13e twice with 28 and 29LP and report timings. 28M rels or fewer for 29 bit! I'm not going to run GNFS140s twice, but I believe the 29/30 transition is around 143144. I'm going to try my next GNFS141 with 13e/30 and see how that does. Perhaps 29/30 transition should happen before 13e/14e! Other tests posted in the forum recently suggest 14e/32 is superior for GNFS164 and up. I only have one sample point, but G156 with 31LP (133M raw rels) was almost as fast as G155 with 30 (80M raw rels). So, I think 30/31 cutoff is 154 or so. So, perhaps a rule of thumb: every 10 digits, go up a bit of LP *or* a siever. That might make 15e/32 correct up to ~174, 15e/33 to ~184, and by then filtering and matrix timeeffects need to be considered. Additional data appreciated! 

Thread Tools  
Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
QS/NFS crossover points  bsquared  Factoring  24  20160125 05:09 
SNFS poly for b^n1, n prime?  ryanp  Factoring  6  20130719 17:23 
Advice for large SNFS jobs?  ryanp  Factoring  69  20130430 00:28 
Large Prime Variation of QS  Sam Kennedy  Factoring  9  20121218 17:30 
32/33 and 15e/16e crossover point  fivemack  Factoring  7  20090421 07:59 