runtime question

[yoyo]

Hello,

I measured nfs() runtime of C120 - C150 on a 4 core system.
How much longer will yafu need if I run the full factor() on it?
Does it take 20% more time (of course if yafu doesn't find factors before nfs phase), or is it more?

What are your assumptions?

yoyo

[bsquared]

Originally Posted by yoyo

Hello,

I measured nfs() runtime of C120 - C150 on a 4 core system.
How much longer will yafu need if I run the full factor() on it?
Does it take 20% more time (of course if yafu doesn't find factors before nfs phase), or is it more?

What are your assumptions?

yoyo

It depends on a number of things. If you are running with default yafu.ini then ecm is run to a t-level of 4/13 of the number of digits for GNFS, or (4/13)/1.2857 for SNFS. For example a C120 GNFS gets t36.9, which is a couple thousand curves at B1=1e6. A C120 SNFS gets t28.7, or a few hundred curves at B1=250k. The amount of time this takes depends on whether you also run ECM multithreaded and of course on the speed of the ECM binary you are using. You can tailor the amount of time spent using your own values for "pretest_ratio" with "plan=custom" in yafu.ini. The docfile has info on how to do this.

tl:dr
Generally between 15-30% more time.