View Single Post
Old 2015-01-03, 20:31   #2
Batalov's Avatar
Mar 2008

22×3×7×109 Posts

You probably have seen this, but just in case.

Somewhere else I think I noticed a comment that Magma does first 80 small primes before trying clever things (and so does your gfnsvCUDA for quadratic nonr). We can look up in Pari's code, too, but I vaguely remember hearing that it does the same... And between Magma and Pari, people surely care about efficiency so it seems like the tried and true way to approach the qnr.
Batalov is offline   Reply With Quote