Issue with rho table when using -I in loop mode
Oliver Kruse reported this bug to me via MersenneForum.
Now to what I think is a bug. The command was:
$ ecm -v -I 1 -c 0 1
Output:
Code:
GMP-ECM 7.0.5 [configured with GMP 6.2.99, --enable-asm-redc] [ECM]
Tuned for x86_64/k8/params.h
Running on chia-debian
2^1277-1
[...]
Using B1=3, B2=3-36, polynomial x^1, sigma=0:12409209765371398839
dF=2, k=2, d=12, d2=1, i0=0
Can't compute success probabilities for B1 <> B2min
But in this case, B1 == B2min! This happens with lots of (or all?) input numbers, when using -I and multiple curves (here at curve 3).
A found it easier to strip out a bunch of unrelated messages with this command
echo "2^1277-1" | ecm -v -I 1 -c 6 1 | grep -v -e " [FGH] " -e "Computing " -e "special div" -e "for NTT"
Wildly speculating this could be happening because B1 is a double and while int(B1) == int(B2min) it's possible that B1 != B2min