Problem with resume in ecm 7 when extending B1...
I discovered that ecm v7 has problems finding factors when resuming ecm with a B1 greater than the B1 in the resume file.
Here is a simple test case:
echo 294940575713329370427113862790368174424655176416322150390973123306949586795213 | ecm -sigma 2215339928 -save test1.txt 1000 1
ecm -resume test1.txt 1000
Since the B1 in the -resume line is the same as in the -save line, this test finds the factor, in stage 2, in both ecm 6 and ecm 7.
The following test, with a smaller B1, will also find the factor in stage 2 in both ecm 6 and ecm 7:
ecm -resume test1.txt 300
However, ecm 7 will fail to find the factor in step 2 with the following test (where ecm 6 succeeds):
ecm -resume test1.txt 1001
And, we can see that resuming with larger B1 values produces different N and X values in the save/resume files in v7 as compared to v6.
echo 294940575713329370427113862790368174424655176416322150390973123306949586795213 | ecm -sigma 2215339928 -save test2.txt 200 1
ecm -resume test2.txt -save test3.txt 300
v6 will find the factor in step 2 of the above test.
v7 will fail to find the factor.
Looking at the resume files:
test2.txt: v6 and v7 have matching N and X values:
v6 has:
N=294940575713329370427113862790368174424655176416322150390973123306949586795213;
X=0x1306f5cd0e5ae4cfda5787416f153b78bbd20f9c91a03d0b86a574868cbd06df6;
v7 has:
N=294940575713329370427113862790368174424655176416322150390973123306949586795213;
X=0x1306f5cd0e5ae4cfda5787416f153b78bbd20f9c91a03d0b86a574868cbd06df6;
test3.txt: v6 and v7 have different N and X values:
v6 has:
N=56422599690332211832212571575408573119160503127419078473534295581;
X=0x64ae340c9e715ffe48f68bca4eeeb7706cd532d4e21f6209d8b28d;
v7 has:
N=294940575713329370427113862790368174424655176416322150390973123306949586795213;
X=0x1de521484a07bff19536b31836ecc4b58c78a11d8d6569b28b37028db429c391f;