Find factors for non base2 candidates

[pepi37]

Finding factors on non base2 candidates many users here found useless. But that is not my concern, I need to at least try to learn,
to understand, what program I can use for that purpose and how fast. For now: only Prime95 can do that.
For finding factor of this candidate I use next settings in Prime95
Pminus1=4,20,770610,1,80000,800000

Prime95 make stage 1 and stage 2 in total about 360 second and max memory usage was around 280 MB
Result is factor for candidate
4*20^770610+1 has a factor: 231221541636698021

Please can you give me what will be similar ( in time, memory consumption) settings in EMC-GMP for candidate 4*20^770610+1 (GPU is excluded from this story, since candidate is too big), or better to say is any other program that will be faster ( or at least have same speed as Prime95) and of course find factors for non base 2 candidates?

Thanks for any suggestion and link(s).

[pepi37]

Since nobody give me any direction or hint: this is small research result(s)

Sieve depth was 80T
Sequence was 4*53^n+1 where N=625000 - 650000 ( it is 254 candidates in test)

Since Prime95 has Pfactor ( as "optimal") I first run 254 tests with those settings and Prime95 choose B1=20000, B2=175000 - testing was very fast and I found 6 candidates

Then I rerun but with my setting B1=80000 and B2=800000 and found every factors as stage before , and 5 new factors ( but run time was 2.5 longer)

Then I remove all factors I found and run final stage with B1=400000 and B2=7M
I found additional 6 factors , but run time was 6x then first attempt.

So factors can be found, but in case that your candidates have less then 2M digits it is faster to run simple sieve process :)

Now back to LLR-ing :)

[pepi37]

This is taken from other message in this forum

Running P-1 is equivalent to running a single ECM curve. If it is known a priori that P-1 is divisible by (say) q, then one
simply reduces the size of the factor you are looking for by log(q) and then chooses the ECM parameters appropriately
for running just one curve.

However, it is clear that rather than run a SINGLE ECM curve to high limits (which is what P-1 to high limits accomplishes)
it is much better to run MANY ECM curves with lower limits.

Now I have to find way to use in "my" case: I run many ECM test, but finding factors rate was many times lower then P-1 on high B1 and B2.