Dropping 41784 (129 digits, 2^5*3*5*7, ECMed C126)
Dropping 321408 (130 digits, 2^3*3*5, ECMed C128)
Dropping 510752 (128 digits, 2^2*7, ECMed C121)

Dropping 66276 (133 digits, 2^3*3, ECMed C123)
Dropping 395442 (140 digits, 2*3*29*ECMed C138)

Dropping 935592 (2^4*3^2, ECMed C121)
Dropping 321660 (2^2*7*11, ECMed C122)
Dropping 415152 (2^3*3*5, ECMed C122)

Reserving 323400 553872 943944 467700 207990 668412 278598 508626 135000

@fivemack: what level do you ECM to when you write ECMed Cxxx ? I'm running aliqeueit mith -m 4 (which should catch a C35 factor)

[QUOTE=ChristianB;400423]
@fivemack: what level do you ECM to when you write ECMed Cxxx ? I'm running aliqeueit mith -m 4 (which should catch a C35 factor)[/QUOTE]

It depends on the size of the cofactor; I run for 10% of the time that I estimate the GNFS will take (where this estimate is exp(0.113*number_of_digits-1.8) seconds). For a C112 that's about 2000 curves at B1=1e6, for a C122 it's about 5000, for a C135 it's about 20000.

[QUOTE=fivemack;400425]It depends on the size of the cofactor; I run for 10% of the time that I estimate the GNFS will take (where this estimate is exp(0.113*number_of_digits-1.8) seconds). For a C112 that's about 2000 curves at B1=1e6, for a C122 it's about 5000, for a C135 it's about 20000.[/QUOTE]

Thanks for the detailed explanation. It seems that the -m 4 option for aliqueit is doing roughly the same. I scanned through the logfile and found that prior to a gnfs run there where usually 524 curves run at B1=1e6 and that indeed the ecm took 10% of the time the gnfs run took after that (for a 100digit cofactor). I increased the ecm level a little bit and monitor the times.

Dropping 323400 (ECMed C129)
Dropping 553872 (ECMed C110)
Dropping 943944 (ECMed C113)
Dropping 467700 (ECMed C123)
Dropping 207990 (ECMed C126)
Dropping 668412 (ECMed C106)
Dropping 278598 (ECMed C109)
Dropping 508626 (ECMed C114)
Dropping 135000 (ECMed C112)

Reserving 452574 303840 467700 121944 242352 974072 574200 864336 676464 546516 296754 486822 496860 120834 721080

Dropping 50892 (127 digits, 2^2*7^2 * ECMed C125)
Dropping 337662 (line number 4933, 122 digits, 2^2*7^2, ECMed C115)
Dropping 981510 (129 digits, 2^2*3, ECMed C121)

337662 has merged with 30352, see detail elsewhere

Taking 30352

Dropping 452574 (size: 124, ECMed C123 cofactor)
Dropping 303840 (size: 128, ECMed C114 cofactor)
Dropping 467700 (size: 132, ECMed C123 cofactor)
Dropping 121944 (size: 122, ECMed C107 cofactor)
Dropping 242352 (size: 121, ECMed C107 cofactor)
Dropping 974072 (size: 123, ECMed C111 cofactor)
Dropping 574200 (size: 121, ECMed C116 cofactor)
Dropping 864336 (size: 121, ECMed C121 cofactor)
Dropping 676464 (size: 121, ECMed C120 cofactor)
Dropping 546516 (size: 123, ECMed C118 cofactor)
Dropping 296754 (size: 121, ECMed C119 cofactor)
Dropping 486822 (size: 125, ECMed C116 cofactor)
Dropping 496860 (size: 122, ECMed C108 cofactor)
Dropping 120834 (size: 123, ECMed C107 cofactor)
Dropping 721080 (size: 124, ECMed C119 cofactor)

Reserving 982840 812562 222834 422382 682704 312150 452880 912192 342504 634098 984654 697872 328592 742350 652590 663120

@ChristianB A lot of those smaller cofactors (< 120 digits) are relatively straightforward, and I'd recommend that you try to keep going on these sequences until they reach a cofactor of at least 120 digits.

Releasing 283410 (134 digits, ECMed C131); 506400 (129 digits, ECMed C123); 729480 (131 digits, ECMed C125)

To piggyback on the discussion above on ECM depth, I use yafu for factoring and rely on its internal algorithm to determine when ECM is done.