[QUOTE=smh;190641]I searched for a poly over night. This is the best I've got[CODE]n: 233515345448704331188831767635577487430288807946401538680318081261615453927577744159848068718415496556100755527513286988731533493638561
# Murphy_E 3.76e-011
[/CODE][/QUOTE]
I'm sure I saw a 3.9 in my msieve run somewhere (I'll post it once the run completes).

[QUOTE=axn;190645]I'm sure I saw a 3.9 in my msieve run somewhere (I'll post it once the run completes).[/QUOTE]

The run is still going, but I have a 4.1
[CODE]# norm 5.128167e-13 alpha -7.282260 e 4.058e-11
skew: 1254739.13
c0: -395037057008070589224358816043760
c1: -28535510064887720543782442688
c2: -135722229269228350074457
c3: 34430682441748408
c4: 27658482452
c5: 7920
Y0: -124141415960610734878798301
Y1: 554190238882283

# norm 5.160900e-13 alpha -7.945421 e 4.123e-11
skew: 1961621.62
c0: -30221871906743529330007758282799725
c1: -46542571898730018876164434500
c2: -132986730345137014294411
c3: 37295435382966856
c4: 28665550052
c5: 7920
Y0: -124141401866998769863459328
Y1: 554190238882283

# norm 5.185762e-13 alpha -7.395020 e 4.020e-11
skew: 1676500.91
c0: 84656145541206932058167824889938527
c1: -110557669146445107172084358976
c2: -85004987739719311592251
c3: 76609624029985384
c4: 40001683652
c5: 7920
Y0: -124141243221175845987834050
Y1: 554190238882283

# norm 4.940195e-13 alpha -7.019987 e 4.006e-11
skew: 948264.70
c0: -3445305906913627771932546686653764
c1: -11136793885810423915288480818
c2: 78017854131225697445836
c3: 52382648081793707
c4: -53561324900
c5: 14700
Y0: -109698626343103870506786481
Y1: 544600877303671
[/CODE]

That 4.1 was the best.

[QUOTE=axn;190757]That 4.1 was the best.[/QUOTE]

I did some tests:

[code]rlim: 5000000
alim: 5000000
lpbr: 27
lpba: 27
mfbr: 54
mfba: 54
rlambda: 2.6
alambda: 2.6
[/code]
[code]$ for i in `seq 1000000 1000000 16000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
Warning: lowering FB_bound to 999999.
total yield: 510, q=1000507 (0.03010 sec/rel)
Warning: lowering FB_bound to 1999999.
total yield: 508, q=2000503 (0.02593 sec/rel)
Warning: lowering FB_bound to 2999999.
total yield: 605, q=3000523 (0.02451 sec/rel)
Warning: lowering FB_bound to 3999999.
total yield: 590, q=4000511 (0.02654 sec/rel)
total yield: 637, q=5000503 (0.02648 sec/rel)
total yield: 928, q=6000503 (0.02813 sec/rel)
total yield: 608, q=7000517 (0.03113 sec/rel)
total yield: 403, q=8000507 (0.03471 sec/rel)
total yield: 425, q=9000517 (0.03301 sec/rel)
total yield: 473, q=10000511 (0.03154 sec/rel)
total yield: 310, q=11000537 (0.03761 sec/rel)
total yield: 294, q=12000509 (0.03813 sec/rel)
total yield: 492, q=13000511 (0.03577 sec/rel)
total yield: 409, q=14000507 (0.03900 sec/rel)
total yield: 388, q=15000509 (0.03570 sec/rel)
total yield: 261, q=16000507 (0.04475 sec/rel)
$ for i in `seq 17000000 1000000 18000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
total yield: 226, q=17000507 (0.03721 sec/rel)
total yield: 216, q=18000527 (0.05153 sec/rel)
$ for i in `seq 19000000 1000000 20000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
total yield: 333, q=19000511 (0.03808 sec/rel)
total yield: 326, q=20000503 (0.04362 sec/rel)
[/code]

[code]rlim: 7500000
alim: 7500000
lpbr: 27
lpba: 27
mfbr: 54
mfba: 54
rlambda: 2.6
alambda: 2.6
[/code]
[code]$ for i in `seq 1000000 1000000 16000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
Warning: lowering FB_bound to 999999.
total yield: 524, q=1000507 (0.03164 sec/rel)
Warning: lowering FB_bound to 1999999.
total yield: 524, q=2000503 (0.02697 sec/rel)
Warning: lowering FB_bound to 2999999.
total yield: 624, q=3000523 (0.02535 sec/rel)
Warning: lowering FB_bound to 3999999.
total yield: 609, q=4000511 (0.02764 sec/rel)
Warning: lowering FB_bound to 4999999.
total yield: 662, q=5000503 (0.02715 sec/rel)
Warning: lowering FB_bound to 5999999.
total yield: 1025, q=6000503 (0.02820 sec/rel)
Warning: lowering FB_bound to 6999999.
total yield: 711, q=7000517 (0.03097 sec/rel)
total yield: 476, q=8000507 (0.03447 sec/rel)
total yield: 523, q=9000517 (0.03126 sec/rel)
total yield: 560, q=10000511 (0.03193 sec/rel)
total yield: 372, q=11000537 (0.03691 sec/rel)
total yield: 351, q=12000509 (0.03729 sec/rel)
total yield: 611, q=13000511 (0.03426 sec/rel)
total yield: 487, q=14000507 (0.03889 sec/rel)
total yield: 456, q=15000509 (0.03616 sec/rel)
total yield: 309, q=16000507 (0.04482 sec/rel)
[/code]

[code]
rlim: 10000000
alim: 10000000
lpbr: 27
lpba: 27
mfbr: 54
mfba: 54
rlambda: 2.6
alambda: 2.6
[/code]
[code]$ for i in `seq 1000000 1000000 16000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
Warning: lowering FB_bound to 999999.
total yield: 533, q=1000507 (0.03415 sec/rel)
Warning: lowering FB_bound to 1999999.
total yield: 542, q=2000503 (0.02851 sec/rel)
Warning: lowering FB_bound to 2999999.
total yield: 636, q=3000523 (0.02714 sec/rel)
Warning: lowering FB_bound to 3999999.
total yield: 626, q=4000511 (0.02919 sec/rel)
Warning: lowering FB_bound to 4999999.
total yield: 675, q=5000503 (0.02890 sec/rel)
Warning: lowering FB_bound to 5999999.
total yield: 1039, q=6000503 (0.03018 sec/rel)
Warning: lowering FB_bound to 6999999.
total yield: 728, q=7000517 (0.03238 sec/rel)
Warning: lowering FB_bound to 7999999.
total yield: 499, q=8000507 (0.03579 sec/rel)
Warning: lowering FB_bound to 8999999.
total yield: 571, q=9000517 (0.03236 sec/rel)
total yield: 623, q=10000511 (0.03287 sec/rel)
total yield: 418, q=11000537 (0.03801 sec/rel)
total yield: 389, q=12000509 (0.03933 sec/rel)
total yield: 683, q=13000511 (0.03512 sec/rel)
total yield: 554, q=14000507 (0.03919 sec/rel)
total yield: 516, q=15000509 (0.03667 sec/rel)
total yield: 357, q=16000507 (0.04457 sec/rel)
[/code]

Appears that alim & rlim = 7.5M gets the needed 17M relations (for 27-bit) in about 3% less time than either 5M or 10M.

28-bit:
[code]
rlim: 7500000
alim: 7500000
lpbr: 28
lpba: 28
mfbr: 56
mfba: 56
rlambda: 2.6
alambda: 2.6
[/code]
[code]$ for i in `seq 1000000 1000000 16000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I13e -a test.poly -f $i -c 500; done
Warning: lowering FB_bound to 999999.
total yield: 867, q=1000507 (0.01938 sec/rel)
Warning: lowering FB_bound to 1999999.
total yield: 960, q=2000503 (0.01506 sec/rel)
Warning: lowering FB_bound to 2999999.
total yield: 1189, q=3000523 (0.01372 sec/rel)
Warning: lowering FB_bound to 3999999.
total yield: 1175, q=4000511 (0.01477 sec/rel)
Warning: lowering FB_bound to 4999999.
total yield: 1288, q=5000503 (0.01436 sec/rel)
Warning: lowering FB_bound to 5999999.
total yield: 1971, q=6000503 (0.01518 sec/rel)
Warning: lowering FB_bound to 6999999.
total yield: 1328, q=7000517 (0.01681 sec/rel)
total yield: 976, q=8000507 (0.01717 sec/rel)
total yield: 1001, q=9000517 (0.01680 sec/rel)
total yield: 1076, q=10000511 (0.01674 sec/rel)
total yield: 743, q=11000537 (0.01903 sec/rel)
total yield: 701, q=12000509 (0.01930 sec/rel)
total yield: 1147, q=13000511 (0.01846 sec/rel)
total yield: 959, q=14000507 (0.02006 sec/rel)
total yield: 851, q=15000509 (0.01954 sec/rel)
total yield: 628, q=16000507 (0.02236 sec/rel)
[/code]

28-bit primes much better? sec/rel is about 51% of the time as for 27-bit primes, and a bit less than 2x as many rels will be needed yes? So 28-bit is better.

Siever 14e:
[code]
rlim: 7500000
alim: 7500000
lpbr: 28
lpba: 28
mfbr: 56
mfba: 56
rlambda: 2.6
alambda: 2.6
[/code]
[code]$ for i in `seq 3000000 1000000 8000000`; do ~/ggnfs/trunk/bin/gnfs-lasieve4I14e -a test.poly -f $i -c 500; done
Warning: lowering FB_bound to 2999999.
total yield: 2382, q=3000523 (0.02003 sec/rel)
Warning: lowering FB_bound to 3999999.
total yield: 2362, q=4000511 (0.02089 sec/rel)
Warning: lowering FB_bound to 4999999.
total yield: 2611, q=5000503 (0.01995 sec/rel)
Warning: lowering FB_bound to 5999999.
total yield: 4069, q=6000503 (0.02015 sec/rel)
Warning: lowering FB_bound to 6999999.
total yield: 2829, q=7000517 (0.02146 sec/rel)
total yield: 2080, q=8000507 (0.02181 sec/rel)
[/code]
Siever 14e is too slow.

So looks like 28-bit, alim & rlim = 7.5M using siever 13e. Sieve from Q=1M to about Q=14M to get 26M-28M rels.

[QUOTE=jrk;190770]So looks like 28-bit, alim & rlim = 7.5M using siever 13e. Sieve from Q=1M to about Q=14M to get 26M-28M rels.[/QUOTE]

A more interesting approach might be to sieve 2-11M, with 4-7M (the sweetspot) being sieved by 14e siever, and the rest by 13e siever!

[QUOTE=axn;190777]A more interesting approach might be to sieve 2-11M, with 4-7M (the sweetspot) being sieved by 14e siever, and the rest by 13e siever![/QUOTE]
I don't think it would help. Siever 14e is just too slow for this number. I really should not have even tested it.

Next line (2448) is c157, not much ECM yet. (half through with B1=5e4, which is for 25 digit factors) Edit: Finished 214@5e4, starting 25e4. I'll probably finish this and then stop (i.e. not also run any at 1e6) Also ran P-1 and P+1 at 10M unsuccessfully. Status all in DB. Edit 2: finished 25e4 (30-digit), no factors

How much ECM will the c157 need before gnfs?

I will start 1000 curves @ B1=43000000, B2=388112953420. ETA monday morning unless factor is found first.

[QUOTE=jrk;191199]How much ECM will the c157 need before gnfs?

I will start 1000 curves @ B1=43000000, B2=388112953420. ETA monday morning unless factor is found first.[/QUOTE]

I think the last one had about 8000 curves at 43M.

[QUOTE=jrk;191199]I will start 1000 curves @ B1=43000000, B2=388112953420.[/QUOTE]
Finished. No factors.

I've queued another 1000 curves @ B1=43000000, B2=388112953420. ETA is Wednesday morning.