1600@3e6 and 1600@11e6 are done on c145. Now running some curves with B1=43e6.

Msieve polyfind is still running, so it may take me a bit more than 4 days for the gnfs (I was guessing based on a recent c140).

In the meantime I've been running the recommended ecm progression:
904 @ 1e6
2350 @ 3e6
4480 @ 11e6
2650 (and counting) @ 43e6
Plus aliqueit's usual P-1/P+1 at each level. Nothing yet...

I did a quick 4*1234 @ 1e7 overnight without success

Another 1000 43e6's here (and smaller curves too).
Time to gnfs.

Note that you can kill msieve -np process, and the script should progress fine; the poly is probably too good to be true by now -- you don't need to overcook it.


P.S. ...we can all have a look at the list of the polys and vote! Just for the fun of it.
[I]grep -n alp *.p |sort +6rg |head[/I]

OK, I don't have a good feel for the polyfind step yet so I was just going with the factMsieve.pl default -- I just checked the log file and it had set the timeout to 97 hours, which does seem like a long time. I don't think it adjusts for the number of cores so its heuristics may be skewed if the sieving will go ~10x faster than the polyfind. It's on my list to try to make a multi-threaded polyfind script, but it's a little daunting...

Anyway I have started ggnfs, so hopefully we'll have results in a few days.

[QUOTE=bchaffin;236334]OK, I don't have a good feel for the polyfind step yet so I was just going with the factMsieve.pl default -- I just checked the log file and it had set the timeout to 97 hours, which does seem like a long time. I don't think it adjusts for the number of cores so its heuristics may be skewed if the sieving will go ~10x faster than the polyfind. It's on my list to try to make a multi-threaded polyfind script, but it's a little daunting...

Anyway I have started ggnfs, so hopefully we'll have results in a few days.[/QUOTE]

A timeout of 97 hours for a c145 seems OK (or even too short!) for me: I recently had a c145 and found NOT A SINGLE POLYNOMIAL within these 97 hours. (Afterwards I did a search on the GPU overnight on a randomly chosen range and luckily found a good poly.)

P.S.: How fast is your i7? 4 days (even if counted only for sieving) looks a bit optimistic to me.

You are right -- some parameter settings are too tight for a c145 and would better be relaxed. After a few hours on GPU, I've only got one poly; that's not good (the single poly looks ok-ish, though):

[FONT=Arial Narrow]# norm 3.679537e-014 alpha -7.355264 e 8.861e-012 rroots 5[/FONT]

41 @ 43e6 without success

430@11e6, no factor

Interesting. I ran msieve for about 28 hours, and got 85 polys. The best was:
# norm 4.742410e-14 alpha -7.760362 e 1.030e-11 rroots 5

How do I know what's good? Batalov says his looks "ok-ish" but I don't know how you make that judgement... so much to learn. :)

I'm running on 8 single-threaded i7 cores @ 2.93GHz, but sharing the machine with some lower-priority tasks which take some of the cycles. After ~19 hours of sieving, I have 8.2M relations; factMsieve wants 10.6M before going any further.

[QUOTE=bchaffin;236502]Interesting. I ran msieve for about 28 hours, and got 85 polys. The best was:
# norm 4.742410e-14 alpha -7.760362 e 1.030e-11 rroots 5

How do I know what's good? Batalov says his looks "ok-ish" but I don't know how you make that judgement... so much to learn. :)

I'm running on 8 single-threaded i7 cores @ 2.93GHz, but sharing the machine with some lower-priority tasks which take some of the cycles. After ~19 hours of sieving, I have 8.2M relations; factMsieve wants 10.6M before going any further.[/QUOTE]

I think you will need ~23M to 25M relations, so approx. 36 more hours. (unless this is a 29 bit job (i.e. if lpbr and lpba would be 29)), then you will need approx. 40M relations.