[CODE]
no factor for M3321929173 from 2^75 to 2^78 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929173 from 2^78 to 2^79 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929173 from 2^79 to 2^80 [mfaktc 0.08-pre5 95bit_mul32]

no factor for M3321929179 from 2^75 to 2^78 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929179 from 2^78 to 2^79 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929179 from 2^79 to 2^80 [mfaktc 0.08-pre5 95bit_mul32]

no factor for M3321929197 from 2^75 to 2^78 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929197 from 2^78 to 2^79 [mfaktc 0.08-pre5 95bit_mul32]
no factor for M3321929197 from 2^79 to 2^80 [mfaktc 0.08-pre5 95bit_mul32]
[/CODE]

Oliver

[CODE]
no factor for M3321929519 from 2^76 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321929519 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
no factor for M3321929519 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]

no factor for M3321929563 from 2^76 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321929563 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
no factor for M3321929563 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]

no factor for M3321929573 from 2^76 to 2^78 [mfaktc 0.08 95bit_mul32]
no factor for M3321929573 from 2^78 to 2^79 [mfaktc 0.08 95bit_mul32]
no factor for M3321929573 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
[/CODE]

Oliver

Level 14?

[CODE]
no factor for M3321928777 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
no factor for M3321928927 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32]
no factor for M3321928963 from 2^79 to 2^80 [mfaktc 0.08 95bit_mul32][/CODE]

Oliver

[QUOTE=TheJudger;218738]Level 14?[/QUOTE]

Indeed, that finished Level 14. :wblipp:

William

From level 13 to level 14 was the 2nd fastest levelup. :smile:
Only from level 2 to level 3 was faster!

Oliver

[QUOTE=TheJudger;218746]From level 13 to level 14 was the 2nd fastest levelup.[/QUOTE]I haven't been paying sharp attention, but my suggestion would be: to take the lower bit level numbers (75 and 76) run 15 of them all the way to 81. Then to go to level 16, take those that remain at 76 etc. and run 16 of them to 82. Repeat.


Luigi, what about opening a new part of the range and mark it "GPU's Stay Out!"? The work from 60 to 70 would be nice for the tired old Pentii.

[QUOTE=Uncwilly;218755]I haven't been paying sharp attention, but my suggestion would be: to take the lower bit level numbers (75 and 76) run 15 of them all the way to 81. Then to go to level 16, take those that remain at 76 etc. and run 16 of them to 82.[/QUOTE]

Yepp, that is what I did for most exponents when moving from level 13 to 14. I've choosen most exponents which where TFed to 2^75 before and took them to 2^80.

Oliver

[QUOTE=wblipp;218745][QUOTE=TheJudger;218738]Level 14?[/QUOTE]Indeed, that finished Level 14. :wblipp:

William[/QUOTE]Outstanding!

Another option for proceeding would be to go deep rather than broad. In other words take an individual candidate to a much higher bit level. I believe that this is an approach taken by the 100 million digits prefactor project.

Judger, if it takes you around a day* to get one candidate from 79 to 80 bits, then to go from 80 to 84 bits would take about a month, two months to go to 85 bits. This is around the time that leading edge exponents take to test for GIMPS on leading edge hardware.

*I pulled this number out of the air, I have no idea how long it takes, but with your lightning pace I figure I can't be that far out.

[QUOTE=Uncwilly;218755]Pentii[/QUOTE]

[pedant]The proper Latin plural would be Pentia.[/pedant]

[QUOTE=Uncwilly;218755]I haven't been paying sharp attention, but my suggestion would be: to take the lower bit level numbers (75 and 76) run 15 of them all the way to 81. Then to go to level 16, take those that remain at 76 etc. and run 16 of them to 82. Repeat.


Luigi, what about opening a new part of the range and mark it "GPU's Stay Out!"? The work from 60 to 70 would be nice for the tired old Pentii.[/QUOTE]

I asked about it in the forum 2 weeks ago, but nobody replied.

If no one replies I will open new CPU-Only ranges next weekend. :smile:

GPU should kick in only after 74-75 bits, do you agree?

Luigi

[QUOTE=lavalamp;218783]Judger, if it takes you around a day* to get one candidate from 79 to 80 bits, then to go from 80 to 84 bits would take about a month, two months to go to 85 bits. This is around the time that leading edge exponents take to test for GIMPS on leading edge hardware.

*I pulled this number out of the air, I have no idea how long it takes, but with your lightning pace I figure I can't be that far out.[/QUOTE]

One day is a [B]good[/B] assumption. With a decent CPU (one core needed) it takes a little bit less than a day on a GTX 275.
A GTX 480 can take 3 exponents from 79 to 80 bits in the same time. 3 CPU cores are needed to do enough sieving. It would work with 2 cores and 2 exponents but I guess it will only be 15% faster per exponent so the throughput decreases.

[QUOTE=ET_;218813]If no one replies I will open new CPU-Only ranges next weekend. :smile:

GPU should kick in only after 74-75 bits, do you agree?

Luigi[/QUOTE]

Agreed!
But keep in mind that there are huge differences in GPU speed. E.g. a GTX 480 yields >40x the throughput of a 8400GS.
"fast" GPUs primary target: above 2^78?
"slow" GPUs primary target: above 2^75?
A "fast" GPU is a GTX 260 and above?

Of course we have to see how many volunteers will use GPUs for OBD.

Oliver