Works perfectly fine, but I havent noticed the difference between 5.0 and 4.2.
MMFactor=127,600e12,700e12 (not reserved, just benching) yielded in the same 142s per classes and 152M/s raw rate.
Used one GTX 480, with no monitor attached to it, so nothing could have interfered.
Maybe the speedup is specific for some gpu type?
Like sm_21 or sm_30 ?

V4 works good for me except the ctrl/c issue which is still there (crashing when ctrl/c is pressed or when the last assignment is finished). V5 is asking for a new driver ("you have an older driver then the one used to compile the program, need to update", beside of using the right cudart), which I am not going to install, I had lots of headache to make the current one to work as I want, so if it works, don't fix it. V4 works ok and it is bloody fast!
[code]
===========================
v4.2
===========================
[Tue Sep 18 01:59:45 2012]
F31 has a factor: 46931635677864055013377 [TF:75:76:mmff 0.23 mfaktc_barrett89_F32_63gs]
[Tue Sep 18 01:59:47 2012]
found 1 factor for k*2^33+1 in k range: 5460G to 5470G (76-bit factors) [mmff 0.23 mfaktc_barrett89_F32_63gs]
[Tue Sep 18 01:59:49 2012]
F28 has a factor: 1766730974551267606529 [TF:70:71:mmff 0.23 mfaktc_barrett89_F32_63gs]
[Tue Sep 18 01:59:49 2012]
found 1 factor for k*2^36+1 in k range: 25709M to 25710M (71-bit factors) [mmff 0.23 mfaktc_barrett89_F32_63gs]
[Tue Sep 18 01:59:50 2012]
F52 has a factor: 1475547810493913550438096961537 [TF:100:101:mmff 0.23 mfaktc_barrett108_F32_63gs]
[Tue Sep 18 01:59:54 2012]
found 1 factor for k*2^54+1 in k range: 81900G to 81911G (101-bit factors) [mmff 0.23 mfaktc_barrett108_F32_63gs]
[Tue Sep 18 01:59:56 2012]
F96 has a factor: 8453027931784477309850388309101819121893377 [TF:142:143:mmff 0.23 mfaktc_barrett152_F96_127gs]
[Tue Sep 18 01:59:56 2012]
found 1 factor for k*2^101+1 in k range: 3334G to 3335G (143-bit factors) [mmff 0.23 mfaktc_barrett152_F96_127gs]
[Tue Sep 18 01:59:59 2012]
F133 has a factor: 3836232386548105510567872577199319351015739156856833 [TF:171:172:mmff 0.23 mfaktc_barrett172_F128_159gs]
[Tue Sep 18 02:00:00 2012]
found 1 factor for k*2^135+1 in k range: 88000M to 88100M (172-bit factors) [mmff 0.23 mfaktc_barrett172_F128_159gs]
<here it crashes>
===========================
v5.0
===========================
not tested

[/code]

A limit of mfaktc is that it cannot trial factor Mersenne numbers with exponents larger then 2[SUP]32[/SUP]-1 = 4,294,967,295.

A good thing with mmff is that it can do trial factoring above that limit. If I would like to trial factor a Mersenne number too large for mfaktc, like a 10Billion digit Mersenne number is there any way I can use mmff for this (now or in the future)?

[QUOTE=aketilander;312190]A good thing with mmff is that it can do trial factoring above that limit. If I would like to trial factor a Mersenne number too large for mfaktc, like a 10Billion digit Mersenne number is there any way I can use mmff for this (now or in the future)?[/QUOTE]

No. I've disabled standard Mersenne factoring in mmff (in case my many changes broke it). It would be easier to add that feature to mfaktc.

New number types like Mersenne factoring exponents more than 32-bits, generalized Fermats, bigger double-Mersennes, bigger Fermats, etc. requires writing new kernels.

[QUOTE=Prime95;312193]No. I've disabled standard Mersenne factoring in mmff (in case my many changes broke it). It would be easier to add that feature to mfaktc.

New number types like Mersenne factoring exponents more than 32-bits, generalized Fermats, bigger double-Mersennes, bigger Fermats, etc. requires writing new kernels.[/QUOTE]

OK I see, thank you for the answer.

Version 0.24

Another minor upgrade.

1) Fixed bug in calculating which Fermat number is divisible by a found factor.
2) If GPUSievePrimes is not set in mmff.ini, then mmff chooses a default value based on each entry in worktodo.txt. It may not choose the optimal GPUSievePrimes value, but it should be in the ballpark. Report to me any instances where it chooses a wildly non-optimal setting.
3) The -st and -st2 mfaktc command line arguments (self-test) are now ignored.
4) Some uninitialized mfaktc CPU sieving pointers are no longer freed at exit. Maybe this will solve the crash some were reporting at exit.

Because the mmff version number is written to save files, finish your current range before doing any upgrading or try the lightly-tested -nocheck command line argument to force using an old save file.

Sources:

Linux 64-bit executable:

Windows 32-bit & 64-bit executables:

Tested. Ctrl+C works now (Thanks George! BTW-what lines did you change, I tried to find it and couldn't?)

If anyone needs a different CUDA version, let me know.

[CODE]F31 has a factor: 46931635677864055013377 [TF:75:76:mmff 0.24 mfaktc_barrett89_F32_63gs][/CODE]

[QUOTE=flashjh;312401]Ctrl+C works now (Thanks George! BTW-what lines did you change, I tried to find it and couldn't?)[/QUOTE]

The very end of mfaktc.c -- I commented out the calls to destroy streams and sieve_free.

[QUOTE=Prime95;312396]Version 0.24

Another minor upgrade.

1) Fixed bug in calculating which Fermat number is divisible by a found factor.
2) If GPUSievePrimes is not set in mmff.ini, then mmff chooses a default value based on each entry in worktodo.txt. It may not choose the optimal GPUSievePrimes value, but it should be in the ballpark. Report to me any instances where it chooses a wildly non-optimal setting.
3) The -st and -st2 mfaktc command line arguments (self-test) are now ignored.
4) Some uninitialized mfaktc CPU sieving pointers are no longer freed at exit. Maybe this will solve the crash some were reporting at exit.

Because the mmff version number is written to save files, finish your current range before doing any upgrading or try the lightly-tested -nocheck command line argument to force using an old save file.

Sources:[/QUOTE]
One thing I've noticed is that the %g variable is returning huge values (at least it does for me on my computer).
e.g. 19071091533368053000000000000000000000.00

[QUOTE=mognuts;312423]One thing I've noticed is that the %g variable is returning huge values (at least it does for me on my computer).
e.g. 19071091533368053000000000000000000000.00[/QUOTE]

I need a bit more information to investigate this.