[QUOTE=lavalamp;251825]That seems a bit long, shouldn't it be closer to 45 - 60 mins on that system? Or does it significantly slow down when the exponent/bit level goes up?

Based on some old times (for smaller exponents and lower bit levels) on my OC'd i7, I'd estimate about 30 mins.[/QUOTE]


Some things to consider is that number is 61 bits long vs 28 bits for the typical OBD (One Billion Digit) range. Therefore the powermod function will take approx twice as many steps when compared to run on an OBD.

I have a factor on an exponent that's still getting crunched. Do I report that factor (and exponent) now, or should I wait till it's completely done?

Rodrigo

a good news, at least

M3321930173 has a factor: 77840491719563518948103
found 1 factor(s) for M3321930173 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
k= 30517 * 383922211

otherwise...
[quote]
no factor for M3321929929 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321929971 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930011 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930211 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930229 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930323 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930347 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930349 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930371 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930397 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930439 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930461 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930517 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930613 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321930977 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321931043 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
no factor for M3321931057 from 2^76 to 2^77 [mfaktc 0.14-Win barrett79_mul32]
[/quote]

[QUOTE=Rodrigo;251886]I have a factor on an exponent that's still getting crunched. Do I report that factor (and exponent) now, or should I wait till it's completely done?

Rodrigo[/QUOTE]

If the number has been reserved, you can wait and report the whole range at once.

Luigi

[QUOTE=gjmccrac;251845]Some things to consider is that number is 61 bits long vs 28 bits for the typical OBD (One Billion Digit) range. Therefore the powermod function will take approx twice as many steps when compared to run on an OBD.[/QUOTE]

Anyway,

[CODE]
M2305843009213693951 has 0 factors in [2^1, 2^100-1].
[/CODE]

Luigi

Heh, so how long did it take in the end?

Running to 100 bits on that number, the highest value for k would be 274,877,906,944 (well 274,877,906,943 actually, 15 mod 16). The same k for a regular OBD number would put the factor size only around the 70.6 bit level.

So firejuggler, if you want a big factor quickly, just pick some stupidly large exponents. :wink:

What's the max input for factor5 anyway? I know it can trial factor MM127 fine and that's pretty big.

[QUOTE=lavalamp;251935]Heh, so how long did it take in the end?

Running to 100 bits on that number, the highest value for k would be 274,877,906,944 (well 274,877,906,943 actually, 15 mod 16). The same k for a regular OBD number would put the factor size only around the 70.6 bit level.

So firejuggler, if you want a big factor quickly, just pick some stupidly large exponents. :wink:

What's the max input for factor5 anyway? I know it can trial factor MM127 fine and that's pretty big.[/QUOTE]

It took so long because I had ti stop it and restart more than 12 hours later :smile:

So far, no upper limit for Factor5, but There could be some problem (missing factors) if the exponent is larger than 50-60 digits. I should investigate more, and I'll do when the range will be accessed by some distributed project. I could also increase the speed of Factor5 limiting the high end of exponents to 32 bit.


Luigi

[QUOTE=ET_;251937]I could also increase the speed of Factor5 limiting the high end of exponents to 32 bit.[/QUOTE]How much would the speed increase be?

Assuming it's not too much hassle, I absolutely think that would be worth it. All OBD candidates are around the 31.63 bits mark, just below 32 bits. Could the same, or similar, speedup be had for exponents below 53 bits (the upper limit for accurately representing an integer in a double) or 64 bits?

[QUOTE=lavalamp;251948]How much would the speed increase be?

Assuming it's not too much hassle, I absolutely think that would be worth it. All OBD candidates are around the 31.63 bits mark, just below 32 bits. Could the same, or similar, speedup be had for exponents below 53 bits (the upper limit for accurately representing an integer in a double) or 64 bits?[/QUOTE]

Factor5 needs a complete overhaul; I'd like to manage it, but actually my time is limited :no:

I already have ideas about it, but the whole structure of the source should be rewritten, some paths rethinked and so on. I plan to work on it as soon as RealLife[sup](TM)[/sup] lets me.

The speedup should be below 15%.

Luigi

Wow, if work had been spread evenly over all unfactored candidates, they'd be just past 79 bits now.

Based on the CPU calculator at MoreGIMPS, that's a total of 455,600 P90 CPU years.

working on that, lavalamp...