F46,300e12,325e12 reran, and it's the same result as the first run (no surprise there)

no factor for F46 from 2^96 to 2^97 (k range: 300000000000000 to 325000000000000) [mmff 0.20mmff mfaktc_barrett108_F30_61gs]

This is me checking in:

[CODE]
F31 has a factor: 46931635677864055013377 [TF:75:76:mmff 0.20mmff mfaktc_barrett89_F30_61gs]
found 1 factor for F31 from 2^75 to 2^76 [mmff 0.20mmff mfaktc_barrett89_F30_61gs]
no factor for F26 from 2^64 to 2^65 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^65 to 2^66 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^66 to 2^67 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^67 to 2^68 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^68 to 2^69 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^69 to 2^70 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^70 to 2^71 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^71 to 2^72 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^72 to 2^73 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
no factor for F26 from 2^73 to 2^74 [mmff 0.20mmff mfaktc_barrett89_F0_29gs]
[/CODE]
Seems that is working and it is fast.
I saw people do not jump too much for mm89 and mm107, so I took:
[CODE]
MMFactor=89,1e14,2e14
MMFactor=107,1e14,2e14
[/CODE]
I saw the ranges available in the tables. Count me in as LaurV

[QUOTE=LaurV;310882]
I saw people do not jump too much for mm89 and mm107, so I took:
[CODE]
MMFactor=89,1e14,2e14
MMFactor=107,1e14,2e14
[/CODE]I saw the ranges available in the tables. Count me in as LaurV[/QUOTE]

Welcome aboard! :smile:

Luigi

sorry if it's dumb but when i see
[code]
61 132,732,600,000,000 135,795,660,000,000 Andrea (TF) 2012-09-01 2012-09-01
61 3,573,570,000,000,000 5,000,000,000,000,000 Rocke Verser 2012-09-09
[/code]

I wonder, what happened to the k between 135,795,660,000,000 and 3,573,570,000,000,000 ?

[QUOTE=firejuggler;310885]sorry if it's dumb but when i see
[code]
61 132,732,600,000,000 135,795,660,000,000 Andrea (TF) 2012-09-01 2012-09-01
61 3,573,570,000,000,000 5,000,000,000,000,000 Rocke Verser 2012-09-09
[/code]I wonder, what happened to the k between 135,795,660,000,000 and 3,573,570,000,000,000 ?[/QUOTE]
I don't know where 3.573e15 came from. Perhaps others have done the range, but it isn't reported. But since I have the first assignment, like frmky above, I am doing "and below" to the limits of mmff-0.20:
MMFactor=61,96,5e15

[QUOTE=Prime95;310799]Now that testing each factor is slower, it will pay to sieve a little deeper. That's why we need to generate a table of best GPUSieveSize settings for each Fermat/MM number. Long-term I suppose auto-tuning needs to be added to the source code.[/QUOTE]

Tested my Geforce GTX 460 on a Q9450 (Yorksfield) 2.66 Ghz. If I have to use the computer I have to set GPUSieveSize=4 to reduce screenlag otherwise it can be at 16. Here is a table of best GPUSievePrimes for the double mersenneprimes. I can test fermat as well but will wait until the new version is out.

[CODE] Best
GPUSieveSize GPUSievePrimes Speed
MM61 16 300k-400k 248.95 M/s
MM61 4 275k-300k 238.47 M/s

MM89 16 525k-675k 113.12 M/s
MM89 4 500k-550k 109.17 M/s

MM107 16 550k-750k 104.75 M/s
MM107 4 450k-650k 101.19 M/s

MM127 16 900k-1075k 67.01 M/s
MM127 4 600k-800k 64.68 M/s[/CODE]


Btw I'm getting:

[QUOTE]ERROR: Exponentiation falure[/QUOTE]

for MM61 below 96 bit (k<1.7*10[sup]10[/sup]).

Here is the next version of mmff.

The big change is how Fermat factoring is handled. Instead of specifying the Fermat number you want to factor, you now specify the factors (k*2^N+1) to try. This is compatible with the way fermatsearch.org has always done things. If you want to try to find a factor of F33 you'll need to choose an N >= 35 (and you might find a factor of F32, F31, F30, ... instead).

If you reserved a Fermat range from fermatsearch.org, please retest it with the new executable (but add 2 to the exponent in worktodo.txt). For example, if you tested a k range for F30 (FermatFactor=30,1.5e15,1.6e15) using the previous mmff, Luigi had you reserve a range using N=32. With this executable, retest the range using FermatFactor=32,1.5e15,1.6e15.

The good news? This change will double our Fermat factoring throughput.

Other minor changes include validating GPUSieveProcessSize. Output of the range of k's has 6,9,12,or 15 trailing zeros truncated -- replaced with M,G,T, or Q for easier readability.

Why Q? P is what is most commonly used on the forum for 10^15. It stands for Peta-.

Linux 64-bit executable attached


Go to post 173 for a newer version

[QUOTE=henryzz;310916]Why Q? P is what is most commonly used on the forum for 10^15. It stands for Peta-.[/QUOTE]

I didn't know. I chose Q for quadrillion. I'll change it to P next time.

[QUOTE=Prime95;310919]I didn't know. I chose Q for quadrillion. I'll change it to P next time.[/QUOTE]
It doesn't happen that often:smile: but it would confuse some people. It is also being consistent since you are using G instead of B for Billion.