M1061... - Page 5

marc · 2004-10-22, 11:02

Code:

error404   4
xyzzy    126
ET        25
thomasn  400
marc     171

TOTAL    726

Just so we can keep track... have any of the above curves been reported already? We could probably just report each 1000, reset the counter and then go again.

geoff · 2004-10-22, 11:36

One curve with B2=180 billion (gmp-ecm default) counts for more than two with B2=4.29 billion (Prime95 default). We need to keep track how many of each were done.

marc · 2004-10-22, 12:25

Bah. Well all of mine are done with gmp-ecm at B2=180billion.

So on the page http://mersenne.org/ecmm.htm it says 19,300 curves need to be tested. If a gmp-ecm default curve isn't equal to a prime95 curve then which curve equals 1?

R.D. Silverman · 2004-10-22, 15:18

Quote:

Originally Posted by marc

Bah. Well all of mine are done with gmp-ecm at B2=180billion.

So on the page http://mersenne.org/ecmm.htm it says 19,300 curves need to be tested. If a gmp-ecm default curve isn't equal to a prime95 curve then which curve equals 1?

George's goal of ~19K curves is for prime95. If GMP were used throughout,
only about 9K curves would be required. Thus, if you use GMP for step 2,
report it as TWO curves done.

Xyzzy · 2004-10-22, 16:57

152 curves with B1=44e6...
5 curves with B1=11e7...

All with GMP-ECM...

These are raw, with no accounting for the 2:1 ratio...

We need to decide on a formal method for reporting work done to this thread that encompasses all the possible variables... Any suggestions Bob?

geoff · 2004-10-23, 10:21

We could report the number of curves together with a multiplier to this thread, then Xyzzy can do the adding and multiplying and report the total equivalent Prime95 curves to George.

Multipliers for non-default curves can be calculated as P/Q where P is the probability that the curve in question will find a 50 digit factor and Q is the probability that a curve with B1=44e6, B2=4.29e9 would find one. These probabilities can be calculated using the rho.gp program from the output of 'ecm -v'.

The multiplier for a gmp-ecm curve with B1=44e6 and default B2 (the actual value used is 191e9) should be 2.17, calculated as above. Maybe that is unreasonably precise? The value obtained by simply converting the expected curves in the documentation for each program (which assumes B1=43e6) is 19300/9000 = 2.14, so the calculation seems to be reasonable.

ET_ · 2004-10-23, 11:46

Quote:

Originally Posted by geoff

We could report the number of curves together with a multiplier to this thread, then Xyzzy can do the adding and multiplying and report the total equivalent Prime95 curves to George.

Multipliers for non-default curves can be calculated as P/Q where P is the probability that the curve in question will find a 50 digit factor and Q is the probability that a curve with B1=44e6, B2=4.29e9 would find one. These probabilities can be calculated using the rho.gp program from the output of 'ecm -v'.

The multiplier for a gmp-ecm curve with B1=44e6 and default B2 (the actual value used is 191e9) should be 2.17, calculated as above. Maybe that is unreasonably precise? The value obtained by simply converting the expected curves in the documentation for each program (which assumes B1=43e6) is 19300/9000 = 2.14, so the calculation seems to be reasonable.

Maybe a conversion table or formula would do better...

I am actually using Prime95, and maybe others are doing the same.
I think that we should specify which program and which B2 we used, and let the calculations to only one person. This will let everyone free to choose his/her own preferred B2 bound.

Luigi

smh · 2004-10-23, 11:52

For the 55 digit level it might be better to keep track of the gmp-ecm curves (using defaults). B2 for prime95 is to low for this level.

stage 1 can still be done with prime95, but gmp-ecm should be used for stage 2

ET_ · 2004-10-23, 12:17

Quote:

Originally Posted by smh

For the 55 digit level it might be better to keep track of the gmp-ecm curves (using defaults). B2 for prime95 is to low for this level.

stage 1 can still be done with prime95, but gmp-ecm should be used for stage 2

Could you put a link to download the working version of GMP-ECM for cygwin? I'm stuck at 5.0.3 (though still don't use it) as 5.1 seems to have some problems under Windows...

Luigi

smh · 2004-10-23, 14:53

Quote:

Originally Posted by ET_

Could you put a link to download the working version of GMP-ECM for cygwin? I'm stuck at 5.0.3 (though still don't use it) as 5.1 seems to have some problems under Windows...

http://www.geocities.com/greatpsycho/ is the fastest for my P4 that i have found. Dunno about athlons

ET_ · 2004-10-23, 16:16

Quote:

Originally Posted by smh

http://www.geocities.com/greatpsycho/ is the fastest for my P4 that i have found. Dunno about athlons

Hmmm... I can only see version 4 for Win32 out there. Did you try version 5 under Cygwin?

Luigi

2004-10-22, 11:02	#45
marc Jun 2004 UK 139₁₀ Posts	Code: error404 4 xyzzy 126 ET 25 thomasn 400 marc 171 TOTAL 726 Just so we can keep track... have any of the above curves been reported already? We could probably just report each 1000, reset the counter and then go again. Last fiddled with by marc on 2004-10-22 at 11:05

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Predict the number of digits from within the factor for M1061	Raman	Cunningham Tables	12	2013-06-17 21:21
M1061 factored!!!	lycorn	NFS@Home	28	2012-08-30 04:40
Anyone have an ETA for M1061?	Stargate38	NFS@Home	99	2012-08-05 09:38
M1061 - t60	Andi47	Factoring	122	2011-11-25 09:18
P-1 on M1061 and HP49.99	ATH	Factoring	21	2009-10-13 13:16

2004-10-22, 11:36	#46
geoff Mar 2003 New Zealand 10010000101₂ Posts	One curve with B2=180 billion (gmp-ecm default) counts for more than two with B2=4.29 billion (Prime95 default). We need to keep track how many of each were done.

2004-10-22, 12:25	#47
marc Jun 2004 UK 139₁₀ Posts	Bah. Well all of mine are done with gmp-ecm at B2=180billion. So on the page http://mersenne.org/ecmm.htm it says 19,300 curves need to be tested. If a gmp-ecm default curve isn't equal to a prime95 curve then which curve equals 1?

2004-10-22, 16:57	#49
Xyzzy Aug 2002 2×3²×13×37 Posts	152 curves with B1=44e6... 5 curves with B1=11e7... All with GMP-ECM... These are raw, with no accounting for the 2:1 ratio... We need to decide on a formal method for reporting work done to this thread that encompasses all the possible variables... Any suggestions Bob?

2004-10-23, 10:21	#50
geoff Mar 2003 New Zealand 1157₁₀ Posts	We could report the number of curves together with a multiplier to this thread, then Xyzzy can do the adding and multiplying and report the total equivalent Prime95 curves to George. Multipliers for non-default curves can be calculated as P/Q where P is the probability that the curve in question will find a 50 digit factor and Q is the probability that a curve with B1=44e6, B2=4.29e9 would find one. These probabilities can be calculated using the rho.gp program from the output of 'ecm -v'. The multiplier for a gmp-ecm curve with B1=44e6 and default B2 (the actual value used is 191e9) should be 2.17, calculated as above. Maybe that is unreasonably precise? The value obtained by simply converting the expected curves in the documentation for each program (which assumes B1=43e6) is 19300/9000 = 2.14, so the calculation seems to be reasonable.

2004-10-23, 11:52	#52
smh "Sander" Oct 2002 52.345322,5.52471 2245₈ Posts	For the 55 digit level it might be better to keep track of the gmp-ecm curves (using defaults). B2 for prime95 is to low for this level. stage 1 can still be done with prime95, but gmp-ecm should be used for stage 2