mersenneforum.org 100M digits prefactor project.
 Register FAQ Search Today's Posts Mark Forums Read

 2006-02-06, 17:39 #276 gribozavr     Mar 2005 Internet; Ukraine, Kiev 11·37 Posts Thanks Ernst for spotting the problem! It should be fixed now. It was improper quoting of strings in perl script (single vs. double).
2006-02-06, 18:21   #277
ewmayer
2ω=0

Sep 2002
República de California

2×32×647 Posts

Quote:
 Originally Posted by gribozavr Thanks Ernst for spotting the problem! It should be fixed now. It was improper quoting of strings in perl script (single vs. double).
Yes, that looks better! ;)

 2006-02-07, 11:00 #278 Jwb52z     Sep 2002 79910 Posts I hate to burst your bubble, but the link in this thread to the project website now says that the page cannot be found.
 2006-02-07, 17:35 #279 gribozavr     Mar 2005 Internet; Ukraine, Kiev 11×37 Posts Thanks Jwb52z! Hope it works now... I have also made some changes to the stylesheet to improve page rendering in Internet Explorer.
 2006-02-08, 20:42 #280 gribozavr     Mar 2005 Internet; Ukraine, Kiev 6278 Posts I have uploaded Mfactor_win32_bat.tar.gz -- Mfactor automation bat files for win32. They were explained in the guide, but the file wasn't there.
 2006-02-09, 16:23 #281 Footmaster     Mar 2004 3910 Posts I'd like to reserve M332194169 to 2^68 thanks Foots
 2006-02-10, 18:24 #282 gribozavr     Mar 2005 Internet; Ukraine, Kiev 11·37 Posts The next thing to implement is stats. But how to define a "stats point"? Since I'm going to release exponents to deep search at 65 bits, then the first thing that comes to mind is to define 1 point as factoring from 1 to 65 bits. Formula is: POINTS = (2^END_BITS - 2^START_BITS) / 2^65 Then: 2^1 - 2^65 = 1 point 2^65 - 2^66 = 1 point 2^66 - 2^67 = 2 points 2^67 - 2^68 = 4 points 2^68 - 2^69 = 8 points 2^69 - 2^70 = 16 points 2^70 - 2^71 = 32 points 2^71 - 2^72 = 64 points 2^72 - 2^73 = 128 points 2^73 - 2^74 = 256 points 2^74 - 2^75 = 512 points But I see one problem -- it is currently minor, but unacceptable in the long run. The larger the exponent is, less factors per bit are tested and thus stats become unfair. For example, testing 332,192,831 from 65 to 66 bits takes 10 times as long as 3,321,928,319 from 65 to 66 bits. Are there other possible "stats points"?
2006-02-11, 17:16   #283

"Richard B. Woods"
Aug 2002
Wisconsin USA

1E0C16 Posts

Quote:
 Originally Posted by gribozavr Formula is: POINTS = (2^END_BITS - 2^START_BITS) / 2^65 Then: 2^1 - 2^65 = 1 point 2^65 - 2^66 = 1 point 2^66 - 2^67 = 2 points 2^67 - 2^68 = 4 points 2^68 - 2^69 = 8 points 2^69 - 2^70 = 16 points 2^70 - 2^71 = 32 points 2^71 - 2^72 = 64 points 2^72 - 2^73 = 128 points 2^73 - 2^74 = 256 points 2^74 - 2^75 = 512 points But I see one problem -- it is currently minor, but unacceptable in the long run. The larger the exponent is, less factors per bit are tested and thus stats become unfair. For example, testing 332,192,831 from 65 to 66 bits takes 10 times as long as 3,321,928,319 from 65 to 66 bits. Are there other possible "stats points"?
Put a 1/exponent multiplier into your points formula to compensate for the effect you've pointed out.

Last fiddled with by cheesehead on 2006-02-11 at 17:29

 2006-02-11, 17:54 #284 gribozavr     Mar 2005 Internet; Ukraine, Kiev 11×37 Posts Thanks cheesehead for the idea! So, the formula becomes: $\large{\text{POINTS} = \frac{2^{\text{END}} - 2^{START}}{2^{65}} * \frac{\text{EXPMIN}}{EXP}}$ where EXPMIN=332192831 Last fiddled with by gribozavr on 2006-02-11 at 18:00
 2006-02-12, 10:47 #285 Jwb52z     Sep 2002 17×47 Posts Gribozavr, that's an extremely small number if my calculations are correct using your example. It's just a little bit over 9x10^-12.
 2006-02-12, 11:48 #286 gribozavr     Mar 2005 Internet; Ukraine, Kiev 19716 Posts Yes, that is a small number now, but it will grow as the project advances and we start factoring larger exponents. But how much time will pass until then?.. On the other hand, AMD64 and Mfactor can bump the project very much. I have an Athlon64 3000+ 1.8GHz w/ 512Kb cache and spend 95% of its CPU time on 100mdpp. There are many people now with much fatser Athlon64's and even dualcores. Mfactor on AMD64 running under 64-bit Linux is the fastest configuration now. So, if anyone has such configuration, you are welcome to join us! Last fiddled with by gribozavr on 2006-02-12 at 11:49

 Similar Threads Thread Thread Starter Forum Replies Last Post aurashift Software 18 2016-04-14 13:48 xorbe LMH > 100M 189 2010-12-09 08:30 joblack LMH > 100M 1 2009-10-08 12:31 davieddy Lounge 1 2008-10-18 10:40 wblipp ElevenSmooth 0 2003-10-15 16:07

All times are UTC. The time now is 04:21.

Sun Aug 1 04:21:44 UTC 2021 up 8 days, 22:50, 0 users, load averages: 1.46, 2.13, 2.18