Possible idea for improvement

Kevin · 2002-08-26, 04:52

First and foremost, like almost all ideas I have (especially with technology), there could be a large error in logic or some things that don't work out but it's better to to at least get the idea out so someone who's less of an idiot might make it workable, and on to the plan.....

It has to do with P-1 factoring. I'm not sure about it's success rate, but this may make it go faster. Instead of having a person do P-1 factoring during their pre-LL factoring stuff, we could assign it to computers that have enough RAM to fill all 12 data storage spaces (or whatever all that RAM is used for) for stage 2. But that doesn't work as well with the current client. If this idea proves worthy, maybe a small, seperate client could be made for P-1 factoring. It would have an easy stop/start option so users could stop whenever they were using their comp and needed RAM. Also during P-1, from what I've seen in the task manager, is that the CPU is barely used [edit]it does use CPU cycles, but it is usually fairly low, and only occasionally spikes up tothe top[/edit]. This means that "power users" could have Prime95 running full time on idle CPU cycles, and they could run P-1 factoring whenever they're not there (or at least overnight). This would also evenutally eliminate the memory settings in Prime95, or at least move it to prime.ini. I realize it would be too hard to implement this into ranges currently being tested, but maybe we could start on exponents from 20 million and up. Other than writing the code for everything, the only problems I see right now are having all the RAM usage shorten the life of the memory, and implementing this into PrimeNet.

Any specific flaws with this? Or is it just too complicated to implement? Or could I actually be right about something? (probably not....)

Xyzzy · 2002-08-26, 07:57

An ultra simple P-1 client would be cool... All I do is P-1 factor... I don't even need PrimeNet access, since we only get .001 P90 years per exponent... A command line client that accepts input from either a text stream or a file would be ideal...

The current client works fine, but if I am not careful it will try to connect... So I run it is "Stress Test" mode with no user information...

Right now I am in a series of exponents that takes less than an hour each... It doesn't seem to spend much time at all in stage two...

How long do higher exponents take to P-1 factor?

2002-08-26, 04:52	#1
Kevin Aug 2002 Ann Arbor, MI 433 Posts	Possible idea for improvement First and foremost, like almost all ideas I have (especially with technology), there could be a large error in logic or some things that don't work out but it's better to to at least get the idea out so someone who's less of an idiot might make it workable, and on to the plan..... It has to do with P-1 factoring. I'm not sure about it's success rate, but this may make it go faster. Instead of having a person do P-1 factoring during their pre-LL factoring stuff, we could assign it to computers that have enough RAM to fill all 12 data storage spaces (or whatever all that RAM is used for) for stage 2. But that doesn't work as well with the current client. If this idea proves worthy, maybe a small, seperate client could be made for P-1 factoring. It would have an easy stop/start option so users could stop whenever they were using their comp and needed RAM. Also during P-1, from what I've seen in the task manager, is that the CPU is barely used [edit]it does use CPU cycles, but it is usually fairly low, and only occasionally spikes up tothe top[/edit]. This means that "power users" could have Prime95 running full time on idle CPU cycles, and they could run P-1 factoring whenever they're not there (or at least overnight). This would also evenutally eliminate the memory settings in Prime95, or at least move it to prime.ini. I realize it would be too hard to implement this into ranges currently being tested, but maybe we could start on exponents from 20 million and up. Other than writing the code for everything, the only problems I see right now are having all the RAM usage shorten the life of the memory, and implementing this into PrimeNet. Any specific flaws with this? Or is it just too complicated to implement? Or could I actually be right about something? (probably not....)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Windows 10 in Ubuntu, good idea, bad idea, or...?	jasong	jasong	8	2017-04-07 00:23
A new Fermat Factorization improvement?	siegert81	Factoring	5	2011-06-13 02:58
performance improvement with assembly	bsquared	Software	15	2010-09-28 19:00
Possible improvement of quadratic sieve	Random Poster	Factoring	4	2010-02-12 03:09
snewpgen improvement very much appreciated :)	jasong	Software	3	2007-11-23 04:06

2002-08-26, 07:57	#2
Xyzzy Aug 2002 2×3²×13×37 Posts	An ultra simple P-1 client would be cool... All I do is P-1 factor... I don't even need PrimeNet access, since we only get .001 P90 years per exponent... A command line client that accepts input from either a text stream or a file would be ideal... The current client works fine, but if I am not careful it will try to connect... So I run it is "Stress Test" mode with no user information... Right now I am in a series of exponents that takes less than an hour each... It doesn't seem to spend much time at all in stage two... How long do higher exponents take to P-1 factor?