P-1 factoring anyone? - Page 98

Dubslow · 2012-01-24, 20:19

I think those are better than average numbers, if I recall correctly. Someone somewhere once got like two workers on four cores running at like 9 ms per iter, but with all four cores on one worker only got 7 ms, which is clearly quite less than 90% efficiency of the 2 cores per, much less compared to 1 core per. Maybe on your computer, it could be that memory bandwidth is more of a problem than the one I mention?

James Heinrich · 2012-01-24, 20:38

Core2 generation (e.g. Q6xxx, Q8xxxx, Q9xxxx) do lousy at multithreaded assignments.
Core-i generation (e.g. Core i3/i5/i7) do significantly better.

Dubslow · 2012-01-25, 00:19

http://mersenne-aries.sili.net/expon...tails=52567117
What's the rule on composite factors?
whoops mod please move to http://www.mersenneforum.org/showthr...=13977&page=11

kladner · 2012-01-25, 20:44

I am carrying over a digressive discussion from the "Assignment discrepancy" thread-
http://www.mersenneforum.org/showthr...t=16352&page=3

The general drift over there are the relations between RAM Allocation and Number of Relative Primes, Bounds, etc. in P-1 Factoring.

Jwb52z · 2012-01-25, 20:50

How much do I need to worry about the validity of my P-1 results when I noticed over the last week or so that my laptop's clock loses about a minute a day unless I resync it manually?

chalsall · 2012-01-25, 21:28

Quote:

Originally Posted by kladner

The general drift over there are the relations between RAM Allocation and Number of Relative Primes, Bounds, etc. in P-1 Factoring.

I have successfully created a new thread: P-1 Discussion.

Sorry for the hesitation, but the last time I tried to do this I lost several messages.

(The supermods laughed in my general direction, and told me that this was to be expected of new moderators...

)

chalsall · 2012-01-25, 21:42

Quote:

Originally Posted by Jwb52z

How much do I need to worry about the validity of my P-1 results when I noticed over the last week or so that my laptop's clock loses about a minute a day unless I resync it manually?

Do you run NTP?

Jwb52z · 2012-01-27, 01:50

Quote:

Originally Posted by chalsall

Do you run NTP?

I am guessing I don't as I don't knwo what that means. I think I fixed my problem by vacuuming out the vent hole slit spaces on the bottom of my laptop and the keyboard. It's now appearing to be running as if it were new.

James Heinrich · 2012-01-27, 02:57

Quote:

Originally Posted by Jwb52z

I am guessing I don't as I don't knwo what that means.

NTP = Network Time Protocol. If you run Windows, you're familiar with the concept where Windows can sync time over the internet.

Dubslow · 2012-01-27, 03:54

Quote:

Originally Posted by James Heinrich

NTP = Network Time Protocol. If you run Windows, you're familiar with the concept where Windows can sync time over the internet.

Edit: Damn not-noticing-there's-another-page.

What I had written:

Quote:

NTP is an internet time synchronization protocol. If you right click on the clock, you can adjust it; one of the options on those various pages should be something to the effect of "Synchronize with internet servers" or the like. (I'm not using Windows right now so I can't attest to exact directions, but Google should clear up what I can't.)

Mini-Geek · 2012-01-27, 12:57

Quote:

Originally Posted by James Heinrich

I don't think there's any particular limit to the size of factors that could be found with any particular bounds (? someone correct me if I'm wrong, please).

According to http://www.mersennewiki.org/index.ph...ization_Method, E is chosen so that each prime is roughly the size of B1, so I believe there is a maximum. E.g. if it chose 1024 as 2's exponent, I believe it wouldn't find a prime with P-1=2^1026*... This would make the maximum from stage 1 approximately p*B1^pi(B1). Stage 2 adds approximately a factor of B2, (relatively insignificant) and I don't know what sort/number of factors Brent-Suyama adds.

2012-01-25, 00:19	#1070
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 1110000110101₂ Posts	http://mersenne-aries.sili.net/expon...tails=52567117 What's the rule on composite factors? whoops mod please move to http://www.mersenneforum.org/showthr...=13977&page=11 Last fiddled with by Dubslow on 2012-01-25 at 01:18

2012-01-25, 20:44	#1071
kladner "Kieren" Jul 2011 In My Own Galaxy! 2·3·1,693 Posts	Thread Continuation: RAM and P-1 I am carrying over a digressive discussion from the "Assignment discrepancy" thread- http://www.mersenneforum.org/showthr...t=16352&page=3 The general drift over there are the relations between RAM Allocation and Number of Relative Primes, Bounds, etc. in P-1 Factoring.

2012-01-24, 20:19	#1068
Dubslow Basketry That Evening! "Bunslow the Bold" Jun 2011 40<A<43 -89<O<-88 3×29×83 Posts	I think those are better than average numbers, if I recall correctly. Someone somewhere once got like two workers on four cores running at like 9 ms per iter, but with all four cores on one worker only got 7 ms, which is clearly quite less than 90% efficiency of the 2 cores per, much less compared to 1 core per. Maybe on your computer, it could be that memory bandwidth is more of a problem than the one I mention?

2012-01-24, 20:38	#1069
James Heinrich "James Heinrich" May 2004 ex-Northern Ontario 3·5·227 Posts	Core2 generation (e.g. Q6xxx, Q8xxxx, Q9xxxx) do lousy at multithreaded assignments. Core-i generation (e.g. Core i3/i5/i7) do significantly better.

2012-01-25, 20:50	#1072
Jwb52z Sep 2002 2·3·7·19 Posts	How much do I need to worry about the validity of my P-1 results when I noticed over the last week or so that my laptop's clock loses about a minute a day unless I resync it manually?