A few questions :) - Page 2

xtreme2k · 2002-08-17, 16:03

http://www.gchoi.com/images/p95.jpg

Is this a bit slow for 1000 iterations on a P4 2.5GHz?

Tasuke · 2002-08-17, 16:38

You are not yet doing the Iteration (LL)portion of the test. You are currently doing the p-1 factoring. Your timings look about right.

xtreme2k · 2002-08-17, 17:15

Yah I know its not LL yet, it just looks a bit slow to me. Going for about 3% per hour...

bayanne · 2002-08-17, 19:15

Nah, it really is doing fine, don't worry. This is BIG crunching on a 33m

dswanson · 2002-08-18, 00:34

33252977 took me a little over 67 hours on a P4 1.8 with 448 Mb allocated. That's about 1.5% per hour...

[Tue May 28 14:29:20 2002]
Self-test 1792K passed!
[Fri May 31 09:58:40 2002]
UID: dswanson/pc1800A, M33252977 completed P-1, B1=385000, B2=10395000, E=2, WY1: 5CE27B7D

vader · 2002-08-18, 14:18

If you have another computer (non-P4) you may want to do your factoring on that one and manually move the exponent to the P4 when it is time for the LL test.

If I remember correctly my Duron 1000 MHz is about twice as fast as my P4 1600 MHz when it comes to factoring (although I think that was regular trial factoring and not P-1). So it seemed like a waste to let the P4 do anything other than LL, and now I do all factoring on the Duron.

I don't think I ever checked how the two processors compared on P-1. Does anyone have any info on that?

(By the way, all this applies to v21.4, I don't know if anything has happened to factoring speed in the more recent versions)

xtreme2k · 2002-08-18, 14:27

Quote:

Originally Posted by dswanson

33252977 took me a little over 67 hours on a P4 1.8 with 448 Mb allocated. That's about 1.5% per hour...

[Tue May 28 14:29:20 2002]
Self-test 1792K passed!
[Fri May 31 09:58:40 2002]
UID: dswanson/pc1800A, M33252977 completed P-1, B1=385000, B2=10395000, E=2, WY1: 5CE27B7D

Thats the COMPLETE factoring right? Not a Stage 2 P-1? mine is only a Stage 2 P-1 and its already taking 3% her hour. Not counting the initial trial factor to 2^68 and Stage 1 P-1 I think mine is way over 4 days.

bayanne · 2002-08-18, 14:34

Quote:

Originally Posted by vader

(By the way, all this applies to v21.4, I don't know if anything has happened to factoring speed in the more recent versions)

In version 22.7 :

New features in Version 22.7 of prime95.exe

"Trial factoring above 2^64 on a P4 now uses the SSE2 instructions. This is more than four times as fast as the previous version. "

Worth getting the upgrade :D

dswanson · 2002-08-18, 22:42

Quote:

Originally Posted by xtreme2k

Thats the COMPLETE factoring right? Not a Stage 2 P-1? mine is only a Stage 2 P-1 and its already taking 3% her hour. Not counting the initial trial factor to 2^68 and Stage 1 P-1 I think mine is way over 4 days.

It's a complete stage 1 + stage 2. I didn't have to factor any deeper because the exponent was assigned to me already factored to sufficient depth, i.e. I was given
Test=33252977,68,0
which is now
Test=33252977,68,1

I don't have the breakdown for hours in stage 1 vs hours in stage 2.

The 67 hours is a bit misleading, since my wife crashed the computer in the middle of the stage 2, so it was down for a few hours before I got home from work to rescue it. :(

Stormblade · 2002-08-19, 00:21

Quote:

don't think I ever checked how the two processors compared on P-1. Does anyone have any info on that?

P-1 testing uses the SSE2 code, so the P4 will be much faster. Factoring does not use SSE2 code (except in the beta client, and then only over 64 bits) so that is why the Athlon is faster at that..

garo · 2002-08-19, 06:05

Hey hey!
Kinda offtopic but just wanted to say hello to dswanson. Thanks for dropping in. I am one of those you contend with for small DC's everyday at 0600GMT. It sucks when you beat me to it :) but much better than going to some random fellow who is going to take months

Onwards to proving M38!!

2002-08-18, 14:18	#17
vader Aug 2002 5₁₆ Posts	P4 speed (factoring) If you have another computer (non-P4) you may want to do your factoring on that one and manually move the exponent to the P4 when it is time for the LL test. If I remember correctly my Duron 1000 MHz is about twice as fast as my P4 1600 MHz when it comes to factoring (although I think that was regular trial factoring and not P-1). So it seemed like a waste to let the P4 do anything other than LL, and now I do all factoring on the Duron. I don't think I ever checked how the two processors compared on P-1. Does anyone have any info on that? (By the way, all this applies to v21.4, I don't know if anything has happened to factoring speed in the more recent versions)

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Two questions:	Dubslow	GPU Computing	1	2011-08-05 18:22
Questions about the QS	Carmichael	Factoring	8	2007-04-10 11:30
5 questions	OmbooHankvald	Factoring	6	2005-08-28 19:31
Questions	OmbooHankvald	Prime Sierpinski Project	2	2005-08-01 20:18
LLR questions	OmbooHankvald	Math	6	2005-06-23 11:42

2002-08-17, 16:03	#12
xtreme2k Aug 2002 2·3·29 Posts	http://www.gchoi.com/images/p95.jpg Is this a bit slow for 1000 iterations on a P4 2.5GHz?

2002-08-17, 16:38	#13
Tasuke Aug 2002 2³×3² Posts	You are not yet doing the Iteration (LL)portion of the test. You are currently doing the p-1 factoring. Your timings look about right.

2002-08-17, 17:15	#14
xtreme2k Aug 2002 2×3×29 Posts	Yah I know its not LL yet, it just looks a bit slow to me. Going for about 3% per hour...

2002-08-17, 19:15	#15
bayanne "Tony Gott" Aug 2002 Yell, Shetland, UK 2²×83 Posts	Nah, it really is doing fine, don't worry. This is BIG crunching on a 33m

2002-08-18, 00:34	#16
dswanson Aug 2002 2³×5² Posts	33252977 took me a little over 67 hours on a P4 1.8 with 448 Mb allocated. That's about 1.5% per hour... [Tue May 28 14:29:20 2002] Self-test 1792K passed! [Fri May 31 09:58:40 2002] UID: dswanson/pc1800A, M33252977 completed P-1, B1=385000, B2=10395000, E=2, WY1: 5CE27B7D

2002-08-19, 06:05	#22
garo Aug 2002 Termonfeckin, IE 2²·691 Posts	Hey hey! Kinda offtopic but just wanted to say hello to dswanson. Thanks for dropping in. I am one of those you contend with for small DC's everyday at 0600GMT. It sucks when you beat me to it :) but much better than going to some random fellow who is going to take months Onwards to proving M38!!