So what tasks should I be doing?

PadainFain · 2002-11-13, 22:22

I have a P4 1.5GHz and an Athlon XP 2000+.

What type of work should I assign each of them for the most fruitful use of my CPU time?

Thanks for your advice.

Pad.

Oh and how would I change the work if I wanted to? Both are doing P-1 factoring right now.

quicky · 2002-11-13, 23:13

I think it's always the best to just let the primenet server assign you the work that it thinks fits your pc most.

I think you will get first primality tests on both pc's and that's what it should be.

btw, P-1 factoring is part of a first primality test, so they are both doing those right now.

Deamiter · 2002-11-14, 05:16

to change your type of work unit, go to 'test' and then 'primenet' and check/uncheck away! Now that you've started standard work units, though, I'd stick with them. That gives you the best chance of finding a prime, and both your machines will work at a good pace.

PadainFain · 2002-11-14, 15:50

Thanks for your responses!

Pad.

QuintLeo · 2002-11-16, 00:31

IMO, Any P4 should be doing LL work - either 1'st check or second check. It's a total waste to have them doing factoring.

If that Athlon was mine, I'd have it doing LL work too - not as efficient as a P4, but still quite fast enough....

2002-11-23, 15:06

"What the best work to do" is now complicated by the recent improvements in the factoring program. While the P4 with SSE2 is perhaps 30% faster for the same clock speed than a P3 for LL-testing, for trial factorisation it is 300% faster on a P4 (from Version prime95.exe 22.7 and above) and a P4 1.5 will do a test factorisation in just over a day. So it could now be argued that the P4 should be used for factoring rather than a P3/AMD. A good compromise might be to use the Atherlon for LL-testing only and to select both LL-testing and factorisation work for the P4 (do this by varying the work requested and days or work queued up), then the work can be alternated by cut and pasting in the worktodo.txt file. While it may seem that quote "Any P4 should be doing LL work" I would conjecture that only a SSE2 (e.g. P4) should now do factoring !

Xyzzy · 2002-11-23, 19:11

Up to 64 bits the P4 still is too slow... A friend and I were testing this last night... He has a 1500MHz XP and I have a 2400MHz P4... His computer is 2x faster than mine, even though it is 900MHz slower...

Of course, running a 20M exponent to 64 bits only takes a short time compared to going from 64 to 66 bits...

Joe O · 2002-11-23, 20:24

Does anyone know why George restricted the SSE2 optimization to trial factoring above 64 bits? Why not just use it all the time?

sdbardwick · 2002-11-24, 01:10

Quote:

Originally Posted by Xyzzy

Of course, running a 20M exponent to 64 bits only takes a short time compared to going from 64 to 66 bits...

Yeah, and that's where the SSE2 really kicks in to make P4s quicker overall.
My 2.2GHz P4 takes about 10 hours to factor a M21357997ish number to 66 bits, while my Athlon XP 1900+ (1600MHz) takes about 23 hours to do the same.

2002-11-13, 22:22	#1
PadainFain Nov 2002 2² Posts	So what tasks should I be doing? I have a P4 1.5GHz and an Athlon XP 2000+. What type of work should I assign each of them for the most fruitful use of my CPU time? Thanks for your advice. Pad. Oh and how would I change the work if I wanted to? Both are doing P-1 factoring right now.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Stop receiving new tasks	gromphedyret	Information & Answers	4	2018-01-18 17:21
How do I get computer to do lots of little tasks?	jasong	Information & Answers	4	2007-10-04 20:40
Can I manually add tasks BEFORE the current one? If not, how do I add after anyway?	Nazo	Software	7	2005-09-13 23:00

2002-11-13, 23:13	#2
quicky Sep 2002 Leiden (NL) 14₈ Posts	I think it's always the best to just let the primenet server assign you the work that it thinks fits your pc most. I think you will get first primality tests on both pc's and that's what it should be. btw, P-1 factoring is part of a first primality test, so they are both doing those right now.

2002-11-14, 05:16	#3
Deamiter Sep 2002 3²·13 Posts	to change your type of work unit, go to 'test' and then 'primenet' and check/uncheck away! Now that you've started standard work units, though, I'd stick with them. That gives you the best chance of finding a prime, and both your machines will work at a good pace.

2002-11-14, 15:50	#4
PadainFain Nov 2002 2² Posts	Thanks for your responses! Pad.

2002-11-16, 00:31	#5
QuintLeo Oct 2002 Lost in the hills of Iowa 2⁶·7 Posts	IMO, Any P4 should be doing LL work - either 1'st check or second check. It's a total waste to have them doing factoring. If that Athlon was mine, I'd have it doing LL work too - not as efficient as a P4, but still quite fast enough....

2002-11-23, 15:06	#6
slattm01 1,319 Posts	"What the best work to do" is now complicated by the recent improvements in the factoring program. While the P4 with SSE2 is perhaps 30% faster for the same clock speed than a P3 for LL-testing, for trial factorisation it is 300% faster on a P4 (from Version prime95.exe 22.7 and above) and a P4 1.5 will do a test factorisation in just over a day. So it could now be argued that the P4 should be used for factoring rather than a P3/AMD. A good compromise might be to use the Atherlon for LL-testing only and to select both LL-testing and factorisation work for the P4 (do this by varying the work requested and days or work queued up), then the work can be alternated by cut and pasting in the worktodo.txt file. While it may seem that quote "Any P4 should be doing LL work" I would conjecture that only a SSE2 (e.g. P4) should now do factoring !

2002-11-23, 19:11	#7
Xyzzy Aug 2002 2⁶×3³×5 Posts	Up to 64 bits the P4 still is too slow... A friend and I were testing this last night... He has a 1500MHz XP and I have a 2400MHz P4... His computer is 2x faster than mine, even though it is 900MHz slower... Of course, running a 20M exponent to 64 bits only takes a short time compared to going from 64 to 66 bits...

2002-11-23, 20:24	#8
Joe O Aug 2002 525₁₀ Posts	Does anyone know why George restricted the SSE2 optimization to trial factoring above 64 bits? Why not just use it all the time?