P4 question

[Stricker]

why does my P4 2.8GHz run slower than my Athlon XP 1900 (1.6GHz)? With my Athlon being 1.2GHz slower than the P4 normally you would think it would be faster but even with the sse2 client it is only as fast as my AXP WHILE watching TV on my tv card the AXP while watching TV is getting 333kp/s when i'm not watching TV i'm getting 430
the P4 using the sse2 is getting 320-340 using cmov it only get 290

[vjs]

Stricker,

For the best of both portions of this project, I wouldn't sieve with a P4.

If regular test are taking to long do double check using the user name supersecret.

Or try some P-1 factoring, you have a nice P4 that would be fasted in sieve

[Stricker]

i was more hoping to be able to use the power of all 3 of my machines on seive for a while since i have 1 laptop (P4) and 2 AXP

what i decided to do, instead, was toss the P4 back in the main project since it takes ~3 days per test in the main project

its just i chose a seive range that is a little larger than i probably should have, since with both AXP it will probably take ~3 months

[Mystwalker]

Originally posted by Stricker
i was more hoping to be able to use the power of all 3 of my machines on seive for a while since i have 1 laptop (P4) and 2 AXP

what i decided to do, instead, was toss the P4 back in the main project since it takes ~3 days per test in the main project

You could consider putting the P4 to P-1 factoring. In 3 days, it should find 1-2 factors, each saving us a PRP (--> main project) test wihtin approx. 3 weeks and a double check test in the distant future...

its just i chose a seive range that is a little larger than i probably should have, since with both AXP it will probably take ~3 months

That's no problem. As long as you check in factors from time to time (say, every second week), nobody minds.

[royanee]

I've actually found laptops (Windows ones at least) to respond poorly to sieve. I haven't done anything else with a laptop, but they start fast, then seem to slow down. I think the heat is getting to them and they underclock themselves or something... I had an Athlon mobile 1900 or 2000 or something around there, and it said 540 kp/s for the first 3 iterations, and then 340 kp/s until the end of the range. Sadness.

[Mystwalker]

Originally posted by royanee
I've actually found laptops (Windows ones at least) to respond poorly to sieve. I haven't done anything else with a laptop, but they start fast, then seem to slow down. I think the heat is getting to them and they underclock themselves or something... I had an Athlon mobile 1900 or 2000 or something around there, and it said 540 kp/s for the first 3 iterations, and then 340 kp/s until the end of the range. Sadness.

Really sounds like notebooks with underdimensioned cooling.
My Pentium-M-powered notebook has no problems, though, capable of doing ~500kp/s.

Besides, when it is a cooling problem, a different OS can't make a difference.

[vjs]

Stricker,

A 3 month range within the active range is a little big but no problems whatsoever, like it was said earlier as long as you inform us everyonce in a while and submit factors everyone is happy.

I try to reserve very large ranges 10T, 50-100T ahead of the pack... by the time the 90% level catches up it gets a nice 10T boost.

Also on the P4 if the 3day tests are a little long for you why don't you try running under the user name, garbage you'll get 4m tests I think they take about a day. Or better yet run as supersecret those tests are around 1.1m currently and run very fast, couple hours at most.

It's actually better if the "in the know" people run supersecret and garbage for time to time to bring up these levels and the sieve floor.

[Stricker]

Like i had said when i had originally reserved the block it was the first seiving i was going to do in well over a year

i was hoping that all 3 could share the load mostly since the heatsink design that Dell made i feel is inadequet (sp?) and running the laptop at full throttle (2.8GHz) for 3 days straight was making it heat up to over 65C so i wanted to be able to use the speedstep and scale it back and have all 3 machines running about the same clock speed (~1.6GHz) but when i realized that the laptop was slower than the AXPs even when running full throttle i scratched that idea.

i think once it finishes the block it is working on now (one i sneaker-netted from my linux AXP) i'll switch names to supersecret

edit: after making this post i took a compressed air can to my laptop heat sink and there was a cloud of dust belching out of it (i just cleaned the heat sink in august) and the full throttle temp dropped to a more reasonable range 50-55C