p4 - 1.8 ghz - LL tests?

[joblack]

I've inherited a p4 - 1.8 ghz - is it possible to run LL tests on it? How long would it approximately take to finish one?

[10metreh]

Originally Posted by joblack

I've inherited an p4 - 1.8 ghz - is it possible to run LL tests on it?

It is possible, but INCREDIBLY slow. You're better off doing TF of some kind (probably OBD). To find out how long it would take, use the "Time" option in Prime95.

[lfm]

Originally Posted by joblack

I've inherited a p4 - 1.8 ghz - is it possible to run LL tests on it? How long would it approximately take to finish one?

I'd guess somewhere around 100 days. Not too bad in my book.

[Kevin]

Why is TF always the default option for anything that can't run 50M tests in a reasonable amount of time? What's wrong with double-checks or P-1 factoring? That machine is more than twice the minimum threshold to do double checks, and 36 times the minimum threshold to not be banished to LMH/ODB assignments. If it's a dedicated machine with a decent amount of memory it would be an ideal candidate for P-1 (might even be able to get a free memory upgrade to make it P-1 ready from people scrapping their older machines). We probably have 10 times as many people doing TF as we need, and suggesting that anybody without a C2D/C2Q/i7 do trial factoring isn't exactly helping.

[lfm]

Originally Posted by Kevin

Why is TF always the default option for anything that can't run 50M tests in a reasonable amount of time?

I think some people want to see results turned in every day or something. Short attention spans! you know?

[Kevin]

Originally Posted by lfm

I think some people want to see results turned in every day or something. Short attention spans! you know?

That might explain why a lot of people choose to do TF, but it's not a valid reason for pushing them towards it. I mean, that computer would take less than a week running 24x7 to finish a P-1 test. It's not like the OP is a n00b that we have to worry about losing because of impatience. And even if he were, we should still present all options, let them know the relative amount of time each test would take, and let them know that more people doing TF isn't especially needed.

[lfm]

Originally Posted by Kevin

That might explain why a lot of people choose to do TF, but it's not a valid reason for pushing them towards it. I mean, that computer would take less than a week running 24x7 to finish a P-1 test. It's not like the OP is a n00b that we have to worry about losing because of impatience. And even if he were, we should still present all options, let them know the relative amount of time each test would take, and let them know that more people doing TF isn't especially needed.

I agree actually. I think he could do DCs productively at about 30 days or so. I think thats what I would do with that machine.

[Kevin]

Originally Posted by lfm

I agree actually. I think he could do DCs productively at about 30 days or so. I think thats what I would do with that machine.

Assuming I did my math right (Benchmark page shows someone with 71.89ms iterations on 1280k FFT with a 1.8GHZ P4, times ~22 million iterations for double checks), you get an estimate of 18.3 days.

[garo]

Yup! P-1 if you have over 300MB to spare otherwise doublechecks. P4s were not that great at TF if I remember correctly. If you do TF you'll get about 33% less credit per hour and do something that the project doesn't really need.

[henryzz]

Originally Posted by garo

Yup! P-1 if you have over 300MB to spare otherwise doublechecks. P4s were not that great at TF if I remember correctly. If you do TF you'll get about 33% less credit per hour and do something that the project doesn't really need.

you might get more actually as tf is overcredited

[joblack]

Thanks for the answers - I will use it to find primes ...