Self Test going on way too long HELP HELP HELP

Jwb52z · 2003-01-13, 01:35

The Help files with the program say that there are 16 self tests. I just transferred my Prime 95 directory to my new faster computer and now it is doing the self test again, but there are WAY OVER 16 tests. It's over 200 now. What's wrong? My new computer is a 2.4 GHz laptop. I am running the P4 version 22.13 of the software. Help!!!

outlnder · 2003-01-13, 01:40

Your computer must do a self test for each FFT range. It will do as many tests as it takes in ONE HOUR.

Simply put, your machine will run a self test for one hour when it encounters a new FFT range.

Deamiter · 2003-01-13, 01:53

just curious, does that mean that there are 16 tests in each range that are looped for an hour, or 16 ranges with X tests looped for an hour?

Jwb52z · 2003-01-13, 02:00

I never would have believed GIMPS could run this fast, but on my old machine each iteration used to take about 0.2 or 0.3 seconds. now it takes 0.028 seconds. This is incredible.

outlnder · 2003-01-13, 10:18

It tests the 16 FFT's in a loop.

eepiccolo · 2003-01-13, 13:27

I think I'm a little lost here. Are there 16 different FFT lengths that the program tests, one hour each, in effect reapeating every 16 hours? Or are there 16 different exponents that are tested in each FFT length?

Deamiter · 2003-01-13, 17:29

woah. We're just talking about the self-test it does for each FFT range. There's 16 different exponents and it goes through 400 (or something) iterations of each. When the test finishes all 16, it loops and does them again until an hour is completed. The point is for it to try to make your computer screw up in that hour which would tip you off to a problem in that range.

eepiccolo · 2003-01-13, 17:51

Along the lines of torture tests, will running the test find errors in hardware other than the CPU or memory? And what kind of problems in the CPU or memory will cause Prime95 errors?

Prime95 · 2003-01-13, 19:40

What we have here is a help file bug. Where does it talk about "16 self tests"? I need to know to fix it.

Older prime95 versions had 3 testing modes: the one hour self test when testing an exponent, a menu choice to run all 16 one hour self tests, and a torture test menu choice which continuously ran tests on all 16 FFT sizes.

Later prime95 versions dropped the "run all 16 one hour tests" menu choice. It seemed pointless - run the torture test instead.

Also, the torture test now tests many more than 16 different FFT sizes.

Jwb52z · 2003-01-14, 01:45

It's in the program's help files for the newest section in the program under the version info links. I hope what I am saying makes enough sense for you to know what I mean.

2003-01-13, 01:35	#1
Jwb52z Sep 2002 863 Posts	Self Test going on way too long HELP HELP HELP The Help files with the program say that there are 16 self tests. I just transferred my Prime 95 directory to my new faster computer and now it is doing the self test again, but there are WAY OVER 16 tests. It's over 200 now. What's wrong? My new computer is a 2.4 GHz laptop. I am running the P4 version 22.13 of the software. Help!!!

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
very long int	davar55	Lounge	60	2013-07-30 20:26
Using long long's in Mingw with 32-bit Windows XP	grandpascorpion	Programming	7	2009-10-04 12:13
I think it's gonna be a long, long time	panic	Hardware	9	2009-09-11 05:11
How long is too long?	schickel	Lounge	2	2009-02-22 12:31
How long to use the Prime95 torture test will prove that the system is rock stable ?	Unregistered	Hardware	4	2004-09-09 07:08

2003-01-13, 01:40	#2
outlnder Aug 2002 2·3·53 Posts	Your computer must do a self test for each FFT range. It will do as many tests as it takes in ONE HOUR. Simply put, your machine will run a self test for one hour when it encounters a new FFT range.

2003-01-13, 01:53	#3
Deamiter Sep 2002 165₈ Posts	just curious, does that mean that there are 16 tests in each range that are looped for an hour, or 16 ranges with X tests looped for an hour?

2003-01-13, 02:00	#4
Jwb52z Sep 2002 1101011111₂ Posts	I never would have believed GIMPS could run this fast, but on my old machine each iteration used to take about 0.2 or 0.3 seconds. now it takes 0.028 seconds. This is incredible.

2003-01-13, 10:18	#5
outlnder Aug 2002 2·3·53 Posts	It tests the 16 FFT's in a loop.

2003-01-13, 13:27	#6
eepiccolo Dec 2002 Frederick County, MD 172₁₆ Posts	I think I'm a little lost here. Are there 16 different FFT lengths that the program tests, one hour each, in effect reapeating every 16 hours? Or are there 16 different exponents that are tested in each FFT length?

2003-01-13, 17:29	#7
Deamiter Sep 2002 165₈ Posts	woah. We're just talking about the self-test it does for each FFT range. There's 16 different exponents and it goes through 400 (or something) iterations of each. When the test finishes all 16, it loops and does them again until an hour is completed. The point is for it to try to make your computer screw up in that hour which would tip you off to a problem in that range.

2003-01-13, 17:51	#8
eepiccolo Dec 2002 Frederick County, MD 2·5·37 Posts	Along the lines of torture tests, will running the test find errors in hardware other than the CPU or memory? And what kind of problems in the CPU or memory will cause Prime95 errors?

2003-01-13, 19:40	#9
Prime95 P90 years forever! Aug 2002 Yeehaw, FL 17·487 Posts	What we have here is a help file bug. Where does it talk about "16 self tests"? I need to know to fix it. Older prime95 versions had 3 testing modes: the one hour self test when testing an exponent, a menu choice to run all 16 one hour self tests, and a torture test menu choice which continuously ran tests on all 16 FFT sizes. Later prime95 versions dropped the "run all 16 one hour tests" menu choice. It seemed pointless - run the torture test instead. Also, the torture test now tests many more than 16 different FFT sizes.

2003-01-14, 01:45	#10
Jwb52z Sep 2002 863 Posts	It's in the program's help files for the newest section in the program under the version info links. I hope what I am saying makes enough sense for you to know what I mean.