Random " % complete" ??
 

i just started getting what seems to be a random % complete message; see pic below. is this a problem? what if anything should i do?

thanks.

[IMG]http://i260.photobucket.com/albums/ii38/markg_2008/prime95.jpg[/IMG]

One thing you haven't told us is which version of Prime95 you are using but on a wild guess it is 25.x. Running two instances of factoring the same number on a pentium 4 would probably give similar results. Try giving a bit more info when asking that is pertinent to the question at hand.

nelson

[quote=Nelson;133651]One thing you haven't told us is which version of Prime95 you are using but on a wild guess it is 25.x. Running two instances of factoring the same number on a pentium 4 would probably give similar results. Try giving a bit more info when asking that is pertinent to the question at hand.

nelson[/quote]

i am using the current version, 24.14
if i was using anything OTHER than the current version, i would have mentioned it.

now that you know what version it is, and that i am not running two instances of factoring the same number, any ideas of whats wrong?

Nelson is just trying to help and since your first inquiry was not very informative...

The % complete numbers are not completely random : one part of them progress from 32,29% to 35,14%, the other from 38,39% to 41,38%...

Both progressed by a similar amount. The time interval between screen output seems to fluctuate quite a lot.

That interval depends on the number of iterations you set int the adhoc menu of Prime95 (Options / Preferences / Iterations between screen outputs) What is your setting there ?

Are you running heavy programs at the same time ? Is it a hyperthreaded machine ? Are you running more than one instance of Prime95 ? What happenned when the low numbers (32,29% to 35,14% and onwards) reached the starting points of the high numbers (38,39%) ?

Those are all questions that come to my mind.

It is a strange behaviour indeed...

Jacob

Thanks Kubik(CZ),

[quote=S485122;133677]The % complete numbers are not completely random : one part of them progress from 32,29% to 35,14%, the other from 38,39% to 41,38%...

Both progressed by a similar amount. The time interval between screen output seems to fluctuate quite a lot.

That interval depends on the number of iterations you set int the adhoc menu of Prime95 (Options / Preferences / Iterations between screen outputs) What is your setting there ?

Are you running heavy programs at the same time ? Is it a hyperthreaded machine ? Are you running more than one instance of Prime95 ? What happenned when the low numbers (32,29% to 35,14% and onwards) reached the starting points of the high numbers (38,39%) ?

Those are all questions that come to my mind.

It is a strange behaviour indeed...

Jacob[/quote]

That was what made me think it might be 25.x even though the window looked like 24.x. it still looks like 2 instances are running in the same window and on a hyperthreaded machine I don't know how 24.14 would do that unless the Ax option was being used, when I tried that I had two instances of Prime 95 with separate windows. Iteration speeds were comparable though.

mfg nelson

P.S. might just try stopping so that interim files are saved and exit. Then look in Prime95 directory and check how many versions of Px00000 and qx00000 are present more than one of each is a sure indicator of something buggy. if only the two are there restart and see if the behavior continues.

wnp

[quote=S485122;133677]
That interval depends on the number of iterations you set int the adhoc menu of Prime95 (Options / Preferences / Iterations between screen outputs) What is your setting there ?

Are you running heavy programs at the same time ? Is it a hyperthreaded machine ? Are you running more than one instance of Prime95 ? What happenned when the low numbers (32,29% to 35,14% and onwards) reached the starting points of the high numbers (38,39%) ?

[/quote]

everything was set to the default except user information and primenet options.

specifically; iterations is set to 10000 (default)
there is only one m and one p file
not running heavy programs; in fact this is the ONLY program running other than virus etc. background stuff
it is not hyperthreaded; 5 year old machine.
there seems to be two sets of numbers running PLUS another seemingly random numbers; one set in the 80's, one in the 90's, and a few turned up in the mid 70's (which seemingly already completed before it hit the 80s).

also, the time to complete the iterations seem strange; one string is high, the other low.

i figured i would let it run a bit and see if it finishes off, but it makes little sense. the other machines work fine.

latest screen shot follows

[IMG]http://i260.photobucket.com/albums/ii38/markg_2008/prime952.jpg[/IMG]

Gosh Mark,

I'm as stumped as you are. Is this by any chance your T40p mentioned in another thread actually the only important thing is why two processes at the same time. They are definintely factoring the same number. That still doesn't explain two instances and/or the wide discrepancy in timing.

There appear to be at least two cores in operation which doesn't make sense. AFAIK ver 24.14 must be started twice to get to simultaneous operation with the affinity parameter A- to define which core to use. The only other explanation with one core would be that one test stops while the other one takes over but that wouldn't account for outputs at exactly the same time even as much as 60 seconds apart.

Starting a second Prime95 instance results with a second tray icon and two separate windows. Even at that I would expect timings on two cores on the same number to be very similar if not exactly the same. One instance appears about to end so maybe it will correct itself but that may require some input from Prime95 himself. I can't figure out how to reproduce this situation with the current information. ???does worktodo.ini have two jobs or more in it or two with the same number or any thing so bizarre.

could you give a screen shot of Procexp or task manager I guess we'd need the processes tab with "Show processes from all users" box checked and window stretched vertically to show all active processes.

nelson

Mark,

It is a bit as if Prime95 was working from two start points at the same time in one instance. Did you try Menu / Test / Status to see the estimated endtime ? I think that exponent is to near to completion. Once it is finished, if it finishes at all, I Would save the result.txt, prime.log an wortodo.ini files. Remove all files, uninstall and reinstal the software. If you keep the local.ini and prime.ini I presume they hold the key to the explanation. You could post them here (if you whish you can remove the password from prime.ini : the line starting with "UserPWD=")

Jacob

Mark,

a real puzzle!

Are any files in prime95's folder readonly or with permissions that would stop prime95 writing to them? Just a thought - not sure if that would cause what you're seeing.

After you open prime95's window to see progress, is there indeed only one instance of prime95 in task manager?

I ask this because a few months ago I saw some occasional odd behaviour on a WinXP machine. Prime95 was starting as a service. Occasionally when I opened its window, a second instance would appear in task manager, with ~0% cpu. Using prime95's menu to exit would close the window & remove the new instance but leave the first running. When I killed the first instance from task manager, not only did it not save, but I found it had not saved any progress since the machine started... :sad:

I have no idea what caused it. It was only a few times, and I have not seen it recently, and I've been checking!

Mark

only one instance running; i had checked this.

and this was the same machine i had an issue with described on another thread; i deinstalled/reinstalled EVERYTHING; ie all new options files etc., then selected factoring only under configure primenet (the machine is kind of slow, so i figured this was better work)

the factoring (the one at issue) finished ok (according to the results file, but didnt give me credit on the primenet status page), started a new factor, found a factor almost immediately (did give me credit for this one), then started yet another which it is working on now.

here is the results page: (i xxxx out any personal data)

[SIZE=1][Wed May 07 10:52:33 2008][/SIZE]
[SIZE=1]Self-test 2560K passed![/SIZE]
[SIZE=1][Thu May 08 05:50:13 2008][/SIZE]
[SIZE=1]UID: markg, User: xxxxxxxxxxxxxxxxxx[/SIZE]
[SIZE=1]Change UID from S818510 to markg[/SIZE]
[SIZE=1][Tue May 13 12:55:40 2008][/SIZE]
[SIZE=1]UID: markg/IBM-T40p, M48210497 no factor to 2^69, Wc1: CED140D7[/SIZE]
[SIZE=1][Tue May 20 15:48:48 2008][/SIZE]
[SIZE=1]UID: markg/IBM-T40p, M48486617 no factor to 2^69, Wc1: D16542AB[/SIZE]
[SIZE=1][Tue May 20 17:14:59 2008][/SIZE]
[SIZE=1]UID: markg/IBM-T40p, M48491761 has a factor: 21780543116224729279[/SIZE]


prime.ini:

[SIZE=1]UsePrimenet=1[/SIZE]
[SIZE=1]DialUp=0[/SIZE]
[SIZE=1]DaysOfWork=1[/SIZE]
[SIZE=1]WorkPreference=1[/SIZE]
[SIZE=1]AskedAboutMemory=1[/SIZE]
[SIZE=1]OldUserID=[/SIZE]
[SIZE=1]OldUserPWD=[/SIZE]
[SIZE=1]UserID=markg[/SIZE]
[SIZE=1]UserPWD=xxxxxxxx[/SIZE]
[SIZE=1]UserName=xxxxxxxxxx[/SIZE]
[SIZE=1]UserEmailAddr=xxxxxxxxxxx[/SIZE]
[SIZE=1]Newsletters=0[/SIZE]
[SIZE=1]OutputIterations=10000[/SIZE]
[SIZE=1]ResultsFileIterations=999999999[/SIZE]
[SIZE=1]DiskWriteTime=30[/SIZE]
[SIZE=1]NetworkRetryTime=2[/SIZE]
[SIZE=1]NetworkRetryTime2=60[/SIZE]
[SIZE=1]DaysBetweenCheckins=28[/SIZE]
[SIZE=1]TwoBackupFiles=1[/SIZE]
[SIZE=1]SilentVictory=0[/SIZE]
[SIZE=1]Left=811[/SIZE]
[SIZE=1]Top=515[/SIZE]
[SIZE=1]Right=1395[/SIZE]
[SIZE=1]Bottom=803[/SIZE]
[SIZE=1]Windows95Service=1[/SIZE]

local.ini

[SIZE=1]OldCpuType=12[/SIZE]
[SIZE=1]OldCpuSpeed=598[/SIZE]
[SIZE=1]CPUHours=20[/SIZE]
[SIZE=1]DayMemory=64[/SIZE]
[SIZE=1]NightMemory=256[/SIZE]
[SIZE=1]DayStartTime=450[/SIZE]
[SIZE=1]DayEndTime=1410[/SIZE]
[SIZE=1]Affinity=99[/SIZE]
[SIZE=1]ComputerID=IBM-T40p[/SIZE]
[SIZE=1]LastEndDatesSent=1210681625[/SIZE]
[SIZE=1]RollingStartTime=0[/SIZE]
[SIZE=1]SelfTest2560Passed=1[/SIZE]
[SIZE=1]RunOnBattery=0[/SIZE]
[SIZE=1]RollingAverage=1000[/SIZE]
[SIZE=1]VacationEnd=0[/SIZE]
[SIZE=1]VacationOn=1[/SIZE]

I see one funny item in your local.ini :
Affinity=99
You can remove that line.

Not getting credit for your exponent 48486617 is strange... You can check if your result shows up on the V5 server in the comming days : [url]http://v5www.mersenne.org/report_factoring_effort/?exp_lo=48486617&exp_hi=&bits_lo=0&bits_hi=999&B1=Get+Data[/url]

Since you are using Prime95 v24.14 I don't think George Woltman will be interested in debugging what happened.

Jacob

[quote=S485122;133819]I see one funny item in your local.ini :
Affinity=99
You can remove that line.

Not getting credit for your exponent 48486617 is strange... You can check if your result shows up on the V5 server in the comming days : [URL]http://v5www.mersenne.org/report_factoring_effort/?exp_lo=48486617&exp_hi=&bits_lo=0&bits_hi=999&B1=Get+Data[/URL]

Since you are using Prime95 v24.14 I don't think George Woltman will be interested in debugging what happened.

Jacob[/quote]

its been running fine since it finished the prior factor;
i removed the affinity line (must have been put there by the sw)
still no credit; not the end of the world, but it did show up in the status file, so i dont get this

[quote=markg;133764]
there is only one m and one p file
[/quote]
isn't that a bit unusual? i.e. factoring and LL is running at the same time?

(maybe not unusual if things are done according to load / day/ night time or other settings, but my machines always did one after the other ; I don't remember having had a p and an m file at the same time.)

PS: and since you have [SIZE=1]TwoBackupFiles=1, whenever there's a p*** there should be a q***. (not later than 30 mins after the p*** started existing)

[/SIZE]

[quote=markg;133999]its been running fine since it finished the prior factor;
[U]i removed the affinity line (must have been put there by the sw)[/U]
still no credit; not the end of the world, but it did show up in the status file, so i dont get this[/quote]

:blush:yes I think that may have been my bad the context for the affinity
switch is -An not A-. A-1 would convert to 99 being the complement of 1
(if you add 1 to 99 the result is 0 with an overflow when your calculator
has a fixed number of 2 digits. Digital calculators just switch to floating
point notation so most people never see it.
The overflow as the result of a subtraction which is commonly used in
binary arithmetic is expected and not an error and adds to the sign bit
changing to 0 in binary resulting in a positive number and 1 without an
overflow resulting in a negative number. In the case of base 100, 99
is the largest number and -100 is the smallest that can be repesented in
some arbitrary Accumulator with 2 digits) Happens a lot with a digital
ohmmeter, just can't be used for arithmetic

:showoff:
.. 02 base 100 complement. 02 because the number is positive
. -01 base 100 complement. 99 because the number is negative
add
= 01 ......................... =(1)01 the overflowing 1 is lost but not
unexpected a test for overflow indicates we dont have to complement
the result to get back to the correct result.

.. 01 base 100 complement 01 because the number is positive
. -02 base 100 complement 98 because the number is negative
add
=-02 ....................... =(0)99 because there was no overflow it
must be complemented resulting in 100 -99 = 1 the negative result
. -01 :showoff:

A simpler way of looking at it would be just subtract 1 from 100 equals 99.

the process is simpler in binary but a programmer has to be careful to test
for invalid situations probably Prime95 didn't expect such an error to occur
and didn't put a check in for that or just dealt with it by setting the
number size to 2 digits not expecting a machine with more than 100 or
more likely 16 CPU's to be running his program. And guess who stepped
into that doo-doo yours truly who should have known better. I'm not sure
it was the cause of your problem because you had the problem prior to my
erroneous posting.

You can also use negative numbers in your memory settings and prime 95
will try to use all memory available except the specified negative amount.
Meaning that all other programms including the OS are expected to be
able to run within that much memory. helps keep the hard drive from
thrashing due to insufficient memory.

Information on start parameters and other settings can be found in the help window.

If you are interested on more information on Complement arithmetic
googling on "Two's Complement" comes directly to the wikipedia entry and
the wiki(hawaiian for quick/short)-pedia is an excellent source of
information but not always quick.

The above may be more information than you are interested in but may be
of some use to others looking in.

:tu:Glad the program is working correctly now. Are you looking at your
personal status page? or the overall status page? Your personal status
page will show as uncleared somewhere near the bottom until some preset
date when synchronisation is done. The overall status will probably
be updated at that time too.

How to get to your personal status page can be found on Prime Net
pagelink of mersenne.org

nelson

Affinity=99 has nothing to do with the problem. Affinity 99 means "run on any CPU" of a multi-core machine.

Two backup files only applies to LL testing. Trial factoring never generates a q file

[quote=Nelson;134071]Glad the program is working correctly now. Are you looking at your
personal status page? or the overall status page? Your personal status
page will show as uncleared somewhere near the bottom until some preset
date when synchronisation is done. The overall status will probably
be updated at that time too.

[/quote]

im looking at both my personal status and the overall status.

it may be that im confused as to how this works; i have only been doing LL testing until recently, and each one of those shows up when completed.

however, i changed the (offending) laptop over to factoring, and originally thought that all results would show some record; it finished 4 factoring jobs so far, but only one showed up in my personal page.

but the one that did show up FOUND a factor, the other three didnt, so perhaps thats what is supposed to happen? so if you dont find a factor there is no (obvious) credit/record of the work on the personal status page?

When the program finishes trial factoring an exponent you get credit, when the program finds a factor you get a bit more credit (depending on the number of bits of the found exponent) and your factor found count goes up by one.

What you saw is normal behaviour.

Jacob (who should have looked up the meaning of affinity=99)

[quote=m_f_h;134018]isn't that a bit unusual? i.e. factoring and LL is running at the same time?

(maybe not unusual if things are done according to load / day/ night time or other settings, but my machines always did one after the other ; I don't remember having had a p and an m file at the same time.)

PS: and since you have [SIZE=1]TwoBackupFiles=1, whenever there's a p*** there should be a q***. (not later than 30 mins after the p*** started existing)[/SIZE]

[/quote]

could be a remnant; it was originally doing LL tests, started crashing continuously, so i deinstalled/reinstalled and kept the files that i wasnt sure about (erasing the config files though since the original remove/reinstall didnt help). when i was done i told it to do only factoring.