USERID changing to blank problem
 

This weekend I experienced this problem on one of the Linux machines. The results were returned as ANONYMOUS but the CPU Name was still correct. Is there a way to relink the results, or should we just email our results file to
George?

[QUOTE=S485122;149636]Which OS are the computers experiencing that problem using ?
[/QUOTE]

PIII running Win XP (SP3), Prime95 located in C:\Prime\.

[quote=S485122;149636]Which OS are the computers experiencing that problem using ?[/quote]
I have 3 machines, and all 3 were affected. Two of them are P4's running XPSP3 (32 bit). The other, my quad-core machine, is running XP64 + SP2. In all 3 machines Prime95 was installed in C:\Program Files\Prime95. (The quad-core machine is running the 64-bit version of Prime95.)


-FES

[QUOTE=fes016;149623]However, on the CPU page I no longer have the option to drop the CPU -- there is no checkbox next to the name. (I don't want to drop the CPU anyway, but there used to be a checkbox there.) Also, I can't click on the name to see its properties, like I used to.[/QUOTE]

I changed this yesterday. The CPU description and the CPU properties page were meaningless for v4_computers. v4_computers do not send their properties to the v5 server and even if they did how could I display the many conflicting v4_computers' properties on one web page.

[quote=S485122;149636]Which OS are the computers experiencing that problem using ?[/quote]

Ubuntu 8.10/32

[QUOTE=Prime95;149567]By forcing clients to factor further, I'm reducing the HTTP workload on the v5 server.[/QUOTE]

But on the other hand, factoring to low limits was a work type very useful for old machines (e.g. AMD Thunderbird and/or PIII type, that were very good at TFing numbers to 2^62). Would it be too hard to allow this lower limits to be assigned to machines identified as being below a certain standard?

[QUOTE=lycorn;149726]But on the other hand, factoring to low limits was a work type very useful for old machines (e.g. AMD Thunderbird and/or PIII type, that were very good at TFing numbers to 2^62). Would it be too hard to allow this lower limits to be assigned to machines identified as being below a certain standard?[/QUOTE]

Even a 486 should be able to rip through trial factoring an exponent around 300 million to 2^64.

You do bring up a good point. LMH is not a good place for a fancy Core2 or Phenom machine.

[QUOTE=lycorn;149726]But on the other hand, factoring to low limits was a work type very useful for old machines (e.g. AMD Thunderbird and/or PIII type, that were very good at TFing numbers to 2^62). Would it be too hard to allow this lower limits to be assigned to machines identified as being below a certain standard?[/QUOTE]

I had a PIII 866 Mhz that since v5 was getting assigned TF from 58-61 bits.
It was finishing each assignment in 2 minutes ... over 700 a day.
While it was cool to see my lowly PIII finishing 2,000 completions after 3 days I had to admit the assignments were too small.

A couple days ago that was changed to TF from 58-64 bits.
It is now finishing in just under 30 minutes with results reported at 58-62, 63 and 64 bits.
To me that is a more reasonable assignment.
I wouldn't have a problem going another bit or two. It should still finish in a few hours.
I'm guessing that even a 100 Mhz machine wouldn't take more than a day to 64 bits.

I suspect there are not many machines below that level on GIMPS, mind you, there was another thread where someone asked if Prime95 would run on an Intel 486.

Considering LL tests are now even taking the newest hardware weeks or months I don't think spending hours or a day or two factoring is excessive.

[QUOTE=Prime95;149689]Right now I'm reprocessing all the data that has gone through the v4 bridge. This will tell me if any data was not processed.
[/QUOTE]

Thanks for the update.

Not sure if this test case will help but as I suggested in another post according to my ciphering several results completed between Oct 20-27 are NOT accounted for:

If I cut off this list at the right place it should account for the difference of about 100 points between the total on the summary report and the total off all my v5 results listed.

[CODE]v4_computers 50699557 NF 2008-10-27 19:51 0.0 [UNASSIGNED] 2.2846
v4_computers 50700259 NF 2008-10-27 12:51 0.0 no factor to 2^69 2.2846
v4_computers 50700233 NF 2008-10-27 12:51 0.0 no factor to 2^69 2.2846
v4_computers 50700229 NF 2008-10-27 12:51 0.0 no factor to 2^69 2.2846
v4_computers 50700217 NF 2008-10-27 12:51 0.0 no factor to 2^69 2.2846
v4_computers 50092069 NF 2008-10-27 08:46 0.0 no factor to 2^69 2.3123
v4_computers 50700197 NF 2008-10-27 05:51 0.0 no factor to 2^69 2.2846
v4_computers 50643877 NF 2008-10-27 05:51 0.0 no factor to 2^69 2.2133
v4_computers 50699533 NF 2008-10-27 04:41 0.0 no factor to 2^69 2.2846
v4_computers 50699501 NF 2008-10-27 04:41 0.0 no factor to 2^69 2.2846
v4_computers 50699483 NF 2008-10-27 04:41 0.0 no factor to 2^69 2.2846
v4_computers 50699443 NF 2008-10-27 04:41 0.0 no factor to 2^69 2.2846
v4_computers 42286817 C 2008-10-27 04:41 0.0 380B3599FA8513__ 73.4146[/CODE]

What happened to my V4 LL assignment?
 

[quote=fes016;149623]
{snip}

My User Summary page shows the correct number of assignments (36), but the Assignment Details page does not (only 31). The assignments under v4_computers (4) are accounted for; it's on my quad-core system where the discrepancy occurs. If I go to the Computer Properties page for my quad-core system, it shows 32 work units assigned. My worktodo.txt file also shows 32 exponents. The assignment details page only shows 27. I haven't yet determined which exponents are missing.

{snip}

I realize that TPTB on GIMPS are working hard and doing their best during the V5 transition, but I'm frustrated with all of the glitches that are going on. And I only have three computers to manage; I can't imagine how some of you with scores of computers are coping.


-FES[/quote]

In a similar vein, I am a little worried. My V4 computer, running v24, which I don't have much access to (about once every 6 months, it's my Dad's who is in his 70s), was nicely working away, but now the Assignment Details page no longer has the next assignment it would have/will work on next. The Exponent Status page reports it has been assigned to someone else, see attachment on posting that follows. But it still shows as me having it included in the Account Summary page, see attachment on posting that follows.

The question is, Does my remote V4 client know the exponent has been re-assigned, or is it going to try the LL only to find someone else has done it/is doing it?

The exponent is the 1st in the red box on the attached old screen grab.

Can someone help me?
T.I.A.

Attachment for posting above
 

Attachment for posting above

2nd Attachment for posting above
 

2nd Attachment for posting above

Looking much better....

Before you read on ... I'm not trying to be picky or trivial but I hope that by identifying what I notice I am helping give symptoms that can help pinpoint problems that likely most of you are having.

On the plus side:
- My "Stats for the Last 365 Days" looks right now
- My lost v4 Assignments have returned
- The Top Producers reports now include the V4 results
- My team results now include "most" of my results including v4, v5 and the transition week (see next section)

Still not quite there:
- The completed v4 assignments are still listed
- My "Lifetime Completed by Result-Type" is still missing the completions listed in post #114
- My completions for the period between when the userid blanked (a couple days ago) and was fixed by me are NOT showing up and I assume NOT credited.
- My team total is about 200 credits below my account total (and 100 below my Lifetime completed) ... not sure I can be sure what is missing and/or possibly double counted because the #of results and #of credits are different in more than 1 row.

Thanks for the progress

Thank you Scott for "fixing" (i.e. ".") the V5 server's user id code. I have logged on and linked my V4 user id. My LL and TF V4 credit in P90 years matches the values on Team Prime Rib's "frozen" 10/20/08 status pages so it appears that my V4 credit was successfully retrieved from anonymous.

One of the nicest things about participating in GIMPS is the fact that George and you pay attention to your users. Thanks again.
Roger

CPU Duplicated....
 

From my CPU Report:

[QUOTE=petrw1;149874]From my CPU Report:[/QUOTE]

Just a thought ... I had to reinstall XP on this PC on Monday but I did save the entire Prime95 directory before hand and restore it after the reinstall.

[QUOTE=petrw1;149877]Just a thought ... I had to reinstall XP on this PC on Monday but I did save the entire Prime95 directory before hand and restore it after the reinstall.[/QUOTE]

The client generates a unique identifier for each computer. Some Windows ID goes into generating the computer ID. I'll bet the Windows reinstall caused the client to generate a new computer ID.

[QUOTE=Prime95;149886]The client generates a unique identifier for each computer. Some Windows ID goes into generating the computer ID. I'll bet the Windows reinstall caused the client to generate a new computer ID.[/QUOTE]

Another curiosity is that both assignments shown on the CPU report show as complete by this same CPU in the last day.

My double-check exponent 19050917 was credited to Anonymous on 11/16 while my user name was blanked out at work (computer name: work-new). It is still credited to Anonymous and has not changed with any of the fixes.

Also much of my V4 work was double checking and is credited as first-time testing.

Version 5 only sends work to one of four cores
 

Before loading the latest version (5) I was running four instances of prime95, one for each core of a quad. I was able to see in 'performance' on the task manager that all four cores went to 100% busy when load testing.
After I installed 5, I noticed that when I torture test, all cores go 100% busy. When I am doing prime finding work only cpu3 is 100% busy. The other cores are doing nothing. I can play games and so on.

CPU Report Ghz-Days Discrepancy for 1 CPU
 

On report [url]http://www.mersenne.org/cpus/[/url] there is a column labelled GHz-Days

For every computer this number seems to be quite accurate EXCEPT Dungeon which is VERY LOW ... it show 0.1509 and if I add up all my results I get closer to 6.0. I can't find a subset of results that adds up to 0.1509 but then I believe it has been 0.1509 almost from the day I upgraded to v5 on Oct 27.

[QUOTE=petrw1;150052]On report [url]http://www.mersenne.org/cpus/[/url] there is a column labelled GHz-Days

For every computer this number seems to be quite accurate EXCEPT Dungeon which is VERY LOW ... it show 0.1509 and if I add up all my results I get closer to 6.0. I can't find a subset of results that adds up to 0.1509 but then I believe it has been 0.1509 almost from the day I upgraded to v5 on Oct 27.[/QUOTE]

Follow up....this CPU completes about 50 2^58 to 2^64 tests per day. It reports in at 62, 63 and 64 bits with credits of .0022, .0029 and .0055 credits respectively, for a total of .0106 credits per TF.

In the past day this CPU received .0029 credits: One 63-64 bit TF.

I am unable to change my password on the v5 server. Is there a special place to do this? I've tried going to "My Account"->"Update Settings", and then entering a new password at the top of that page twice, but I just get an error message at the top of the screen saying, "Password incorrect on existing account - please re-enter.".

[QUOTE=WraithX;150348]I am unable to change my password on the v5 server. Is there a special place to do this? [/QUOTE]

There is no way to do this presently

The internal links have all disappeared
 

I think it must be a JS thing, but the home page of mersenne.org has no way for me to navigate about the place.

Can this behaviour please be changed. The <noscript> tag should do the trick nicely.

Please don't say to me "just enable JS" because that is not an option for me, office policy will not allow it.

hmm, I can navigate and login without Java & JS with Mozilla.

[QUOTE=Phantomas;150368]hmm, I can navigate and login without Java & JS with Mozilla.[/QUOTE]There are a lot of links hidden without JS. I can see them in the source code:[code]<script language="javascript" type="text/javascript">
objTreeMenu_1 = new TreeMenu("/TMimagesXL", "objTreeMenu_1", "_self", "", true);
newNode = objTreeMenu_1.addItem(new TreeNode('My Account', null, '/account/', false, true, 'tmenu0textMyAccount'));
newNode_1 = newNode.addItem(new TreeNode('Summary', null, '/account/', false, true, 'tmenu1textMyAccount'));
newNode_1.iconImageWidth=20;
newNode_1.iconImageHeight=20;
newNode_1.lineImageWidth=15;
newNode_1.lineImageHeight=12;
newNode_2 = newNode.addItem(new TreeNode('Update Settings', null, '/update/', false, true, 'tmenu1textMyAccount'));
newNode_2.iconImageWidth=20;
newNode_2.iconImageHeight=20;
newNode_2.lineImageWidth=15;
newNode_2.lineImageHeight=12;
newNode_3 = newNode.addItem(new TreeNode('CPUs', null, '/cpus/', false, true, 'tmenu1textMyAccount'));
newNode_3.iconImageWidth=20;
newNode_3.iconImageHeight=20;
newNode_3.lineImageWidth=15;
newNode_3.lineImageHeight=12;
newNode_4 = newNode.addItem(new TreeNode('Assignments', null, '/workload/', false, true, 'tmenu1textMyAccount'));
... <snip about 250 lines> ...
newNode_8 = newNode.addItem(new TreeNode('Legal', null, '/legal/', false, true, 'tmenu1textAbout'));
newNode_8.iconImageWidth=20;
newNode_8.iconImageHeight=20;
newNode_8.lineImageWidth=15;
newNode_8.lineImageHeight=12;
newNode.iconImageWidth=20;
newNode.iconImageHeight=20;
newNode.lineImageWidth=15;
newNode.lineImageHeight=12;

objTreeMenu_1.drawMenu();
objTreeMenu_1.resetBranches();
</script>[/code]Currently I have to manually type each link into the address bar, annoying!

Neither the links {My Account, My Team, PrimeNet Summary, Top Producers, Top Teams, Progress Reports, Results Queries, Manual Testing, About Gimps} nor their sub-links appear in the left pane in Seamonkey without Javascript.

Yes, that's true for [URL]http://v5www.mersenne.org/account/?user_login=[/URL]...... But retina asked for the home page of mersenne.org, and [URL="http://www.mersenne.org"]www.mersenne.org[/URL] is operable without JS for me.


addendum: ok, I'm complete wrong. You'r right, dosen't work without JS...
sorry for the confusion....

I've just noticed that v5 gives credit to double-checks even when they disagree with the result of the first-time check.

I guess credit isn't taken away until a triple-check discredits one of the first two tests?

Stats being mis-reported
 

I can well imagine that TPTB on GIMPS have many things to do, but since the fix for the Anonymous bug my stats have been misreported. If you look at the attachment you can see that for the last 365 days that the LL slice of the pie chart is shown as almost the same as the V4-LL, but the figures are quite different.

Also the number of TF results is wildly different for lifetime and 365 days, when they should be the same. Anyone else got this kind of problem?

Clearly the integrity of the database is of the utmost importance, but it would also be nice to have the stats working as well.

M.

[QUOTE=g0ods;150434]
Also the number of TF results is wildly different for lifetime and 365 days, when they should be the same. Anyone else got this kind of problem?[/QUOTE]

Nearly everyone should be similarly affected. I know what happened and how to correct it. I've been working on some other matters.

[QUOTE=Prime95;150438]Nearly everyone should be similarly affected. I know what happened and how to correct it. I've been working on some other matters.[/QUOTE]

I thought you would have it covered, as ever there just aren't enough hours in the day. Thanks for your hard work and also your prompt reply.:tu:

[QUOTE=petrw1;150122]Follow up....this CPU completes about 50 2^58 to 2^64 tests per day. It reports in at 62, 63 and 64 bits with credits of .0022, .0029 and .0055 credits respectively, for a total of .0106 credits per TF.

In the past day this CPU received .0029 credits: One 63-64 bit TF.[/QUOTE]

And now on my new Quad on which one core is doing TF ... in the last 16 hours it shows credit on the CPU report equal to only one completion (0.0735) even though it has completed 26 for a total of 1.6901.

On the CPU Details:
[QUOTE]GHz-days credited 0.0735 GHz-days
Work units completed 26 (in the last 365 days) [/QUOTE]

However, it appears that all the credit is showing up on the Account Totals reports.

Looking for work that uses less memory. Huh?
 

[CODE][Nov 24 19:53] Worker starting
[Nov 24 19:53] Setting affinity to run worker on logical CPU #1
[Nov 24 19:53] Optimal P-1 factoring of M47710997 using up to 256MB of memory.
[Nov 24 19:53] Assuming no factors below 2^69 and 2 primality tests saved if a factor is found.
[Nov 24 19:53] Optimal bounds are B1=530000, B2=9672500
[Nov 24 19:53] Chance of finding a factor is an estimated 5.33%
[Nov 24 19:53] Using FFT length 2560K
[B][U][Nov 24 19:53] Exceeded limit on number of workers that can use lots of memory.
[Nov 24 19:53] Looking for work that uses less memory.[/U][/B]
[Nov 24 19:53] Optimal P-1 factoring of M47726677 using up to 256MB of memory.
[Nov 24 19:53] Assuming no factors below 2^70 and 2 primality tests saved if a factor is found.
[Nov 24 19:53] Optimal bounds are B1=505000, B2=8585000
[Nov 24 19:53] Chance of finding a factor is an estimated 4.67%
[Nov 24 19:53] Using FFT length 2560K
[Nov 24 19:53] M47726677 stage 1 is 7.49% complete.[/CODE]

I have a Q9550
4 Gb RAM
Vista 64

I thought with that much RAM I would be home free so why the messages in bold. I allocated 256M in the GIMPS Setup.

I have 3 cores doing LL in the 47M Range and 1 doing TF.
When 2 of the 3 LL cores finished Phase 1 of P-1 they generated the messages above and started Phase 1 of P-1 on another exponent while 1 core went into Phase 2.

[QUOTE=petrw1;150599]
I have a Q9550
4 Gb RAM

I thought with that much RAM I would be home free so why the messages in bold. I allocated 256M in the GIMPS Setup.[/QUOTE]

If you want prime95 to use more of that 4GB (and avoid the bold messages) then let prime95 use more than 256MB of memory.

[QUOTE=Prime95;150600]If you want prime95 to use more of that 4GB (and avoid the bold messages) then let prime95 use more than 256MB of memory.[/QUOTE]

Thanks
I had assumed (in error) that the 256 was for each worker, not in total. As soon as I changed it to 1024 then 1600 it went back to the Phase 2 of the previous P-1 tests. I hope it didn't lose the Phase 1 progress of the test just abandoned (temporarily, I hope)

Oh and it also ignored the optimal B1/B2 setting and used the smaller values from the save file, probably back when it was trying to manage with 256 Mb of RAM.

[B][Nov 24 19:53] Worker starting
[Nov 24 19:53] Setting affinity to run worker on logical CPU #2
[Nov 24 19:53] Optimal P-1 factoring of M47711143 using up to 256MB of memory.
[Nov 24 19:53] Assuming no factors below 2^69 and 2 primality tests saved if a factor is found.
[Nov 24 19:53] Optimal bounds are B1=530000, B2=9672500
[Nov 24 19:53] Chance of finding a factor is an estimated 5.33%
[Nov 24 19:53] Using FFT length 2560K
[Nov 24 19:53] Exceeded limit on number of workers that can use lots of memory.
[Nov 24 19:53] Looking for work that uses less memory.
[Nov 24 19:53] Optimal P-1 factoring of M47726717 using up to 256MB of memory.
[Nov 24 19:53] Assuming no factors below 2^69 and 2 primality tests saved if a factor is found.
[Nov 24 19:53] Optimal bounds are B1=530000, B2=9672500
[Nov 24 19:53] Chance of finding a factor is an estimated 5.33%
[Nov 24 19:53] Using FFT length 2560K
[Nov 24 19:53] M47726717 stage 1 is 50.33% complete.
[Nov 24 21:55] M47726717 stage 1 is 66.67% complete. Time: 7322.365 sec.
[Nov 24 23:00] Restarting worker with new memory settings.
[Nov 24 23:00] Optimal P-1 factoring of M47711143 using up to 1024MB of memory.
[Nov 24 23:00] Assuming no factors below 2^69 and 2 primality tests saved if a factor is found.
[Nov 24 23:00] Optimal bounds are B1=560000, B2=14980000
[Nov 24 23:00] Chance of finding a factor is an estimated 5.82%
[Nov 24 23:00] Using FFT length 2560K
[Nov 24 23:00] Ignoring suggested B1 value, using B1=530000 from the save file
[Nov 24 23:00] Ignoring suggested B2 value, using B2=9672500 from the save file
[Nov 24 23:01] Available memory is 419MB.
[Nov 24 23:01] Using 414MB of memory. Processing 15 relative primes (45 of 480 already processed).
[Nov 24 23:01] M47711143 stage 2 is 9.46% complete.[/B]

Stats fixed!
 

The stats in the user summary are much better now! The v4 TF & v4 LL counts and credit issues look to be sorted.

George and Scott - thank you - nice work with the new server. I know there's "a few" features to add, tweak or fix, but it's still quite an achievement.

error 7 invalid character
 

Hello,

oi am new to this forum. I read the thread but i did'nt find an answer. My questeion, prime V25 tells me there is an invalid character in my User ID. In my user id i have a dot "." xxx.xxxxx. I do'nt get in contact with my prime account. What can i do to solve ist.
I linked my V4 account to V5.

Sorry when the answer is in this thread or in another tzhread. Then i was to dumb to fiind.

May somebody help?

best wishes
Axel

[QUOTE=markr;150620]George and Scott - thank you - nice work with the new server. I know there's "a few" features to add, tweak or fix, but it's still quite an achievement.[/QUOTE]

Here, Here.

ECMF top producers page
 

I would like to refer to the attachment in g0ods post above (#150).
[URL]http://www.mersenneforum.org/attachment.php?attachmentid=2948&d=1227485959[/URL]

It looks like he (or she) has a ranking for the ECMF worktype (22 of 34). However, I am unable to view the top producers page for ECMF work. I can't find the link in the sidebar in the GIMPS homepage.

I'm rather new here and am not really familiar with ECM. Would someone be kind enough to explain to me why do we need to do ECM? Aren't we already sure that there won't be any primes among them? In particular, Mersenne primes?

[QUOTE=Axel1;150628]
oi am new to this forum. My questeion, prime V25 tells me there is an invalid character in my User ID. In my user id i have a dot "." xxx.xxxxx.[/QUOTE]

Welcome! I've fixed a bug - it should work now.

[QUOTE=hockmeng;150641]However, I am unable to view the top producers page for ECMF work. I can't find the link in the sidebar in the GIMPS homepage.

I'm rather new here and am not really familiar with ECM. Would someone be kind enough to explain to me why do we need to do ECM? Aren't we already sure that there won't be any primes among them? [/QUOTE]

Click any of the top producer reports. Then click customize. You can get the ECMF report from there.

You are correct, we know these numbers are not prime. ECM tries to find factors. We do this simply for the fun of it.

Hi Prime95,

"Welcome! I've fixed a bug - it should work now. "

thank you, it looks like to work now.

[QUOTE=markr;150620]The stats in the user summary are much better now! The v4 TF & v4 LL counts and credit issues look to be sorted.
[/QUOTE]

I agree. The pie chart now shows almost all of my credit coming from v4-LL, as it should be.

Stats looking much better
 

[QUOTE=markr;150620]The stats in the user summary are much better now! The v4 TF & v4 LL counts and credit issues look to be sorted.
[/QUOTE]

Same here, the totals all tally from 'last 365' and 'lifetime', (now where can I get the cash for a wizz bang Core2 Duo!:wink:)

Miles

Are there any plans to differentiate between V4-LL and V4-LL-D in the stats? Much of my V4 work was double checking, not first time tests. Thanks for the great job so far....

[QUOTE=richs;150737]Are there any plans to differentiate between V4-LL and V4-LL-D in the stats? Much of my V4 work was double checking, not first time tests.[/QUOTE]

Nope. The v4 server did not keep enough information to do this.

[QUOTE=petrw1;150568]And now on my new Quad on which one core is doing TF ... in the last 16 hours it shows credit on the CPU report equal to only one completion (0.0735) even though it has completed 26 for a total of 1.6901.
[/QUOTE]

Similar problem here:
I have a PC (kangoo-64) doingTF, that has so far completed ~3.89 GHz-days of work. But If go to Myaccount/CPUs, I see it has been credited 0.3738, although it shows 57 work units completed, which is right.

[QUOTE=lycorn;151073]Similar problem here:
I have a PC (kangoo-64) doingTF, that has so far completed ~3.89 GHz-days of work. But If go to Myaccount/CPUs, I see it has been credited 0.3738, although it shows 57 work units completed, which is right.[/QUOTE]

All fixed.

[QUOTE=Prime95;151127]All fixed.[/QUOTE]

And so it has ... thanks

:tu:

Two more requests for Santa :smile:
1. Credit transfer tool.
2. Increase number of characters allowed in team display name to accomodate "Ars Technica Team Prime Rib" - that's 27.

Should I be manually Unreserving my v4_Computers Assignments that finished days or weeks ago ... or will that eventually be cleaned up by the server?

It appears that as v4_Computers tests complete they still stay on the Assignment list but show as over due.

[QUOTE=garo;151145]Credit transfer tool.[/QUOTE]

I want to make sure all credit bugs are fixed before allowing this. Otherwise, it could make repairing problems that much more complicated.

[QUOTE=petrw1;151148]Should I be manually Unreserving my v4_Computers Assignments that finished days or weeks ago ... or will that eventually be cleaned up by the server?

It appears that as v4_Computers tests complete they still stay on the Assignment list but show as over due.[/QUOTE]

It wouldn't hurt to do unreserves, but I'll probably be able to sort it all out without you having to do that.

[QUOTE=petrw1;151148]
It appears that as v4_Computers tests complete they still stay on the Assignment list but show as over due.[/QUOTE]

I presume this only happens for TF assignments

[QUOTE=Prime95;151170]I presume this only happens for TF assignments[/QUOTE]

There are LL too. For example: 28865267, 25501093, 28004047.
They all showed a percentage for a while but without a Stage but the percentage counter stopped before it finished. Can't remember why.

[CODE]
v4_computers 0 28865267 LL 90.50% 2008-10-27 09:27 33 2008-10-28 20:25 2008-10-29 20:25 2008-10-31 20:25 -29
v4_computers 0 25501093 LL 49.20% 2008-10-27 02:35 33 2008-10-28 19:44 2008-10-29 19:44 2008-11-07 19:44 -22
v4_computers 0 28004047 LL 8.60% 2008-10-27 20:34 33 2008-10-29 03:40 2008-10-30 03:40 2008-11-13 03:40 -16[/CODE]

[quote=Prime95;151163]I want to make sure all credit bugs are fixed before allowing this. Otherwise, it could make repairing problems that much more complicated.[/quote]

Makes sense.

is it possible to combine pcs
i have one pc that shows up twice on the v5 server

[QUOTE=henryzz;151250]is it possible to combine pcs
i have one pc that shows up twice on the v5 server[/QUOTE]

I reported the same on another thread.

I have one PC twice. In its case I reinstalled XP and it got a new CPU ID.

I have another PC 3 times. My new Vista Q9550 first registered as a Xeon, then when it reported in completion dates it re-registered as a Q9550 but as a new computer, then I rebooted a few days later and it reported again as a Xeon and after a Man. Comm. it was Q9550 again but a third PC was added. So now I have 1 Xeon and 2 Q9550 but they are all the same PC.

According to my account summary page, I have completed 2196 trial factoring assignments. While I'm sure I have done a lot of TF assignments, I don't think I did that many - probably only a few hundred of them at most.

Can anyone explain this?

[quote=ixfd64;151277]According to my account summary page, I have completed 2196 trial factoring assignments. While I'm sure I have done a lot of TF assignments, I don't think I did that many - probably only a few hundred of them at most.

Can anyone explain this?[/quote]

For TF, an assignment (usually) is one additional bit level. Trial factoring a single exponent from 63 to 68 bits would be 5 assignments.

[QUOTE=ixfd64;151277]According to my account summary page, I have completed 2196 trial factoring assignments. While I'm sure I have done a lot of TF assignments, I don't think I did that many - probably only a few hundred of them at most.

Can anyone explain this?[/QUOTE]

My counts are out of whack that way too and I suspect the same thing as ckdo proposed. I am doing TF to Low Limits on one PC. It's assignments are 2^58 to 2^64 with updates and credit awarded at 62, 63 and 64. I haven't figured out yet if that counts as 3 completions or 5.

[QUOTE=henryzz;151250]is it possible to combine pcs
i have one pc that shows up twice on the v5 server[/QUOTE]

I'll add it to the wish list, but it is low priority.

Where else can I look for Start Up
 

One of my PCs is a core duo running XP. In version 24 I had to start two instances of Prime95 (one with the -A1 switch) with worktodo.ini and work0001.ini files. With Version 25 it is smarter and I only need 1 instance and it gives me two workers. This is ALL GOOD!!!

After the upgrade to v25 these .ini files were renamed to .txt. Still goodness.

However, now whenever the PC reboots it starts up using the work0001.txt file. All I have to do is Test Stop; Test Exit and then restart it and then it uses the worktodo.txt file.

My question is where can I look to find where it is starting from at bootup?
- It is NOT in Start...Programs...Startup.
- It is NOT in Control Panel...Administration Tools...Services.

[QUOTE=ixfd64;151277]According to my account summary page, I have completed 2196 trial factoring assignments. While I'm sure I have done a lot of TF assignments, I don't think I did that many - probably only a few hundred of them at most.

Can anyone explain this?[/QUOTE]

You inherited 2160 from the v4 server.

[QUOTE=henryzz;151250]is it possible to combine pcs
i have one pc that shows up twice on the v5 server[/QUOTE]I second the request -- I've got multiple sets of computers listed, for two different reasons:

a) one is the issue [i]petrw1[/i] has been reporting, where the same computer seems to be cloning itself into multiple entries

b) I've reinstalled modified hardware inside a machine enough to give it a new GUID

I would like to be able to merge these all down to what I know to be the actual 3 or 4 real computers I have, not the 17 that show up in my account :smile:

it should be in services...appearing as "Prime95 Service - 0" and "Prime95 Service - 1"...if not i don't know

Rejoice!
 

More v4 CPU credit has been found for many users. Check your updated stats.

[QUOTE=petrw1;151296]One of my PCs is a core duo running XP. In version 24 I had to start two instances of Prime95 (one with the -A1 switch) with worktodo.ini and work0001.ini files. With Version 25 it is smarter and I only need 1 instance and it gives me two workers. This is ALL GOOD!!!

After the upgrade to v25 these .ini files were renamed to .txt. Still goodness.

However, now whenever the PC reboots it starts up using the work0001.txt file. All I have to do is Test Stop; Test Exit and then restart it and then it uses the worktodo.txt file.

My question is where can I look to find where it is starting from at bootup?
- It is NOT in Start...Programs...Startup.
- It is NOT in Control Panel...Administration Tools...Services.[/QUOTE]Since it starts up with -a1 switch (work[b]0001[/b].txt) it is definitely a leftover from the previous installation. I suppose you have only one instance of Prime95 in services... That is the culprit. Look at its properties : the "Path to executable" will be "...\Prime95.exe -A1".

If it is :
to change that exit Prime95, start the Prime95 service, In the now running instance of Prime95 go to the menu Options and uncheck "Start at bootup". Now you restart your OS. No Prime95 instance should starts, go to your Prime95 directory start Prime95, go to the options menu and check "Start at bootup", the resulting service should be free of the -A1 switch.
or a quicker way to correct this is to use regedit and go to the key HKLM/System/CurrentControlSet/Services/Prime95 and correct the value of ImagePath. [b]Of course use regedit at your own risk.[/b]
finally another dirtier way to live with this problem is just to delete the 0001 files and rename your good parameter files to their 0001 counterpart. (local.txt to loca0001.txt, ... and worktodo.txt to work0001.txt) put I would advise against this last bypass but it is quick and dirty.

Jacob

Hello,

is there a solution for the "V5UserID changed to ANONYMOUS" problem?
Yesterday evening I've setup some v25.7 clients (linux) doing some P-1 work.
I've checked my status page and there were 48 assignments, so far, so good.
Today all 24 assignments where no longer listed in my status page so I've checked the status of some of them, they are assign to ANONYMOUS.
I've checked the prime.txt: in 9 out of 16 cases the V5UserID was changed to ANONYMOUS.

btw.: all clients are using the same HardwareGUID, I've set FixedHardwareUID=1, this should be OK, right?

TheJudger

[QUOTE=TheJudger;151369]
is there a solution for the "V5UserID changed to ANONYMOUS" problem?
Yesterday evening I've setup some v25.7 clients (linux) doing some P-1 work.
I've checked my status page and there were 48 assignments, so far, so good.
Today all 24 assignments where no longer listed in my status page so I've checked the status of some of them, they are assign to ANONYMOUS.
I've checked the prime.txt: in 9 out of 16 cases the V5UserID was changed to ANONYMOUS.

btw.: all clients are using the same HardwareGUID, I've set FixedHardwareUID=1, this should be OK, right?[/QUOTE]

Ugh, I thought I'd squashed any anonymous problems. Please send me the prime.log file from one or two affected machines.

I don't know why you would use the FixedHardwareUID option to set up the client on different computers. It is little tested, and perhaps the cause of the trouble.

Unrejoice
 

[QUOTE=Prime95;151317]More v4 CPU credit has been found for many users. Check your updated stats.[/QUOTE]

Oops. I applied the newly found credit with a 5.5 multiplier rather than the correct 5.075 multiplier. I'll fix it. You'll now see a drop in v4 credit.

[QUOTE=petrw1;149792]- My team total is about 200 credits below my account total (and 100 below my Lifetime completed) ... not sure I can be sure what is missing and/or possibly double counted because the #of results and #of credits are different in more than 1 row.

Thanks for the progress[/QUOTE]

Fixed...thanks

[QUOTE=S485122;151345]I suppose you have only one instance of Prime95 in services... That is the culprit. Look at its properties : the "Path to executable" will be "...\Prime95.exe -A1".

Jacob[/QUOTE]

No, NONE in services.

[QUOTE=petrw1;151384]No, NONE in services.[/QUOTE]There is a setting somewhere that starts the program... Try a search for Prime95 in the registry via regedit. Note the location of each instance found. Did you look in all startup directories (All User, your login...) Do a search of your disks to find worktodo and work0001 files, look at the associated files prime, primXXXX, local locaXXXX .ini or .txt. i there something pointing to a service (the relevant file should be local.txt : ServiceName=your service name, DisplayName=your service display name) If needed you can post the content of those files or PM them to me.

Jacob

[QUOTE=petrw1;151293]My counts are out of whack that way too and I suspect the same thing as ckdo proposed. I am doing TF to Low Limits on one PC. It's assignments are 2^58 to 2^64 with updates and credit awarded at 62, 63 and 64. I haven't figured out yet if that counts as 3 completions or 5.[/QUOTE]

And the answer is....

Even though it is one assignmen it counts as 3 results - the same 3 that show up in my Results Page:
1. 58-62
2. 63
3. 64

[QUOTE=petrw1;151178]There are LL too. For example: 28865267, 25501093, 28004047.
They all showed a percentage for a while but without a Stage but the percentage counter stopped before it finished. Can't remember why.

[CODE]
v4_computers 0 28865267 LL 90.50% 2008-10-27 09:27 33 2008-10-28 20:25 2008-10-29 20:25 2008-10-31 20:25 -29
v4_computers 0 25501093 LL 49.20% 2008-10-27 02:35 33 2008-10-28 19:44 2008-10-29 19:44 2008-11-07 19:44 -22
v4_computers 0 28004047 LL 8.60% 2008-10-27 20:34 33 2008-10-29 03:40 2008-10-30 03:40 2008-11-13 03:40 -16[/CODE][/QUOTE]

Update: The exponents listed above and a couple more recent LL completions now show up as D (Double Check) on my Assignment list.

[CODE]
v4_computers 0 28865267 D 90.50% 2008-10-27 09:27 36 2008-10-28 20:25 2008-10-29 20:25 2008-10-31 20:25 -32
v4_computers 0 28912867 D 0.00% 2008-10-27 09:27 36 2008-10-28 20:25 2008-10-29 20:25 2008-11-21 20:25 -11
v4_computers 0 25502293 D LL, 98.50% 2008-10-27 02:35 36 2008-11-29 12:11 2008-11-30 12:11 2009-03-19 20:32 107
v4_computers 0 28004047 D 8.60% 2008-10-27 20:34 36 2008-10-29 03:40 2008-10-30 03:40 2008-11-13 03:40 -19
v4_computers 0 28912867 D 0.00% 2008-10-27 09:27 36 2008-10-28 20:25 2008-10-29 20:25 2008-11-21 20:25 -11
[/CODE]

Since I recently did the LL I cannot do the DC.

Hey more progress.....
 

I followed the advice of a previous poster (sorry I don't remember who or which forum ... but thanks).

1. I put all my currently assigned exponents left over from v4 in the worktodo.add file. (P.S. Not sure why I had to do them 1 by 1. Even when the worktodo.add had many entries only the first was added to worktodo.txt but all were removed from worktodo.add)
2. This assigned them a GID(?) from the v5 server.
3. They still showed up as v4_computers assignments but I got stage/% updates on them.
4. ...that is until today. They now all show up as belonging to the appropriate CPU. :smile:

Now all I have left assigned to v4_computers are a bunch of v4-era assignments that finished but stayed as assigned.
- The LL tests completed and later were listed as D tests. NOT good since one CPU can't do both.
- The TF tests completed but stayed as assigned TF tests.
However, NONE of these assignments show up in any worktodo.txt file on any of my CPUs.

I would simply Unassign all of these but I hesitate in case George still wants to do some of his own verification and clean-up on them.

I await direction with the patience of Job.

What happened to the Completed link that used to be in the Lifetime Stats section of your account report? Is there another way to see what has completed?

[QUOTE=1997rj7;151797]What happened to the Completed link that used to be in the Lifetime Stats section of your account report? Is there another way to see what has completed?[/QUOTE]

Results under the My Account menu on the left.

[QUOTE=henryzz;151250]is it possible to combine pcs
i have one pc that shows up twice on the v5 server[/QUOTE]I noticed a new [Merge Checked CPUs] button on the Computer Details page... so I tried it out and it seems to work ok, my only comment so far is that there's no way of specifying which computer details to keep, that is, what is the current-good-valid GUID/name/etc and which are the old-duplicate-invalid ones I want to merge into the first one. What happens if my current CPU gets merged into an old one, and then I continue to report results with the new one -- will the new one reappear, or does the merging process add a permanent pointer?

[QUOTE=James Heinrich;151887] What happens if my current CPU gets merged into an old one?[/QUOTE]

Can't happen. It always merges into the CPU that last reported to the server.

As a safety measure you can't merge cpu A into cpu B if A communicated with the server after B was created.

[QUOTE=Prime95;151891]Can't happen. It always merges into the CPU that last reported to the server.

As a safety measure you can't merge cpu A into cpu B if A communicated with the server after B was created.[/QUOTE]

COOL!!! Worked great for me ... even merging 3 into 1 (in two passes of two).

The GHz-days credited and Work units completed on the CPU Properties page are not accurate but I don't think it is related to the Merge. It seems to be more out of whack for CPUs doing TF.

Are my assumptions and calculations out to lunch?
 

Is it possible TF-LMH gets more credit per hour than TF?

My scenario (I think I have included the relevant stats):

CPU1:
PIII 866 doing TF-LMH
Averaging about 0.5 points per day
CPU Rolling Average 88%

CPU2:
PIII 1300 doing TF
Averaging about 0.33 points per day
CPU Rolling Average 55%

Both are set to run 24 hours a day but I think CPU2 may get shut down occasionally. I assume that is responsible for the 55%.

If I am ciphering correct then:
CPU1 is running at 866X88%=762Mhz Equivalent
CPU2 is running at 1300X55%=715Mhz Equivalent: 94% of CPU1 but getting about 66% the points.

[quote=Prime95;151891]As a safety measure you can't merge cpu A into cpu B if A communicated with the server after B was created.[/quote]

Bummer.

I had some machines which were running two copies of mprime - one processing v4 work and one processing v5 work. Now in order to join them I will probably have to remove the GUID, let it create a fresh one, and merge all three, yeah?

[QUOTE=petrw1;151897]
The GHz-days credited and Work units completed on the CPU Properties page are not accurate .[/QUOTE]

Fixed

[QUOTE=ckdo;151909]Bummer.

I had some machines which were running two copies of mprime - one processing v4 work and one processing v5 work. Now in order to join them I will probably have to remove the GUID, let it create a fresh one, and merge all three, yeah?[/QUOTE]

Just for you --- Merge the two machines, you should get an error message. Then add "&date_override=99" to the URL and try again. Let me know if that works.

CPU Properties anomaly?
 

[QUOTE]Software Version Windows,Prime95,v25.7,build 3
Model Intel(R) Core(TM)2 Duo CPU E6550 @ 2.33GHz
Features Dual core, Prefetch,SSE,SSE2
[B][U]Speed 2.327 GHz (7.915 GHz P4 effective equivalent) [/U][/B]
L1/L2 Cache 32 / 4096 KB
Computer Memory 2048 MB configured usage 200 MB day / 200 MB night [/QUOTE]

[QUOTE]Software Version Windows64,Prime95,v25.7,build 4
Model Intel(R) Core(TM)2 Quad CPU Q9550 @ 2.83GHz
Features 4 core, Prefetch,SSE,SSE2,SSE4
[B][U]Speed 2.833 GHz (5.557 GHz P4 effective equivalent) [/U][/B]
L1/L2 Cache 32 / 6144 KB
Computer Memory 4094 MB configured usage 1600 MB day / 1600 MB night[/QUOTE]

It seems odd that the Core2 Processor while slower has a P4 equivalent 42% more than the Quad Processor.

Could it be that for the Core2 it is accounting for both cores and for the Quad it is only accounting for 1?

Or does Build 4 compute differently?

James Heinrich noticed a new [Merge Checked CPUs] button on the Computer Details page...

... and so did I, but there are no checkboxes for me to check my 2 registered computers. (I tried just checking the "I am sure ..." box and clicking on the "Merged checked CPUs" button, but when the page redisplayed, there was no change.)

Details that may matter:

The first computer listed is "Manual testing". Last Activity date is "[COLOR=#00ff00]2008-08-13 11:54[/COLOR]", in this hard-for-me-to-read yellowish (or-greenish?) color. (I have red-green-color-deficient vision.) Under "status" is "[IMG]http://mersenne.org/images/yellow_light.gif[/IMG][SIZE=1][COLOR=#00ff00]!![/COLOR][/SIZE] U".

The second one is "v4_computers". Last Activity date is in normal color. Under "status" is "[IMG]http://mersenne.org/images/yellow_light.gif[/IMG] U".

Then there's a footnote in hard-for-me-to-read color: "[SIZE=1][COLOR=#00ff00]!!At least one CPU is 90 days or more unreported and appears lost.[/COLOR][/SIZE]".

I'm using Firefox 3.0

[quote=Prime95;151936]Just for you --- Merge the two machines, you should get an error message. Then add [spoiler]"&date_override=99"[/spoiler] to the URL and try again. Let me know if that works.[/quote]

:tu: Works. :bow:

[QUOTE=petrw1;151949]It seems odd that the Core2 Processor while slower has a P4 equivalent 42% more than the Quad Processor.

Could it be that for the Core2 it is accounting for both cores and for the Quad it is only accounting for 1?

Or does Build 4 compute differently?[/QUOTE]

The P4 equivalent numbers are changing (slightly) so I incorrectly assumed it is some calculated measure of performance based solely on the hardware ... it must have something to do with results returned ... and if that is the case then my previous question MIGHT BE moot.

Never mind: (Now where is that icon that represents "research first; ask questions later"
[url]http://www.mersenneforum.org/showpost.php?p=148564&postcount=6[/url]

[QUOTE=cheesehead;151961]Details that may matter:

The first computer listed is "Manual testing".

The second one is "v4_computers". [/QUOTE]

First: Results you have manually submitted
Second: ALL pooled results prior to the v5 cutover (Oct 20-27) + any results since the cutover from machines you had/have NOT yet upgraded to Prime95 v25.7.

These two DO NOT have checkboxes for me either; probably because they do NOT represent actual specific CPUs so it does NOT make sense to try to merge them.

[quote=petrw1;151992]+ any results since the cutover from machines you had/have NOT yet upgraded to Prime95 v25.7.

These two DO NOT have checkboxes for me either; probably because they do NOT represent actual specific CPUs so it does NOT make sense to try to merge them.[/quote]Oh, I had failed to connect what I read here about v4_computers to what I see on my reports. Thanks.

Team links and questions.
 

Hi folks,

I'm fiddling with the team panels on Primenet 5.0

1 - I'm noticing that links related to team details do not work. Is it the same for you?

2 - Is it possible (George, Scott?) to change Team password? Older teams have their password spread out of the Internet due to installation purposes. That's not fine for security purposes...

3 - Once a user has joined a team, is it possible for the user to still see his personal stats ranking together with team stats ranking?

Thank you for the attention.

Luigi

[QUOTE=ET_;152059]
1 - I'm noticing that links related to team details do not work. Is it the same for you?

2 - Is it possible (George, Scott?) to change Team password? Older teams have their password spread out of the Internet due to installation purposes. That's not fine for security purposes...

3 - Once a user has joined a team, is it possible for the user to still see his personal stats ranking together with team stats ranking?[/QUOTE]

The team web pages need a lot of work - they will improve over time. There is no way to change the password.

Any idea on the TF Counts Summary vs Results?
 

Summary (Last 365 days):
[QUOTE]TF 105 1518 425.1299 10,408 [/QUOTE]
Results:
[QUOTE]petrw1 has 8,717 results in the last 365 days
(Of these, 8,700 are TF)[/QUOTE]

So I am overcounted by 1,708 TF

In case this helps:
Nov 9-17: 5,084 TF 2^58 - 2^61: Reports as one result each
Nov 17-: 924 TF 2^58 - 2^64: Reports as 3 results at 58-62, 63 and 64
In recent days the Summary count and Results count are going up the same so whatever caused the discrepancy was happening in the past and has been fixed now but the totals still mismatch.

Thanks