[QUOTE=garo;116251]Happened at about 2100 UTC Oct 11th at my end.[/QUOTE]

hmm I'm in the Uk so I should be on UTC+1 which means mine glitched 4 hours later

Hi, I have a quad core Q6600 CPU, and when I tried to stress my CPU using Prime95 25.5 (in windows xp x64), with the test "In-place large FFTs (maximum heat, power comsumption, some RAM tested)", I have errors and WITHOUT ANY OVERCLOCK.

I bought the processor 5 days ago, and I haven't got any error in windows... (only in prime 95 I detect this error)

Could be that the processor is damaged? or could be possible that the program don't run properly yet?

If you need any file, tell me.


Thank you, and sorry for my horrible english.

if you're new to this project you should probably use 24.14 on [url]www.mersenne.org[/url] and stress test with that. If you get the same problem then yes it is probably a hardware fault. 25.5 is for a beta test and not for general use at the moment

[quote=Bundu;116300]if you're new to this project you should probably use 24.14 on [URL="http://www.mersenne.org"]www.mersenne.org[/URL] and stress test with that. If you get the same problem then yes it is probably a hardware fault. 25.5 is for a beta test and not for general use at the moment[/quote]

Thank you. But, how can I probe my 4 cores with 24.14?

[quote=Brian-E;116262]
(2) Another possible bug noticed: no torture test was performed before starting to test a new assigned exponent (double check in the M19000000 series) even though I started with a completely new user account.[/quote]

I don't think any of the 25.x versions do a test with each new FFT length. This may be by design.

[quote=ed_co;116301]Thank you. But, how can I probe my 4 cores with 24.14?[/quote]
[SIZE=2]
You could test with the affinity set between 0 and 3 in the "Advanced", "Affinity" dialog.

Or, you could run four instances of Prime95 simultaneously.[/SIZE][SIZE=2]

I would suggest removing one ram module at a time while running the tests. I have bought more than my share of faulty DDR2 modules lately.
[/SIZE]

[QUOTE=AES;116307]I would suggest removing one ram module at a time while running the tests. I have bought more than my share of faulty DDR2 modules lately.[/QUOTE][url]http://www.memtest.org/[/url] would be a better way of testing RAM.

You can start multiple instances of Prime95 v24.x with the -A parameter to set affinity, so open a command prompt and start [b]prime95.exe -A1[/b] (-A2, -A3, etc) for as many instances as you need.

If you run more than 1 instance of Prime95 v24.x torture test, make sure you don't let it run any torture test type that tests any significant amount of RAM on more than 1 instance, otherwise it'll try and allocate far more RAM than your system has and will either thrash your harddrive, or just die. You can do Blend on one torture test, and Small FFTs on the other instances.

I don't know if this is anything or not, but I noticed it reported the intermediate status on the line [i]after[/i] it found a factor:[code][Oct 14 20:08] Starting trial factoring of M132320527 to 2^57
[Oct 14 20:08] M132320527 has a factor: 3175692649
[Oct 14 20:08] Trial factoring of M132320527 to 2^50 is 12.49% complete.[/code]From my brief observation it seems that if the factor is found early on (before 2^51) it seems to always report that "factoring to 2^50 is 12.49% complete" [i]after[/i] it reports the factor, but if the factor found is larger than 2^51 no status report is printed after the reported factor, for example:[code][Oct 14 20:18] Starting trial factoring of M132323119 to 2^57
[Oct 14 20:18] Trial factoring of M132323119 to 2^50 is 0.09% complete.
[Oct 14 20:18] Trial factoring of M132323119 to 2^51 is 6.25% complete.
[Oct 14 20:18] Trial factoring of M132323119 to 2^52 is 6.25% complete.
[Oct 14 20:18] Trial factoring of M132323119 to 2^53 is 6.25% complete.
[Oct 14 20:18] Trial factoring of M132323119 to 2^54 is 4.33% complete.
[Oct 14 20:18] M132323119 has a factor: 18332786974311353[/code]Again, this probably isn't significant, but I just noticed it so I thought I'd mention it.

This is not a bug. Prime95 does factoring in passes and sometimes it goes on to complete a pass even after a factor has been found. I've seen the same behaviour too.

Enhancement Request....
 

The following is a scenario from version 24 ... but I'd like to request that version 25.x be enhanced as described at the bottom of this note...
==============================
I have a small team of GIMPS computers, some more readily accessible than others. So I thought I would try an experiment: I tried to "Unreserve" an exponent from the assignment list of Computer "B" while sitting in front of Computer "A".

It sort of worked:
1. Computer "A" responded with the standard "Stopping / Starting ..." message. Then when I rechecked my team list of assignments the "Unreserve"d exponent was no longer there.

2. However, Computer "B" did not receive another assignment to replace the unreserved exponent. I checked a few times over the next couple days.

3. 2 days later I was able to get to Computer "B" and noticed it was displaying the error: "Error 3: Exponent not assigned ..." every time it contacted the Primenet server. This exponent was also missing from the Computer "B" worktodo.ini file. I tried manual communication ... same error. I tried adding the exponent back into the worktodo.ini file and unreserving it from Computer "B" ... same error.

Now, I seem to recall from past posts that if I just wait a few days the server will clear up this error ... I can live with that.

My only request would be to know if this is supposed to be a supported feature or something that should not have worked. That is, I would like the ability to do some team maintenance from any team PC. Such as:
- unreserve exponents
- send new expected completion dates
- change work type (LL, DC, Fact)
- change "at least..." days of work
- change frequency of updates
- set vacation
- ... pretty much anything I can do from my own PC I'd like to be able to for any team PC.

There is a personal web page in the v5 server that will allow you to do all this and more.