sb client continue to crunch after it found a P-1 factor

[Aillas]

Hi,

I was looking at my result.txt and prime.log files and found this:

On december 10, sb found a factor for 22699*2^17456374

In Result.txt
[Thu Dec 10 01:11:26 2009]
P-1 found a factor in stage #2, B1=40000, B2=170000.
UID: Ludovic, 22699*2^17456374+1 has a factor: 2449061290113623401, AID: 000000000000000000000000000F7A93

In prime.log
[Thu Dec 10 01:11:26 2009 - ver 25.11]
Sending result to server: UID: Ludovic, 22699*2^17456374+1 has a factor: 2449061290113623401, AID: 000000000000000000000000000F7A93

So the result was sent to the server.

and then I found this:

In Result.txt
[Sun Jan 3 03:41:46 2010]
UID: Ludovic, 22699*2^17456374+1 is not prime. Wd2: D1BA2149,00000000, AID: 000000000000000000000000000F7A93

In prime.log
[Sun Jan 3 03:41:46 2010 - ver 25.11]
Sending result to server: UID: Ludovic, 22699*2^17456374+1 is not prime. Wd2: D1BA2149,00000000, AID: 000000000000000000000000000F7A93



Hopefully, it occurs only on 1 of my exponents, but what a waste of time.
I hope there are not a lot of cases like this one.

Hope this help

Regards

[opyrt]

Hi.

I have no answer to your post, but I would like to remind you that you should not post the residues (RES64) to the public.

[Aillas]

Hi.

I have no answer to your post, but I would like to remind you that you should not post the residues (RES64) to the public.

I was not aware of this.
Removed.

[Frodo42]

I sure hope this gets fixed ... I haven't seen it on my cruncher here (Linux 64bit) but I haven't found any factors ... if everyone could check their result.txt files so that we can see if this is a general problem and Louie will be able to fix it fast when he finds the time.

Hi.

I have no answer to your post, but I would like to remind you that you should not post the residues (RES64) to the public.

Completely of topic ... but why?

[imwithid]

I've checked my results.txt (prior to this thread I've never really looked through it). I was surprised to see several round off errors recently (in the last two months, I've seen six "ERROR: ROUND OFF (0.5) > 0.40"). Again, not to detract from the current thread, but:

1. What does this error mean? If I get this error should I turn the "Round off checking" feature on (even though the readme.txt states that there is no good reason for having this feature) or do I have a hardware problem?
2. Like the above user had mentioned, what's with the RES64 output? Why is this to be kept confidential?

Regards,
Dan.

[philmoore]

The RES64 output would enable another user to fraudulently claim that they had double-checked this number and therefore receive credit for it. Masking the last 2 characters (8 bits) would in all likelihood prevent this. So for example: 7AE52134B5669A__ reveals most of the residue, but not the entire thing.

By the way, the round-off error > 0.4 probably indicates a problem with your computer hardware, and needs to be investigated. Turning off round off checking is not a good idea!

[imwithid]

By the way, the round-off error > 0.4 probably indicates a problem with your computer hardware, and needs to be investigated. Turning off round off checking is not a good idea!

I've turned round off checking on to see what would happen.

[Work thread Jan 29 02:51] Iteration: 15181500 / 17450262 [86.99%]. Round off: 0.0000000000 to 0.0937500000. Per iteration time: 0.046 sec.

I now get the above in addition to the "per iteration" time. Is this within norms or should it be 0?

[enderak]

That's within norms, I believe. However, you stated earlier that you have seen several Round Off > 0.4 errors. You might want to investigate this. From the prime95 readme:

If you do not get the "Disregard last error..." message or this happens
more than once, then your machine is a good candidate for a torture test.
See the stress.txt file for more information.

Running with round off's > 0.5 will result in bad results. It's not fatal to the project, since second-pass checks are done, but it's not good either. You could just be wasting your electricity and cpu cycles, and I think if a second-pass and subsequent error check found your result in error, your credit for that test gets taken away from your score.

[imwithid]

I will be looking into it. I've done several memtest86+ as well as a full torture test with no problems. I've tested the temperature on the die and it is +1-2 °C above maximum recommended limits (one possible cause).

I suspect that it was while I was burning a couple of discs. It does coincide with the last error date and time and I did have two bad burn sessions due to some memory error in the logs (I've had a couple of other bad burns in the past few months - not sure if they relate with the error). Sometimes I forget to close Prime95 while doing other CPU and memory intensive applications.

I'll attempt to recreate the error upon the completion of this number.

[Update]

I just checked my logs and I have another error. This time I am certain of the cause. It was the result of a forced suspend from a third party application. I have recently been using a media player to suspend the system after completion of a play list (if I leave the room / asleep, etc.).

The following is a portion of the window output:

[Work thread Jan 29 10:35] Iteration: 15705000 / 17450262 [89.99%].  Round off: 0.0429687500 to 0.0703125000.  Per iteration time: 0.047 sec.
[Work thread Jan 29 12:59] Iteration: 15879500 / 17450262 [90.99%].  Round off: 0.0429687500 to 0.0781250000.  Per iteration time: 0.049 sec.
[Work thread Jan 29 22:35] Iteration: 15888001/17450262, ERROR: ROUND OFF (0.5) > 0.40
[Work thread Jan 29 22:35] Continuing from last save file.
[Work thread Jan 29 22:35] Resuming PRP test of 10223*2^17450249+1 using all-complex FFT length 1536K
[Work thread Jan 29 22:35] Iteration: 15885439 / 17450262 [91.03%].  Round off: 0.0429687500 to 0.0781250000.
[Work thread Jan 30 00:49] Iteration: 16054000 / 17450262 [91.99%].  Round off: 0.0429687500 to 0.0781250000.  Per iteration time: 0.048 sec.

[Frodo42]

Some time ago I had a corruption in either CPU or Motherboard (never figured out which part it was) some guys at Gentoo forums where very helpful at identifying this for me, maybe some of the tricks here can also help you identify your problem even if you don't use Gentoo.

http://forums.gentoo.org/viewtopic-t-690802.html

[imwithid]

Some time ago I had a corruption in either CPU or Motherboard (never figured out which part it was) some guys at Gentoo forums where very helpful at identifying this for me, maybe some of the tricks here can also help you identify your problem even if you don't use Gentoo.

http://forums.gentoo.org/viewtopic-t-690802.html

I appreciate the info. It was an interesting read and perhaps further evidence that a seemingly stable computer after a couple of torture tests and ram tests may not be so stable.

I am very confident that it is a suspend issue. I viewed the log and the times coincide within the time period that I run the application and the expected auto suspend time.

[enderak]

I think you are correct.

It looks like it recovered fine, whatever the problem was. The roundoff error is at iteration 15888001 and then it goes back and starts again from 15885439, and is able to continue without incident. Looks like your hardware is fine, just need to stop it from going into suspend mode.

[engracio]

To get back into the topic, doesn't this mean that Prime should not have been crunching this wu??

[Sat Feb 06 12:23:44 2010]
10223*2^17592329+1 does not need P-1 factoring.
[Sat Feb 06 18:52:18 2010]
UID: engracio, 10223*2^17592329+1 completed P-1, B1=60000, B2=510000, Wd1: C74C1AE4, AID: 000000000000000000000000000F966D

Just wondering why it is currently running and on 2% completed.

[Frodo42]

To get back into the topic, doesn't this mean that Prime should not have been crunching this wu??

[Sat Feb 06 12:23:44 2010]
10223*2^17592329+1 does not need P-1 factoring.
[Sat Feb 06 18:52:18 2010]
UID: engracio, 10223*2^17592329+1 completed P-1, B1=60000, B2=510000, Wd1: C74C1AE4, AID: 000000000000000000000000000F966D

Just wondering why it is currently running and on 2% completed.

As far as I can figure from the above it didn't find a factor so crunching it should be ok, looks a bit strange that it first says it doesn't need P-1 factoring and then later gives back results from a P-1 test.

[engracio]

As far as I can figure from the above it didn't find a factor so crunching it should be ok, looks a bit strange that it first says it doesn't need P-1 factoring and then later gives back results from a P-1 test.

Coolest, thinking back maybe it decided to factor the wu after I changed the memory settings from 300mb to 512mb. Since it is being crunched now, P-1 factoring did not find any factor??

[Frodo42]

Coolest, thinking back maybe it decided to factor the wu after I changed the memory settings from 300mb to 512mb. Since it is being crunched now, P-1 factoring did not find any factor??

if it finds a factor you will find a line with these words in your results text
"P-1 found a factor"
so since you didn't post a line like that I guess no factor was found.

[prime95]

On december 10, sb found a factor for 22699*2^17456374

In Result.txt
[Thu Dec 10 01:11:26 2009]
P-1 found a factor in stage #2, B1=40000, B2=170000.
UID: Ludovic, 22699*2^17456374+1 has a factor: 2449061290113623401, AID: 000000000000000000000000000F7A93

and then I found this:

In Result.txt
[Sun Jan 3 03:41:46 2010]
UID: Ludovic, 22699*2^17456374+1 is not prime. Wd2: D1BA2149,00000000, AID: 000000000000000000000000000F7A93

Has anyone else noticed this problem? I just tried replicating the problem and couldn't. Please check your results.txt files and let me know if this was a one time occurrence or if finding P-1 factors is not then skipping the PRP test.