mersenneforum.org Double credit on GPU72
 Register FAQ Search Today's Posts Mark Forums Read

 2014-08-07, 01:35 #1 kladner     "Kieren" Jul 2011 In My Own Galaxy! 1015810 Posts Double credit on GPU72 I can't figure what to make of this. Here are three examples of assignments being recorded twice, with different times, with credit. As to the results labelled "proven bad", I will take that up in Hardware\GPU where it is better suited. That story got much bigger than my question here. Attached Thumbnails
2014-08-07, 15:51   #2
chalsall
If I May

"Chris Halsall"
Sep 2002

7×1,361 Posts

Quote:
 Originally Posted by kladner I can't figure what to make of this. Here are three examples of assignments being recorded twice, with different times, with credit. As to the results labelled "proven bad", I will take that up in Hardware\GPU where it is better suited. That story got much bigger than my question here.
Hmmm... That's interesting.

Here's what I can quickly see from the database:
Code:
mysql> select ID,Exponent,Assigned,Updated,Completed,CID,CPU from Assigned where Exponent=32254777 and User="8ad778c87c3d88f688fef4ebad36994c";
+---------+----------+---------------------+---------------------+---------------------+----------------------------------+-----+
| ID      | Exponent | Assigned            | Updated             | Completed           | CID                              | CPU |
+---------+----------+---------------------+---------------------+---------------------+----------------------------------+-----+
| 1366659 | 32254777 | 2014-05-23 23:24:01 | 2014-06-13 18:03:35 | 2014-06-14 16:41:28 | 99cfd947b1d524d5f8db2fe36836xxxx |   6 |
| 1390985 | 32254777 | 2014-06-14 16:41:32 | 2014-06-14 16:41:32 | 2014-06-14 18:51:26 | 99cfd947b1d524d5f8db2fe36836xxxx |   4 |
+---------+----------+---------------------+---------------------+---------------------+----------------------------------+-----+
2 rows in set (0.00 sec)
Based on the dates assigned, these requests would have been passed straight through the proxy to Primenet to be fulfilled.

I don't have time at the moment to drill down on the raw logs, but will try to this weekend. It either indicates a bug in GPU72's proxy, and/or a bug with Primenet.

Is there any chance you could look at the logs for your 99cf... ("pod64") machine, and let us know if you received this assignment twice in a row? Based on the different CPUs the assignment was given to, it suggests strongly that it was, even though it shouldn't have been.

Thanks for bringing this forward. Strangeness should always be investigated.

Edit: Oh, also, although I didn't show it in the above query, the AIDs are also different, and are in Primenet's format (read: upper-case).

Last fiddled with by chalsall on 2014-08-07 at 16:10

 2014-08-07, 16:16 #3 kladner     "Kieren" Jul 2011 In My Own Galaxy! 236568 Posts Code: pod64 (4)32254777DC70--2014-06-14 16:41:322014-06-14 18:51:2635.502 pod64 (6)32254777DC70--2014-05-23 23:24:012014-06-14 16:41:2835.502 Here's a start- Obviously, the above is from GPU72's Completed page. The assigned date(s) are highlighted. However, this exponent only shows one instance of Assign -> Complete in prime.log. It appears as a single completion in results.txt: Code: [Fri Jun 13 18:27:27 2014] UID: kladner/pod64, M32254777 is not prime. I will look further as I think of other places, or if you have suggestions as to where. Edit: Results.txt shows Code: AID: E14E7702B47645718D345DDC55xxxxxx .....as does prime.log Last fiddled with by kladner on 2014-08-07 at 16:24
2014-08-07, 16:30   #4
chalsall
If I May

"Chris Halsall"
Sep 2002

100101001101112 Posts

Quote:
 Originally Posted by kladner I will look further as I think of other places, or if you have suggestions as to where.
Hmmmm... Interesting...

Could you please email me your full prime.log and results.txt files for "pod64"? Use the email address given on the GPU72.com contact page. I promise not to do anything nasty with the information contained.

Something very strange happened here and I want to fully understand it, since it will likely also affect others. I will correlate your log files with mine -- this will help me greatly to determine if I've made a SPE.

Thank you again for bringing this forward.

2014-08-07, 16:32   #5
chalsall
If I May

"Chris Halsall"
Sep 2002

7·1,361 Posts

Quote:
 Originally Posted by kladner Edit: Results.txt shows Code: AID: E14E7702B47645718D345DDC55xxxxxx .....as does prime.log
Yup, that's what I see as well for the first assignment. The second was F6D04DE4CAD3E490A0D0D314xxxxxxxx

2014-08-07, 16:54   #6

"Kieren"
Jul 2011
In My Own Galaxy!

2·3·1,693 Posts

Quote:
 Originally Posted by chalsall Yup, that's what I see as well for the first assignment. The second was F6D04DE4CAD3E490A0D0D314xxxxxxxx
OK. Thanks. I'll root around for the second AID.

2014-08-09, 01:55   #7
chalsall
If I May

"Chris Halsall"
Sep 2002

7·1,361 Posts

Quote:
 Originally Posted by kladner OK. Thanks. I'll root around for the second AID.
Just wondering before I start digging through the logs... Did you find any evidence of F6D04DE4CAD3E490A0D0D314xxxxxxxx on your end?

2014-08-09, 06:57   #8

"Kieren"
Jul 2011
In My Own Galaxy!

236568 Posts

Quote:
 Could you please email me your full prime.log and results.txt files for "pod64"? Use the email address given on the GPU72.com contact page. I promise not to do anything nasty with the information contained.
Sorry. I overlooked the above line. I will send the files at once.

2014-08-09, 15:07   #9

"Kieren"
Jul 2011
In My Own Galaxy!

2×3×1,693 Posts

Quote:
 Originally Posted by chalsall Just wondering before I start digging through the logs... Did you find any evidence of F6D04DE4CAD3E490A0D0D314xxxxxxxx on your end?
PM sent with the relevant exponent being reported as 'not prime', and then immediately reassigned in the same communication session.

2014-08-09, 16:22   #10
chalsall
If I May

"Chris Halsall"
Sep 2002

7·1,361 Posts

Quote:
 Originally Posted by kladner PM sent with the relevant exponent being reported as 'not prime', and then immediately reassigned in the same communication session.
Thank you. I copy both the PM and the emailed logs.

I have PM'ed you back, CC'ing in George.

Thanks again for noticing and reporting this! This is exactly how bugs are squashed!

 Similar Threads Thread Thread Starter Forum Replies Last Post Mysticial Software 50 2017-10-30 19:16 jasong jasong 7 2015-08-17 10:56 bayanne GPU to 72 17 2013-12-25 18:16 Uncwilly Puzzles 8 2006-07-03 16:02 ThomRuley PrimeNet 3 2004-05-01 01:12

All times are UTC. The time now is 04:00.

Sun Apr 18 04:00:25 UTC 2021 up 9 days, 22:41, 0 users, load averages: 2.53, 2.29, 2.10