Redundent Result

Message boards : Number crunching : Redundent Result

Author	Message
Zydor Send message Joined: 8 Feb 09 Posts: 252 Credit: 1,309,451 RAC: 0 Level Scientific publications	Message 10909 - Posted: 29 Jun 2009 \| 1:46:49 UTC
	Just started to crunch this one a few minutes ago http://www.gpugrid.net/workunit.php?wuid=560403 Two of us have got it for crunching. There were two other attempts, one was timed out - the other was a "Redundent Result". Should we in fact be doing this one? I was under the impression that "Redundent Result" meant "not needed any more". If we are crunching it, it seems as though it is needed, in which case I've got it wrong as to what "Redundent Result" means. I need some education methinks - help :) Regards Zy
	ID: 10909 \| Rating: 0 \| rate: / Reply Quote

ExtraTerrestrial Apes Volunteer moderator Volunteer tester Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level Scientific publications	Message 10921 - Posted: 29 Jun 2009 \| 21:00:14 UTC - in response to Message 10909. Last modified: 29 Jun 2009 \| 21:00:39 UTC
	Mh, what I'm seeing there doesn't match your description. What I can read out of these results: - the WU was 1st sent out 23 Jun 2009 11:45:37 - a 2nd time on 23 Jun 2009 12:33:11 - the 1st guy timed out at 28 Jun 2009 11:45:37 - it was sent out to you and Paul on 28 Jun 2009 12 o'clock - both of you reported successfully on 29 Jun 2009 15 o'clock - on 29 Jun 2009 17 o'clock the 2nd guy probably contacted the server again and at this point his result was declared redundant MrS ____________ Scanning for our furry friends since Jan 2002
	ID: 10921 \| Rating: 0 \| rate: / Reply Quote

Zydor Send message Joined: 8 Feb 09 Posts: 252 Credit: 1,309,451 RAC: 0 Level Scientific publications	Message 10940 - Posted: 30 Jun 2009 \| 22:36:47 UTC - in response to Message 10921. Last modified: 30 Jun 2009 \| 22:42:54 UTC
	It was declared redundent on the 28th, not the 29th. I did not start to crunch mine until over eight hours after it was declared redundent. For sure it was sent to me four hours before it was declared redundent, but I didnt start to crunch it until around 0135hrs the next day - over eight hours after it was declared redundent. No biggie for me, it just seems those circumstances where I had not started to crunch it at the time it was made redundent - and I didnt start until over eight hours later - would be a good one for server cancellation of suitable copies of the redundent WU that have not yet started to be crunched. The sequence was: 28 Jun 2009 12:48:09 UTC Delivered to me 28 Jun 2009 17:01:28 UTC Redundant result declared 29 Jun 2009 1:35:56 UTC Previous WU in my queue finished by me 29 Jun 2009 01:36:00 UTC (approx) I started to crunch the redundent result 29 Jun 2009 14:58:40 UTC I finished the WU When the server declared it redundent at 28 Jun 1701 - should it not have also done so for other copies that had not yet been started, eg mine? Regards Zy
	ID: 10940 \| Rating: 0 \| rate: / Reply Quote

ExtraTerrestrial Apes Volunteer moderator Volunteer tester Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level Scientific publications	Message 10946 - Posted: 1 Jul 2009 \| 20:55:23 UTC - in response to Message 10940.
	It was declared redundent on the 28th, not the 29th. You're right - I don't know how the 29 got into my head :p At 28 Jun 2009 17:01:28 the host nr 1 was well beyond the 5 days deadline. So it may be that he contacted the server again and had not yet started this WU. The server told him not to run the WU (because it would be too late anyway), but because he contacted the server the message is not "no reply - timeout" as for the other host. And because he hadn't started yet it's not the message "aborted by server". So I guess the message "redundant result" was chosen, because his result would surely have been redundant because the two of you already ran the WU. However, this wouldn't mean that your results weren't needed any more. Notice how it says "redundant result", not "redundant WU" ;) MrS ____________ Scanning for our furry friends since Jan 2002
	ID: 10946 \| Rating: 0 \| rate: / Reply Quote

Zydor Send message Joined: 8 Feb 09 Posts: 252 Credit: 1,309,451 RAC: 0 Level Scientific publications	Message 10949 - Posted: 2 Jul 2009 \| 1:51:22 UTC - in response to Message 10946. Last modified: 2 Jul 2009 \| 1:53:34 UTC
	........ So I guess the message "redundant result" was chosen, because his result would surely have been redundant because the two of you already ran the WU. However, this wouldn't mean that your[/b] results weren't needed any more. Notice how it says "redundant result", not "redundant WU" ;)..... Its not a question of being needed, at the end of the day I will crunch whats given... The whole point about this is that the WU was declared redundent before the final two had even started it .... the final two should therefore have been cancelled, and not allowed to run?? If this means what I think it does, it would seem there is a slight glitch in the recall programme of cancellations, which allows redundent results to be crunched again, when the Project could have got some good ones done. If thats the case, there will be an unknown number going the same way across the whole Project, meaning less work is done for the Project. I'll get back in me box :) Regards Zy
	ID: 10949 \| Rating: 0 \| rate: / Reply Quote

ignasi Send message Joined: 10 Apr 08 Posts: 254 Credit: 16,836,000 RAC: 0 Level Scientific publications	Message 10950 - Posted: 2 Jul 2009 \| 10:50:31 UTC - in response to Message 10949.
	If this means what I think it does, it would seem there is a slight glitch in the recall programme of cancellations,... It could well be. Anyway, we are aware of the confusion that the redundant results stuff brings to, so we are stopping from using it. It was mainly all WU's tagged IBUCH. Now they will have less steps. With no change for the computing length. Instead of -X-10- the later ones are coming with -X-3- which seemed to be enough for us as we have recently seen. This is gonna be quite an optimization for simulations from the computing point of view. cheers, ignasi
	ID: 10950 \| Rating: 0 \| rate: / Reply Quote

Zydor Send message Joined: 8 Feb 09 Posts: 252 Credit: 1,309,451 RAC: 0 Level Scientific publications	Message 10953 - Posted: 2 Jul 2009 \| 18:40:56 UTC - in response to Message 10950.
	.......This is gonna be quite an optimization for simulations from the computing point of view.... Will that mean you get more work in the same time frame, or does it change the length of the WU? Just curious ... Regards Zy
	ID: 10953 \| Rating: 0 \| rate: / Reply Quote

ignasi Send message Joined: 10 Apr 08 Posts: 254 Credit: 16,836,000 RAC: 0 Level Scientific publications	Message 10996 - Posted: 7 Jul 2009 \| 10:57:44 UTC - in response to Message 10953.
	No, no... I mean that 3 steps seem to be enough data for us. However, in the future when faster cards (i.e. GTX 300 series) start to be adopted, we might even condense the current 3 steps into 2, being WUs then 50% longer or so... We will see. salut, ignasi
	ID: 10996 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : Number crunching : Redundent Result

	About	Science	Volunteers	Performance	Forum	Join us	Donate