Welcome to MilkyWay@home

unusual behaviour

Message boards : Number crunching : unusual behaviour
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Lord Tedric
Avatar

Send message
Joined: 9 Nov 07
Posts: 151
Credit: 8,391,608
RAC: 0
Message 30510 - Posted: 11 Sep 2009, 21:58:31 UTC
Last modified: 11 Sep 2009, 21:59:21 UTC

a bunch of wus' are ready to report after 2 secs crunch time........what gives?
ID: 30510 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 30511 - Posted: 11 Sep 2009, 22:01:57 UTC - in response to Message 30510.  

Looks like Matt gave me some bad parameter files to start up searches with. Should fix this real quick-like.
ID: 30511 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Lord Tedric
Avatar

Send message
Joined: 9 Nov 07
Posts: 151
Credit: 8,391,608
RAC: 0
Message 30512 - Posted: 11 Sep 2009, 22:04:18 UTC - in response to Message 30511.  

Looks like Matt gave me some bad parameter files to start up searches with. Should fix this real quick-like.


Now they're starting to Abort....
ID: 30512 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 30513 - Posted: 11 Sep 2009, 22:08:49 UTC - in response to Message 30512.  

I stopped the bad searches that were giving the really small WUs. All the ones running now should be of the right size.
ID: 30513 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Lord Tedric
Avatar

Send message
Joined: 9 Nov 07
Posts: 151
Credit: 8,391,608
RAC: 0
Message 30514 - Posted: 11 Sep 2009, 22:14:06 UTC - in response to Message 30513.  
Last modified: 11 Sep 2009, 22:18:26 UTC

I stopped the bad searches that were giving the really small WUs. All the ones running now should be of the right size.


seems to be back to normal ........
although run times are shorter, hence the lower grated credit?
ID: 30514 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Travis
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 30 Aug 07
Posts: 2046
Credit: 26,480
RAC: 0
Message 30518 - Posted: 11 Sep 2009, 22:45:37 UTC - in response to Message 30514.  

I stopped the bad searches that were giving the really small WUs. All the ones running now should be of the right size.


seems to be back to normal ........
although run times are shorter, hence the lower grated credit?


Fixed this too. All the new workunits should be the right size and right credit.
ID: 30518 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Crunch3r
Volunteer developer
Avatar

Send message
Joined: 17 Feb 08
Posts: 363
Credit: 258,227,990
RAC: 0
Message 30523 - Posted: 11 Sep 2009, 23:35:01 UTC - in response to Message 30518.  

All the new workunits should be the right size and right credit.


The size of the new wus is the same now as before but credit is lower.

Compared to the de_s222_3s_best WUs that took the same time to complete and granted ~74 credits, the new ones only get like 53 credits...

What's up with that ?

Join Support science! Joinc Team BOINC United now!
ID: 30523 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
Message 30524 - Posted: 12 Sep 2009, 0:02:52 UTC - in response to Message 30523.  
Last modified: 12 Sep 2009, 0:07:00 UTC

The size of the new wus is the same now as before but credit is lower.

Compared to the de_s222_3s_best WUs that took the same time to complete and granted ~74 credits, the new ones only get like 53 credits...

What's up with that ?

There is a news about this ;)
Basically MW joins SETI and GPUGrid with the value for the credit multiplier, just doubled here for double precision.

As mentioned there, I'm preparing new apps for ATI (and CPU). They should be calculating more accurate (better than the stock app or all apps out there in the moment). They will probably be ready on monday and will compensate the credit loss. How sounds 47 seconds on a HD4870? :D

Running Milkyway@home ATI GPU application version 0.20 by Gipsel
scaling the wait times with 1.07
CPU: AMD Phenom(tm) II X4 940 Processor (4 cores/threads) 3.00006 GHz (400ms)

CAL Runtime: 1.3.145
Found 1 CAL device

Device 0: ATI Radeon HD 4800 (RV770) 512 MB local RAM (remote 28 MB cached + 256 MB uncached)
GPU core clock: 770 MHz, memory clock: 400 MHz
800 shader units organized in 10 SIMDs with 16 VLIW units (5-issue), wavefront size 64 threads
supporting double precision

1 WUs already running on GPU 0
Starting WU on GPU 0

main integral, 320 iterations
predicted runtime per iteration is 136 ms (33.3333 ms are allowed), dividing each iteration in 5 parts
borders of the domains at 0 320 640 960 1280 1600
Calculated about 8.22242e+012 floatingpoint ops on GPU, 1.23583e+008 on FPU. Approximate GPU time 47.1094 seconds.

probability calculation (stars)
Calculated about 3.60304e+009 floatingpoint ops on FPU.

WU completed.
CPU time: 1.73438 seconds, GPU time: 47.1094 seconds, wall clock time: 92.242 seconds, CPU frequency: 3.00011 GHz
ID: 30524 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Dan T. Morris
Avatar

Send message
Joined: 17 Mar 08
Posts: 165
Credit: 410,228,216
RAC: 0
Message 30529 - Posted: 12 Sep 2009, 1:55:30 UTC
Last modified: 12 Sep 2009, 1:58:18 UTC

You know every time we get this project all smoothed out and the work is flowing nicely to the tune of oh say 400 teraflops, and the credits make you have to stay...Then the politics start! Too much this too much that. Really folks...Can't a project just for once leave a good thing alone?

Oh and advance notice of the drop in credits would also be nice, that way we can have the choice of running with the project or just shutting down our systems.

Hey could someone tell me when the DA will have all the projects at equal rate and we are doing 56 hours of work for the credit of 10 per wu.

I know that this seems like I am a little poed, I am.


Just my thoughts. Ok guys I too have now put on my flame suit.
ID: 30529 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile KWSN imcrazynow
Avatar

Send message
Joined: 22 Nov 08
Posts: 136
Credit: 319,414,799
RAC: 0
Message 30535 - Posted: 12 Sep 2009, 2:36:17 UTC - in response to Message 30524.  


As mentioned there, I'm preparing new apps for ATI (and CPU). They should be calculating more accurate (better than the stock app or all apps out there in the moment). They will probably be ready on monday and will compensate the credit loss. How sounds 47 seconds on a HD4870? :D


Once again CP...You da man!

4870 GPU
4870 GPU
ID: 30535 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 30604 - Posted: 12 Sep 2009, 17:56:22 UTC - in response to Message 30529.  

Oh and advance notice of the drop in credits would also be nice, that way we can have the choice of running with the project or just shutting down our systems.

Hey could someone tell me when the DA will have all the projects at equal rate and we are doing 56 hours of work for the credit of 10 per wu.

I know that this seems like I am a little poed, I am.

The only reasons I get peeved about the whole credit thing is that firstly it was supposed to be based on the fixed capability of a standard system. Faster computers and more efficient applications would just mean that you would get that earning faster.

Secondly, the only pressure ever exerted is on those projects that are felt to be overpaying. There is never any effort expended to get projects that are on the other end of the spectrum to pay up ...

Were truth to be told, both are equally bad if you are supposedly in favor of credit parity ...
ID: 30604 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Cluster Physik

Send message
Joined: 26 Jul 08
Posts: 627
Credit: 94,940,203
RAC: 0
Message 30614 - Posted: 12 Sep 2009, 20:48:39 UTC - in response to Message 30604.  

Secondly, the only pressure ever exerted is on those projects that are felt to be overpaying. There is never any effort expended to get projects that are on the other end of the spectrum to pay up ...

Were truth to be told, both are equally bad if you are supposedly in favor of credit parity ...

Probably you missed that, but I tried to talk to several projects in that direction. And by the way, it was succesful after a while when I came up with good reasons for a change. Every project has seen after a while that server determined fixed credits are best and if they are doing some kind of flops counting they should orient them on the "standard" of SETI.
The idea for the latter actually came after reading from DA that if projects would do so they would probably pay less than optimized SETI. In my opinion this isn't exactly true, especially if you look at GPU applications. So I talked to the GPUGrid staff to raise their credits to the SETI level. That was the ~50% increase some months ago. MW is the second project using the same approach to credits as SETI and still paying more.

Other projects were Spinhenge and Docking which went from benchmark based credits to server side credits connected with an increase. So yes, there is pressure on low paying projects to raise it. One has to be patient, talk to the project guys and finally convince them. This works in both directions.
ID: 30614 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Paul D. Buck

Send message
Joined: 12 Apr 08
Posts: 621
Credit: 161,934,067
RAC: 0
Message 30623 - Posted: 12 Sep 2009, 22:17:53 UTC - in response to Message 30614.  

Other projects were Spinhenge and Docking which went from benchmark based credits to server side credits connected with an increase. So yes, there is pressure on low paying projects to raise it. One has to be patient, talk to the project guys and finally convince them. This works in both directions.

Only 50 other projects to go I guess...

I will admit that I did not see the latter two projects for the simple reason I dropped them because they did pay so low and the work was either intermittent or had other problems.

And for the moment I am concentrating on WCG 2 year badges so it will be a short time before I go looking ...
ID: 30623 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote
Profile Dan T. Morris
Avatar

Send message
Joined: 17 Mar 08
Posts: 165
Credit: 410,228,216
RAC: 0
Message 30628 - Posted: 12 Sep 2009, 23:39:01 UTC

Well I pulled a Travis trick.. I changed my credit status without telling him. 1st by shutting down my systems on this project until CP has the other app. The other side of the coin is that how can you have cross parity when not every project uses ATI. They all seem to love those Nvidia over priced cards.

Come on folks. Those of you who were doing 100k per day or over.. Shut them down until the better app is out or the credits are changed back to what they were.

Here is one more example of the cost being put back on us with out asking.. For some months I was doing 600k per day then I decided to buy 4 more cards. That put me at 1 mil. Then two days latter Travis lowers the credit and I am now back to 600 k and 4oo.oo dollars more of cards. Am I a happy camper? NOT!!!!!!!!!
ID: 30628 · Rating: 0 · rate: Rate + / Rate - Report as offensive     Reply Quote

Message boards : Number crunching : unusual behaviour

©2024 Astroinformatics Group