Advanced search

Message boards : Number crunching : No bonus on replication of 2

Author Message
Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15606 - Posted: 4 Mar 2010 | 17:20:02 UTC

It seems that a replication of 2 doesn't have any bonus credit. Is this by design?

WU:
1212022
1179453
1174203
1194905
1208413
1200051
1178736
____________

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15610 - Posted: 4 Mar 2010 | 20:21:27 UTC - in response to Message 15606.
Last modified: 4 Mar 2010 | 20:22:10 UTC

It looks like GPUGrid is still using the old rules to send out duplicate WUs when it looks like the original will not get processed in time.
Because most WUs are using the new app there are more returning in the gap between when they send out the second copy and when that second copy is completed but from the original issue it is not inside the bonus window. This odd scenario has been around since bonus time was started but was never particularly common, where it is becoming so now. I have had it happen a few times myself but was trying to be patient and let GPUGrid tweak their project settings to produce the best results for them.

So ... GDF, how about a review of the bonus rules or maybe extend the timeout on the first issue of a WU before sending the second?

Just out of curiosity what is the overall time to return percentage breakdown?
How many in 24 hours, 48 hours, etc.
____________
Thanks - Steve

Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15627 - Posted: 6 Mar 2010 | 1:20:43 UTC

anyone from the staff?

i've been aborting task that has replication of 2.
____________

Alain Maes
Send message
Joined: 8 Sep 08
Posts: 63
Credit: 1,650,510,471
RAC: 2,285,513
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15635 - Posted: 7 Mar 2010 | 8:45:50 UTC - in response to Message 15606.

The design here is that a second copy of the WU is send out (to a fast card?) when there is no reply within the first two days. Reason being that the project heavily depends on a fast turnaround of the results to advance.
The credit given still runs along the old logic:
- if the second fast card replies first and within the two days of its deadline it will get the fixed and bonus credit
- if however the first card replies first (but after more than two days) it will only get the fixed credit and automaticaly set a cap on the max credit for that WU. So when then the second fast card comes back it will only get the capped credit, even if it's reply is within the two day timeframe that normally grants the bonus.

No fun I know, but remember it is the best interest of the project.

Kind regards

Alain

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15666 - Posted: 10 Mar 2010 | 12:24:02 UTC - in response to Message 15635.

It would be in the best interest of the project to change that system!

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15670 - Posted: 10 Mar 2010 | 14:14:35 UTC - in response to Message 15666.

We are looking into the code, as it should work.

gdf

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15686 - Posted: 11 Mar 2010 | 13:33:34 UTC - in response to Message 15670.

The problem is that once a canonical result has been chosen, the credits are not computed in the standard way but just assigned as the credits of the canonical result.
I am looking if we can do anything about that.

gdf

Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15951 - Posted: 24 Mar 2010 | 1:01:13 UTC - in response to Message 15686.

The problem is that once a canonical result has been chosen, the credits are not computed in the standard way but just assigned as the credits of the canonical result.
I am looking if we can do anything about that.

gdf

Any updates on this?

Fast returner got no bonus; Slow returner got bonus. Seems like a good idea for slow returner to have bonus.
Here's the example of piggybacking:
2030062
2030049

Its unfair for fast returner to not have a bonus and also for someone to get a bonus (piggybacking the fast returner).

____________

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15954 - Posted: 24 Mar 2010 | 17:36:03 UTC - in response to Message 15951.

We have worked out a fix.
It will be tested out tomorrow and then if it works used on the main server.

gdf

Profile X-Files 27
Avatar
Send message
Joined: 11 Oct 08
Posts: 95
Credit: 68,023,693
RAC: 0
Level
Thr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15965 - Posted: 24 Mar 2010 | 21:35:11 UTC - in response to Message 15954.

We have worked out a fix.
It will be tested out tomorrow and then if it works used on the main server.

gdf

That's good news, Thanks!
____________

Siegfried Niklas
Avatar
Send message
Joined: 23 Feb 09
Posts: 39
Credit: 144,654,294
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 16030 - Posted: 28 Mar 2010 | 14:39:22 UTC - in response to Message 15954.

We have worked out a fix.
It will be tested out tomorrow and then if it works used on the main server.

gdf


Any news about the fix?

(I'm "not amused" about this one: Workunit 1292423 :credit: 7,954.42

Yesterday I got 11,931.63 for (nearly) the same runtime (same card, same WU-type) Workunit 1295734)

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 16062 - Posted: 29 Mar 2010 | 17:17:50 UTC - in response to Message 16030.

We have patched the software related to this problem. Please check it out.

gdf

[boinc.at] Nowi
Send message
Joined: 4 Sep 08
Posts: 44
Credit: 3,685,033
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 16222 - Posted: 9 Apr 2010 | 14:24:44 UTC

I haven´t gotten the full credit bonus, because the first client had errored out.

See this WU: minor bonus WU

Can you fix that?

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 16223 - Posted: 9 Apr 2010 | 15:53:19 UTC

looks right ot me ... perhaps GDF hopped in and fixed it already?
Claimed: 3,977.21
Granted: 5,965.82
____________
Thanks - Steve

[boinc.at] Nowi
Send message
Joined: 4 Sep 08
Posts: 44
Credit: 3,685,033
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 16227 - Posted: 9 Apr 2010 | 19:36:15 UTC
Last modified: 9 Apr 2010 | 19:41:02 UTC

Sorry!
Snow Crash your right!
My disorientation came from the run time of the WU. I had a WU with nearly the same runtime which had about 1800 credits more (s. here). But here it is a problem with the granted credit, not with the bonus calculation.
It´s my fault, so please forget my last post.

Post to thread

Message boards : Number crunching : No bonus on replication of 2

//