credits

Message boards : Number crunching : credits

To post messages, you must log in.

AuthorMessage
Tom Philippart

Send message
Joined: 15 Sep 07
Posts: 9
Credit: 44,303
RAC: 0
Message 258 - Posted: 21 Nov 2007, 17:44:20 UTC

*** I know this topic will create fights about fair/unfair credits. Please try to avoid them this time. Thanks ***

You can look at my computer here.
As you see, there's only one result that has a higher claimed credit value than granted credit value.

Could you please higher the credit distribution in order to be on a similar level than other projects?

Thanks!
(Remember: I'm not crunching for credits, but for the science, still the stats/credits keep it fun in the long run.)
ID: 258 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 260 - Posted: 21 Nov 2007, 20:17:11 UTC - in response to Message 258.  

Enigma is using fixed credits, it doesn't care about the claims you make.
Credit given is 8.35 (for _0_), 15.26 (for _1_) and 17.85 (for _2_).
Jord.

BOINC FAQ Service.
ID: 260 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 267,994,998
RAC: 0
Message 261 - Posted: 21 Nov 2007, 23:43:16 UTC - in response to Message 260.  
Last modified: 22 Nov 2007, 0:03:26 UTC

Runtimes are predictable, so I calibrated credits with benchmarks and timings from my reference machine (Athlon TB 1400), using the standard app. I forgot the exact value, but it's around 7-7.5 credits/hour, which is quite high already. If you use cpu which doubles my reference machine's performance, you'll get same amount of credit, but the workunit runtime will drop to half of the original value, so you'll get twice as much credits/time. I believe it's fair.
Btw, have you tried the optimized app ? :->


M4 Project homepage
M4 Project wiki
ID: 261 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Tom Philippart

Send message
Joined: 15 Sep 07
Posts: 9
Credit: 44,303
RAC: 0
Message 266 - Posted: 22 Nov 2007, 14:36:48 UTC - in response to Message 261.  

@ageless:
I know it is using a fixed credit system, however it should always be around the claimed credit.

@TJM:
I'm using the optimized app, I love it!!
ID: 266 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile Ageless
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 11 Sep 07
Posts: 104
Credit: 155,932
RAC: 0
Message 267 - Posted: 22 Nov 2007, 14:49:47 UTC - in response to Message 266.  

@ageless:
I know it is using a fixed credit system, however it should always be around the claimed credit.

Not necessarily. For what if you then increase your measured integer speed artificially? That's how the claimed credit still works here: benchmarks times time taken.

Enigma doesn't use floating point operations that much, so even setting the amount of fpops taken times a multiplier won't give a claimed credit anywhere near what is granted. The fp_benchmark_weight is set to 0.1, so it has no impact at all.
Jord.

BOINC FAQ Service.
ID: 267 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote
Profile TJM
Project administrator
Project developer
Project scientist
Avatar

Send message
Joined: 25 Aug 07
Posts: 843
Credit: 267,994,998
RAC: 0
Message 268 - Posted: 22 Nov 2007, 15:14:19 UTC - in response to Message 267.  

BOINC (default) benchmark doesn't show the real CPU speed. A quick example: Athlon Thunderbird 1400 c2d 1.86
The second one has only 50% higher benchmarks while in fact it's usually 3-5 times faster, depends on project.

M4 Project homepage
M4 Project wiki
ID: 268 · Rating: 0 · rate: Rate + / Rate - Report as offensive    Reply Quote

Message boards : Number crunching : credits




Copyright © 2024 TJM