It seems I'm regularly claiming about 1/2 of my 'wingmans' claimed credit.
I feel like I'm 'cheating', in a bad way. ;)
Is this an intended behaviour or is it something I'm doing?
Host 28305
It seems I'm regularly claiming about 1/2 of my 'wingmans' claimed credit.
I feel like I'm 'cheating', in a bad way. ;)
Is this an intended behaviour or is it something I'm doing?
Host 28305
As far as i can see you pretty much get what you asked for.
Which result(s) do you mean?
It seems I'm regularly claiming about 1/2 of my 'wingmans' claimed credit.
I feel like I'm 'cheating', in a bad way. ;)
Is this an intended behaviour or is it something I'm doing?
Host 28305
It's just because the BOINC benchmarks don't match AstroPulse processing very well. Your system does relatively better on AP, so the credit claims based on time and benchmarks are lower than many other systems. Josh will get the fpops method working soon and all hosts will claim the same.
Joe
As far as i can see you pretty much get what you asked for.
Which result(s) do you mean?
Yeah, I get what I ask for...but the other guy/gal doesn't. I'm not being greedy, I just feel I'm robbing them of credit they earned... :)
Great to hear it's that way by design (not my fault), and is being worked on.
/Anton
For a fairly extreme disparity in platforms, check out ap_27fe08ac_B3_P1_00093_20080425_27217:
[url=http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=8617]My G5 Mac[/url] [url=http://setiweb.ssl.berkeley.edu/beta/show_host_detail.php?hostid=17120]Pentium D[/url]
2.3 GHz 3.0 GHz
v4.31* v4.30
1283 1506 Mflops (W)
3395 2360 Miops (D)
151.6 157.5 CPU-hours taken
1477 1269 CS claimed
The discrepancy in claims is not nearly as large as those mentioned upthread, but this gives a taste of the foibles of benchmarking.
* Unofficial PPC build from Dotsch
____________
Yeah, I get what I ask for...but the other guy/gal doesn't. I'm not being greedy, I just feel I'm robbing them of credit they earned... :)
/Anton
Don't
If you are conncerned about credit you should not be here.
This is beta so main objexctive is developing new software and yes that means no gauranty is given that there wil ever be credit granted.
I think Joe (as he usualy is) might be on to somthing.
I also have a XEON running which has claimed significantly less than other machines.
You only qualify for the underclaiming club if you meet or exceed the claims in this thread. LOL.
Seems this thread may be the place to pose the following question, and no I am not complaining about credits, it just seems strange. Check 1147111. There are 2 completed tasks that were granted 669.39 credits and a failed task that was granted 1219.34.
This is also a good example of Mr.kjellen's issue as well. I claimed 1322.26 credits but because my wingman completed it a quarter of the time I took we both recieved his claim of 669.39.
____________
Richard U
Seems this thread may be the place to pose the following question, and no I am not complaining about credits, it just seems strange. Check 1147111. There are 2 completed tasks that were granted 669.39 credits and a failed task that was granted 1219.34.
This is also a good example of Mr.kjellen's issue as well. I claimed 1322.26 credits but because my wingman completed it a quarter of the time I took we both recieved his claim of 669.39.
You need to look at the dates when those different results were returned, also the application versions.
The higher credit was manually granted first, the 2 others appear to have been the result of normal validation.
[edit] added a word [/edit]
Seems this thread may be the place to pose the following question, and no I am not complaining about credits, it just seems strange. Check 1147111. There are 2 completed tasks that were granted 669.39 credits and a failed task that was granted 1219.34.
This is also a good example of Mr.kjellen's issue as well. I claimed 1322.26 credits but because my wingman completed it a quarter of the time I took we both recieved his claim of 669.39.
The high claiming tasks in that unit were processed with versions containing the debugging code, so would have taken twice as long as for v4.33.
And until the app is made to do Flop counting claims, the benchmark * time method is always going to produce very different claims depending on cpu type and OS.