Posts by Billy

11)

Message boards : Number crunching : Previous WU's

( Message 1939 )
Posted 3870 days ago by Profile Billy
I regret that this might still be the case, but that has nothing to do with the fact that we changed the seed to have a fixed value. Memo is currently working on the architectural difference problem on Macs. Hopefully he will find a solution for this soon.

Thanks
Andre


Alright. The most likely result is that I will get zero credit and the other 2 will get full credit. Oh well.


I misunderstood. I thought that the new WU's were an attempt to solve the problems with the validation and credit of WU's. I was even thinking of suggesting to you that you manually grant credits to all of the old WU's and start the new WU's to save some time.
12)

Message boards : Number crunching : Previous WU's

( Message 1936 )
Posted 3871 days ago by Profile Billy
Please keep on crunching all of the wu's that are send out. The workunits starting from 19724 are an experiment that we can wait for a couple of days. If workunits with a smaller id are aborted, credit might not be awarded to people who crunched other results in that workunit.

Thanks
Andre

I just downloaded some WU's and the highest number is 19615. Shoud I discard these and wait until the newer, i.e. WU 19724 and higher are sent out?


Alright. The most likely result is that I will get zero credit and the other 2 will get full credit. Oh well.
13)

Message boards : Number crunching : Previous WU's

( Message 1934 )
Posted 3871 days ago by Profile Billy
I just downloaded some WU's and the highest number is 19615. Shoud I discard these and wait until the newer, i.e. WU 19724 and higher are sent out?
14)

Message boards : Number crunching : zero credits for valid result??

( Message 1890 )
Posted 3878 days ago by Profile Billy
Hi Tim,

This is one of the biggest issues we are working on at the moment: the issue with the homogeneous redundancy. I suspect that you have uncovered another case of this problem: P4's give another result than Core 2's and P D's. We'll add that to our list of platform problems and start working on this after the new years holiday! Please understand that this is a project of almost all part time people (I do 2 days a week 'officially') and many of these are students that are on the project to learn (which they do fast!). So things might take a little longer than usual to get resolved. Hope you understand!

Thanks
Andre

Hi,

I too have noticed that one of my hosts has spent time on Docking WU's and has been awarded zero credits when producing valid resuilts.

http://docking.utep.edu/workunit.php?wuid=12518

Time: 36,792.44 - Credits earned: 80.71 - Credits awarded: 0.00

http://docking.utep.edu/workunit.php?wuid=12545

Time: 36,473.42 - Credits earned: 80.73 - Credits awarded: 0.00

and while I'm happy for these to be "one-off's", as a Team founder, I'm concerned that the project is issuing WU's which may continue to offer no reward.

And as churlish as it might seem, people do expect projects to get it right and to offer some incentive for members....

So, I truly hope that:
1) This issue is resolved soonest
2) Past zero credit WU's are rewarded correctly.

If this can happen, then more crunchers will consider joining up to this project and more research can be done.

So, please make this a New Years resolution for 2007....?

regards

Tim
Founder, UK BOINC Team


Do you think that it may have something to do with all the writing to disk that may introduce small errors somehow? This seems to be the only project that has so much disk writing. Can you alter it at all?
15)

Message boards : Macintosh : macInfo script results

( Message 1848 )
Posted 3884 days ago by Profile Billy

I ran it but got an error message,

$ otool -L "/Library/Application Support/BOINC Data/projects/docking.utep.edu/charmm_5.2_i686-apple-darwin"
-bash: otool: command not found

I don't know terminal. It may be that I am not the owner of the computer, but a user with admin privileges. Boinc is in the other directory perhaps?
16)

Message boards : Macintosh : macInfo script results

( Message 1846 )
Posted 3884 days ago by Profile Billy
Created this thread to make space for the results of the script we send out.

I couldn't get it to work on my computer.

s-computer:~ Bill$ % otool -L "/Library/Application Support/BOINC Data/projects/docking.utep.edu/charmm_5.2_i686-apple-darwin"
-bash: fg: %: no such job

But I don't know Jack about terminal. I got the same error message about the Charmm being missing when I did the first script.
17)

Message boards : Web site : Updated FAQ

( Message 953 )
Posted 3955 days ago by Profile Billy
The FAQ is dated. The Mac worker is up to 5.2(5.02).
18)

Message boards : Number crunching : Pending Credits Problem

( Message 866 )
Posted 3961 days ago by Profile Billy

Should I abort all my present WU's that I am working on?


Might not be a bad idea actually, so that way you won't get anymore for which a cannonical result has been found anymore.

Thanks
Andre

I already got rid of them!
19)

Message boards : Number crunching : Validator and Quorum

( Message 840 )
Posted 3961 days ago by Profile Billy
Once it "goes public", it will be far to late to figure that out.

100% agree. The policy on credits should be established within the testing phase, because it's much easier to change rules (even if they were so radical). Also , when the number of the participants is limited, a consensus can be formed more easily.

For my Intel Mac, the credits are evened out for the projects I am doing: (normalized as credits per 100 CPUseconds)

  • Einstein 112.97/22.4= 3.6219
  • Seti (post Sept. 24) 31.6/9.0= 3.511
  • Seti (pre Sept. 29) 33.9/9.1=3.73

There is continuous updating to get the credits more or less equivalent across the projects and between computers based on the work done.

The credits my machine is getting with Docking is:


  • docking 14.55/5.54= 2.63

For my computer, Docking is giving quite a bit less credit than the other projects.

For Rosetta, there has been a torturous analysis and modification of the credit system. It is now trying to establish a new system that uses an "average" of the results for different computers to come up with the "fair" credit. For example, I got these results for 2 Rosetta WU's:


  • rosetta high 49/14.2= 3.45
  • rosetta low 37/14.2=2.61

For my computer, Rosetta is giving similar or slightly less credit than average for the projects I am crunching at.

20)

Message boards : Number crunching : Pending Credits Problem

( Message 839 )
Posted 3961 days ago by Profile Billy
I've just manually granted credit to all the pending, valid results. Hope that makes some people happy :-) I'll keep an eye on any further pending results that come in.


Do you mean for just the 5.01 WUs?

Also, how long will this take to show up? It is now 3 hours since your post, but I still have many pending 5.01 credits. "Valid State" are "Initial".

Whatever you did, it is now invalidating all the results that I just turned in with 5.2. :-( The WUs are 3510 3475 2312 2704 2702 2877 5937 3613

Should I abort all my present WU's that I am working on?


Next 10 posts