Currently, I am just using the quotient of the claimed_credit and model totals for each work unit batch which get saved in a project specific table that we created. I wanted to avoid having to query the result table which would be necessary to get the median. I'd also have to add a project specific column to the result table which would hold the model count and I'm trying to avoid modifying the BOINC tables. I could use a correction factor if the descrepancies are significant enough. Can you point me to the results that you are talking about?

I'm extracting the archived data and will see what I can do. What do other users think about trying to backdate the credits using the new work based system?

can't read ../cache/65/index.php; lastmod 1145943518

# random seed: 3985987
No heartbeat from core client for 31 sec - exiting

well if that was the case would it not have errored out straight after the benchmarks??

But seen as though BOINC left it in memory at the benchmark that can't be the reason for the failure can it?

Would any of those running Ralph be willing to voice their opinion on whether they'd volunteer this info to help track down these annoying bugs?

You know what is it an alpha or beta application?

I crunched several workunit without success and with 0 credits but i'm happy to help to sort these issues :)

P.S.

to the admins :
I suggest to don't give credits at all. it's meaningless in an alpha project and there will ever be users that will not comprend why the application crash and why there are no workunits
____________

Thanks, however more than *one* pc has erroed out some of my results too
*may be, not every alphatester had the patience and time to report here.
Click on the results I posted, and then, on each one, click Workunit
I did this only for a few ... this one for example ...
http://ralph.bakerlab.org/result.php?resultid=14553
*It reports some stackwalker uninitialized *sure cause of sigsegv
btw: my smartd daemon is not reporting any errors on my hda -:)

Signal: Segmentation Violation (SegmentationFault)

This is raised when the program attempts has a bad memory reference such as:

* The pointer is NULL.
* Address not mapped to object (eg, the memory is unallocated, and unmapped by the OS)
* Invalid Permission for mapped object (accessing memory that permissions deny).

This is almost invariably a programming fault.

The default action for this signal is to cause the program to terminate and dump core.

A classic example is to dereference a pointer in C that is either uninitialised, or has already been freed. Here is some C code:

#include <stdio.h>
int main(void) {
int *pointer;

pointer = 0;

printf("Value pointed to by pointer is %dn",
*pointer /* this will cause SEGV */
);
}

I think that the values are being overridden, I updated my prefs to 1 hour then updated BOINC but am still getting 8 hour WU's.

I got another, I had a look at the slots n stdout.txt and found it to be 80,562KB and full of

WARNING:: fullatom scorefxn terms requested
but fullatom flag is false
fullatom terms will not be evaluated


error messages.

Maybe your making the stdout file(s) too big and causing the BOINC disk space to be used up?

If you are correct we need to know. Could you set your time parameter to 4 hours and see if that fixes the problem? If not then try 2 hours

Thanks!

I think i have now understood what the difference is:


"
Rosetta: Faster but less accurate(Bakers analogy sounds to me like he uses kind of brute-force-like methods as used in passwordcrackers); still in development to get more precise answers without losing the speed-advantage, basing on Bakers & team development

F@H, much much slower, but precise enough to do research with it, based on GROMACS
"


Is this right?

I still think it would be a nice idea to implement F@H in BOINC.
Or is it already?
-
And what means troll?
My English is horrible, sorry...

WARNING:: fullatom scorefxn terms requested
but fullatom flag is false
fullatom terms will not be evaluated