21) Message boards : Number crunching : sugestion for downloading BOINC from LHC website (Message 8306)
Posted 5 Jul 2005 by tannengruen
Post:
> It was the step from 4.19 to 4.20 that caused the problem. The installer was
> changed at this point, and the two installers were completely incompatible.
> If the old version and new version are both on the same side of this divide,
> the install will go without the need for an uninstall.
Ah, thanks for that background information!
Unfortunately it's exactly that "version step" one may be concerned with here at LHC@home: from pre-4.20 to 4.45... So it wouldn't be a fault to add some information about how-to upgrade to 4.45 here on the local download-page, don't you agree?
22) Message boards : Number crunching : sugestion for downloading BOINC from LHC website (Message 8302)
Posted 4 Jul 2005 by tannengruen
Post:
OK, thanks, I got it meanwhile.
But IMO there should be at least a notification that de-installing the client does NOT destroy one's settings/WUs etc. Even on the BOINC-home page there is no info about how-to upgrade an existing installation...
23) Message boards : Number crunching : sugestion for downloading BOINC from LHC website (Message 8300)
Posted 4 Jul 2005 by tannengruen
Post:
There's one big feature still missing in the new BOINC installers: The ability to update an existing installation!
I tried to update my v4.19 with the new v4.45, but the installer refused to work telling me I have to uninstall any prior versions before. :-(
This is quite worse, because at least one workunit (or as many as your client has donwloaded for processing) is lost!
Has anyone yet thought about this at LHC?
24) Message boards : Number crunching : Merging computers don't work anymore :( (Message 8232)
Posted 30 Jun 2005 by tannengruen
Post:
Honestly, I feel a bit imposed on by this. :-/
Why not let the user decide which machines one want's to merge instead of preselecting them by an obviously faulty script?
Maybe one adds/replaces some hardware on a certain machine, even though it's still the same computer, it will appear as a new machine with a new number.
This is really odd IMO.
25) Message boards : Number crunching : Merging computers don't work anymore :( (Message 8227)
Posted 30 Jun 2005 by tannengruen
Post:
> I take it the computer you're talking about is the Celeron 1400?
> The description of all computers has changed a while ago. So where it would be
> an Intel(R) Celeron(TM) CPU 1400MHz Pentium it is now a GenuineIntel
> Intel(R) Celeron(TM) CPU 1400MHz.
Strange about it is, that my system didn't change in any ways since then. It's still the same mainboard, cpu, etc. and even the same windows-install.
Maybe the cpu-recognition inside of BOINC has changed meanwhile?
However, thanks for your reply!
26) Message boards : Number crunching : Merging computers don't work anymore :( (Message 8225)
Posted 30 Jun 2005 by tannengruen
Post:
Hi,

I tried to merge two identical accounts for the same machine in my account settings, but it says:

"Check the computers that are the same as wrkstn1 (created 8 Oct 2004 23:03:40 UTC):

Select all
Unselect all"

Unfortunately there *are* not other computers listed I could select from?! This is no browser-error, tried with firefox and IE, the same results.
Please help, thanks!

regards,
Chris
27) Message boards : Number crunching : Oh my Goddness 1040 Sheeps running round the Database Farm in the last few Hours (Message 4562)
Posted 28 Oct 2004 by tannengruen
Post:
Great, so it must have felt like an Atari ST ;-D
28) Message boards : Number crunching : fortran error.. (Message 4486)
Posted 27 Oct 2004 by tannengruen
Post:
I've had a lot of Fortran errors here since my first participation in LHC@home.
One thing I noticed is the fact that nearly all errors seem to be caused by an access violation.

Under normal conditions something like this shall never happen, even if there are errors in the mathematics inside the sixtrack code.

Errors like this can only happen if you read/write to variables or arrays using pointer and accidently do a read/write attempt outside the reserved memory areas.

So from my point of view it is VERY certain that some bad pointer offsets are leading to such failures. Maybe someone should have a deeper look on the source again.
29) Message boards : LHC@home Science : Is there a site to check LHC scientific results? (Message 3914)
Posted 16 Oct 2004 by tannengruen
Post:
> > > Different results for same wu's?
> >
> > Lets coin the term 'Distributed Quantum Computing'
> > Sometimes a bit is 1.. sometimes it's 0.
> >
> And dont forget the best bit!
>
> I can be both! while being nither! WOOO its logic and you know it =)
>
Nope! The best bit is the one that comes in a sixpack and it has to be stored in your fridge, because it won't fit in your computer's ram for sure... ;D

I think I gonna grab one of these "best bits" now :)
30) Message boards : Number crunching : Fortran Error/Computation Error (Message 3840)
Posted 15 Oct 2004 by tannengruen
Post:
another error here: WU334186 (v64tunescanc9s4_6615_1_sixvf_3231)

error summary (Fortran)
error number error level error count
jwe0019j u 1
total error count = 1
31) Message boards : Number crunching : WU 173268 lost?? (Message 3693)
Posted 12 Oct 2004 by tannengruen
Post:
I saw that - regarding the results list - there have never returned any results for this WU: http://lhcathome.cern.ch/workunit.php?wuid=173268
What's going on there?
32) Message boards : Number crunching : Fortran Error/Computation Error (Message 3477)
Posted 10 Oct 2004 by tannengruen
Post:
> Fortran is pretty effective for calculations, because the code can be
> optimized by the compiler pretty well.
>
> Of course, one thing is that there are so much reusable existing scientific
> code done with fortran that it is often easiest to continue using fortran.

OK, I agree that it is for sure a benefit to have lots of already existing code, but I strongly doubt that the Fortran compilers are better in code optimizing than for example the GNU C++ compiler...

Have you thought about adding some kind of "debug log" to the sixtrack code? Maybe this could make bug tracing a bit easier...
33) Message boards : Number crunching : Fortran Error/Computation Error (Message 3460)
Posted 10 Oct 2004 by tannengruen
Post:
> http://www.ibiblio.org/pub/languages/fortran/unfp.html

Thanks for that link, Guido, but I know about Fortran and the 2 revisions since Fortran 77...
But it still misses features like OOP, for example. IMHO mandatory things that have been included in C++ many years ago.
34) Message boards : Number crunching : Fortran Error/Computation Error (Message 3453)
Posted 10 Oct 2004 by tannengruen
Post:
I also had a crash / fortran error. This happened after I clicked "OK" on that error message:

LHC@home - 2004-10-10 15:37:19 - Started upload of v64lhc1000profour20s10_12563_1_sixvf_25675_0_0
LHC@home - 2004-10-10 15:37:22 - Finished upload of v64lhc1000profour20s10_12563_1_sixvf_25675_0_0
LHC@home - 2004-10-10 15:37:22 - No data transferred

I wonder why so many scientists keep sticked to such an old and inflexible programming language... :-/
35) Message boards : Number crunching : @Markku-thanks you and The Staff for WU s for the Weekend-i wonder how fast analyse old results ;-) (Message 3394)
Posted 9 Oct 2004 by tannengruen
Post:
> german: steht die geistige sonne flach ,werfen selbst zwerge riesen schatten
> english: stands if the mental suns flat, even dwarves giant shade throw
> (translate from babelfish)

So that's the reason? Now I see, thanks! :D
36) Message boards : Number crunching : @Markku-thanks you and The Staff for WU s for the Weekend-i wonder how fast analyse old results ;-) (Message 3390)
Posted 9 Oct 2004 by tannengruen
Post:
Hm, what's your motive to spam these forums so extreme? Are you that bored? ;)
Never mind, just my 2 cents...
37) Message boards : LHC@home Science : Is there a site to check LHC scientific results? (Message 3389)
Posted 9 Oct 2004 by tannengruen
Post:
Yeah, it would be nice to know *what* kind of data already has been processed and some statistical results, maybe as percents of the total number of data.
Would that be a big effort? I have no clue about it *g* ;)


Previous 20


©2024 CERN