Message boards : News : Status, March 2014
Message board moderation

To post messages, you must log in.

AuthorMessage
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 857
Credit: 1,619,050
RAC: 0
Message 26300 - Posted: 16 Mar 2014, 8:43:45 UTC

First, in reply to a recent query about 2014 workload, thanks to Msssimo:
"The majority of the 2014 studies will be devoted to LHC upgrade and the rest to understand the nominal
machine. I do not expect any increase in workload when approaching the LHC re-start in 2015, on the
other hand, we will all be locked up in the control room and the resources for performing the
simulations will be reduced."

Second, we have been experiencing major problems with our
Windows executables for several months now.
There are "small" result differences between Windows and Linux.
After extensive testing I believe they are due to the Windows
ifort compiler. This will be verified and fixed as soon as I
return to CERN next week. In addition new builds of SixTrack
for Windows, which now include a call boinc_unzip, are failing
on Windows in at least two ways; there is a problem parsing the
hardware description (/proc/cpuinfo on Linux) and secondly we
get "cannot Create Process" errors. So, we shall first try and
build without the hopefully resposible call, and fix the result
differences. We can then resume development of the case splitting
to smaller WUs and the return of all results.

It is great that your support continues and, when required, we have
lots of capacity. Saw a new record of over 140,000 WUs in
process a couple of weeks ago. Eric.

ID: 26300 · Report as offensive     Reply Quote
Profile yo2013
Avatar

Send message
Joined: 16 Oct 13
Posts: 59
Credit: 342,408
RAC: 0
Message 26301 - Posted: 16 Mar 2014, 10:30:37 UTC - in response to Message 26300.  

Thanks for keeping us informed, Eric :)
ID: 26301 · Report as offensive     Reply Quote
Profile JLDun
Avatar

Send message
Joined: 5 Jul 07
Posts: 21
Credit: 12,058
RAC: 0
Message 26303 - Posted: 18 Mar 2014, 20:53:25 UTC - in response to Message 26300.  


Second, we have been experiencing major problems with our
Windows executables for several months now.
There are "small" result differences between Windows and Linux.
After extensive testing I believe they are due to the Windows
ifort compiler. This will be verified and fixed as soon as I
return to CERN next week. In addition new builds of SixTrack
for Windows, which now include a call boinc_unzip, are failing
on Windows in at least two ways; there is a problem parsing the
hardware description (/proc/cpuinfo on Linux) and secondly we
get "cannot Create Process" errors. So, we shall first try and
build without the hopefully resposible call, and fix the result
differences. We can then resume development of the case splitting
to smaller WUs and the return of all results.


Does this mean WU reissues in the future, or are the errors consistent enough to be correctable "as-is"?

ID: 26303 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 857
Credit: 1,619,050
RAC: 0
Message 26304 - Posted: 19 Mar 2014, 7:18:38 UTC - in response to Message 26303.  

I a hoping the results are usable; in almost all cases the
"lost" turn number is the same. May have to re-issue one
or two studies. I shall be checking now I am back in Geneva.
Eric.
ID: 26304 · Report as offensive     Reply Quote
Profile Ananas

Send message
Joined: 17 Jul 05
Posts: 102
Credit: 542,016
RAC: 0
Message 26309 - Posted: 23 Mar 2014, 7:09:28 UTC - in response to Message 26304.  

LHC@Home has a large user base, if nothing else helps, you could still configure homogenous redundancy level 2 (the basic 4-classes OS level, ignoring the CPU vendors)
ID: 26309 · Report as offensive     Reply Quote
Eric Mcintosh
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist

Send message
Joined: 12 Jul 11
Posts: 857
Credit: 1,619,050
RAC: 0
Message 26310 - Posted: 24 Mar 2014, 12:46:00 UTC - in response to Message 26309.  

Thanks for that; I would still get slightly different results
but it would be my problem at CERN and would avoid
invalids for you. In fact, it should really help me find the
problem as well. I'll see if I can get that done soonest.
Eric.
ID: 26310 · Report as offensive     Reply Quote
Profile Ananas

Send message
Joined: 17 Jul 05
Posts: 102
Credit: 542,016
RAC: 0
Message 26313 - Posted: 25 Mar 2014, 18:06:45 UTC - in response to Message 26310.  
Last modified: 25 Mar 2014, 18:10:40 UTC

Not even one inconclusive in this batch for me :-)

... still get slightly different results ...


Nothing really changed in the result quality.

Before : The inconclusive workunits became either results with windows rounding or with *ix rounding after the third result has been returned. The third result decides about win or ix.

Now : The workunits, that would have been inconclusive become either results with windows rounding or with *ix rounding. They just skip the temporary "inconclusive" state now. The first result decides about win or ix.
ID: 26313 · Report as offensive     Reply Quote
[TA]Assimilator1
Avatar

Send message
Joined: 29 Nov 13
Posts: 58
Credit: 4,010,807
RAC: 28
Message 26346 - Posted: 13 Apr 2014, 13:59:10 UTC - in response to Message 26309.  

LHC@Home has a large user base, if nothing else helps, you could still configure homogenous redundancy level 2 (the basic 4-classes OS level, ignoring the CPU vendors)

What does that mean?
ID: 26346 · Report as offensive     Reply Quote
Richard Haselgrove

Send message
Joined: 27 Oct 07
Posts: 186
Credit: 3,297,640
RAC: 0
Message 26347 - Posted: 13 Apr 2014, 14:34:11 UTC - in response to Message 26346.  

LHC@Home has a large user base, if nothing else helps, you could still configure homogenous redundancy level 2 (the basic 4-classes OS level, ignoring the CPU vendors)

What does that mean?

Homogeneous Redundancy
ID: 26347 · Report as offensive     Reply Quote
[TA]Assimilator1
Avatar

Send message
Joined: 29 Nov 13
Posts: 58
Credit: 4,010,807
RAC: 28
Message 26380 - Posted: 28 Apr 2014, 18:11:43 UTC - in response to Message 26347.  

Thx :thumbsup:
Team AnandTech - SETI@H, Muon1 DPAD, F@H, MW@H, A@H, LHC@H, POGS, R@H, DHEP, CPDN, E@H.
Main rig - Ryzen 3600, MSI B450 Gm Pro C AC, 32GB DDR4 3200, RX580 8GB, Win10 64bit
2nd rig - i7 4930k @4.1 GHz, 16 GB DDR3 1866, HD 7870XT 3GB(DS), Win7 64bit
ID: 26380 · Report as offensive     Reply Quote

Message boards : News : Status, March 2014


©2024 CERN