We are running a .oct through rpiece and we get the 'resolution mismatch in
rpict' error. We have removed -x and -y settings from the .vf files. We
have set the horizontal and vertical angles at 80 degrees and 80 degrees.
Our .opt file sets -x and -y at 2048 and 2048. The only other thought we
have is that the .oct file has been created through adding up various oct
files which have been created using different -r settings in oconv.
Any help appreciated.
Regards
Andrew
Andrew Bissell B.Eng (Hons) C.Eng MIEE
Lighting Consultant
Buro Happold LiT
Canal Suite
Churchgate House
56 Oxford Street
Manchester
M1 6EU
This transmission is confidential and intended solely for the person or
organisation to whom it is correctly addressed. If you are not the intended
recipient of this transmission, you should not take any action in reliance
on it. Further, this transmission may contain confidential design and other
information owned by Buro Happold. You should not copy, distribute, use,
offer for sale or hire such information or in any way infringe the design
and intellectual property rights of Buro Happold.
It is intended that communication by email from Buro Happold or its
employees is limited to communications connected to the services provided by
Buro Happold. Buro Happold accepts no liability for any communications not
connected to the services it provides.
···
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit:
This error sounds suspiciously like a communication problem between rpiece and rpict, since the resolution string is the first thing rpiece tries to read back from rpict. I'm sure it doesn't have anything to do with the octree or the -x and -y options or where they appear. Does rpiece die right away, or only after it's done some work? What machine(s) are you running on? Rpiece talks to its rpict process over a standard Unix pipe/socket, which should be reliable. I'm quite puzzled by this error -- has anyone else on the list gotten this before?
-Greg
···
From: Andrew Bissell <[email protected]>
Date: Wed Nov 26, 2003 10:15:03 AM US/Pacific
We are running a .oct through rpiece and we get the 'resolution mismatch in
rpict' error. We have removed -x and -y settings from the .vf files. We
have set the horizontal and vertical angles at 80 degrees and 80 degrees.
Our .opt file sets -x and -y at 2048 and 2048. The only other thought we
have is that the .oct file has been created through adding up various oct
files which have been created using different -r settings in oconv.
We have encountered this before. Although at the moment I am not sure that I can think of the cause(s). Here are some things to look at:
* are you using a persist file (-PP) - if it is left over some
reason from something else then this can cause problems
* is there an rpict process hanging around - this is also going to
be associated with -PP, this can occur if a rpiece job dies
unexpectedly or is killed and not cleaned up, clean up may need to
occur by hand
* are you using rpiece on something that runs really really fast - I
have noticed that if your run rpiece on a scene that runs super
fast or on a very small image that runs very fast then things can
get confused and die which I believe results in this kind of error
as well.
If you can send further diagnostic information then it may help jar my memory further on the topic.
-Jack
Greg Ward wrote:
···
Hi Andrew,
This error sounds suspiciously like a communication problem between rpiece and rpict, since the resolution string is the first thing rpiece tries to read back from rpict. I'm sure it doesn't have anything to do with the octree or the -x and -y options or where they appear. Does rpiece die right away, or only after it's done some work? What machine(s) are you running on? Rpiece talks to its rpict process over a standard Unix pipe/socket, which should be reliable. I'm quite puzzled by this error -- has anyone else on the list gotten this before?
-Greg
From: Andrew Bissell <[email protected]>
Date: Wed Nov 26, 2003 10:15:03 AM US/Pacific
We are running a .oct through rpiece and we get the 'resolution mismatch in
rpict' error. We have removed -x and -y settings from the .vf files. We
have set the horizontal and vertical angles at 80 degrees and 80 degrees.
Our .opt file sets -x and -y at 2048 and 2048. The only other thought we
have is that the .oct file has been created through adding up various oct
files which have been created using different -r settings in oconv.
This error sounds suspiciously like a communication problem between rpiece and rpict, since the resolution string is the first thing rpiece tries to read back from rpict. I'm sure it doesn't have anything to do with the octree or the -x and -y options or where they appear. Does rpiece die right away, or only after it's done some work? What machine(s) are you running on? Rpiece talks to its rpict process over a standard Unix pipe/socket, which should be reliable. I'm quite puzzled by this error -- has anyone else on the list gotten this before?
-Greg
From: Andrew Bissell <[email protected]>
Date: Wed Nov 26, 2003 10:15:03 AM US/Pacific
We are running a .oct through rpiece and we get the 'resolution mismatch in rpict' error.
I think this only popped up with the official 3.5 release and was fixed by the subsequent HEAD releases.
···
--
Roland Schregle
PhD candidate, Fraunhofer Institute for Solar Energy Systems
RADIANCE Photon Map page: www.ise.fhg.de/radiance/photon-map
I looked back at the release notes, and there was a bug in Radiance 3.1 (released Jan 2002) that caused the -pa option to fail in rpiece. Could this be your problem? I just assumed you were using the latest version. If there is such ann error in the 3.5 official release, this is the first I've heard about it.
-Greg
···
From: Roland Schregle <[email protected]>
Date: Wed Nov 26, 2003 3:50:15 PM US/Pacific
From: Andrew Bissell <[email protected]>
Date: Wed Nov 26, 2003 10:15:03 AM US/Pacific
We are running a .oct through rpiece and we get the 'resolution mismatch in rpict' error.
I think this only popped up with the official 3.5 release and was fixed by the subsequent HEAD releases.
We have tried a few things here to resolve this but with no luck as yet. We
have compiled the latest HEAD distribution. When we run rpiece it waits
about 3 minutes before it comes up with the error, but no files are created
and no other information is written to the command line. We have just taken
a step back and run an older model on the cluster and this ran fine with
rpiece etc.
The model .oct we are trying to run is ~300MB in size, we created it in 7
stages with the -i switch. we do have different -r settings for each .oct
file. Rpeice runs fine with the model on the Linux server machine.
Almost all of the cluster machines are running Linux SuSE 8.1 with a few
running Linux Red Hat 9.0. All have the latest official release and the
server and a few others have today's HEAD distribution.
Your help is appreciated
Regards
Andrew
We are going to rebuild the .oct tonight just in case.
···
-----Original Message-----
From: Greg Ward [mailto:[email protected]]
Sent: 26 November 2003 19:05
To: [email protected]
Subject: Re: [Radiance-general] Resolution Mismatch in Rpict.
Hi Andrew,
This error sounds suspiciously like a communication problem between
rpiece and rpict, since the resolution string is the first thing rpiece
tries to read back from rpict. I'm sure it doesn't have anything to do
with the octree or the -x and -y options or where they appear. Does
rpiece die right away, or only after it's done some work? What
machine(s) are you running on? Rpiece talks to its rpict process over
a standard Unix pipe/socket, which should be reliable. I'm quite
puzzled by this error -- has anyone else on the list gotten this before?
-Greg
From: Andrew Bissell <[email protected]>
Date: Wed Nov 26, 2003 10:15:03 AM US/Pacific
We are running a .oct through rpiece and we get the 'resolution
mismatch in
rpict' error. We have removed -x and -y settings from the .vf files.
We
have set the horizontal and vertical angles at 80 degrees and 80
degrees.
Our .opt file sets -x and -y at 2048 and 2048. The only other thought
we
have is that the .oct file has been created through adding up various
oct
files which have been created using different -r settings in oconv.
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________
We have tried a few things here to resolve this but with no luck as yet. We
have compiled the latest HEAD distribution. When we run rpiece it waits
about 3 minutes before it comes up with the error, but no files are created
and no other information is written to the command line. We have just taken
a step back and run an older model on the cluster and this ran fine with
rpiece etc.
Can you provide the exact error message? (copy and paste).
Both rpiece and rpict can complain about "resolution mismatch",
in different contexts, and the messages vary only slightly.
I don't think we established yet which of the three possible
problems that I can see we're actually talking about.
We are going to rebuild the .oct tonight just in case.
"Resolution mismatch" is about pixel resolution in images,
and has nothing to do with your octree.
-schorsch
···
--
Georg Mischler -- simulations developer -- schorsch at schorsch com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/
-----------
[bhlit@nodea39 191103]# ./royalmillsrpiece
rpict: signal - rpiece: resolution mismatch from rpict
[bhlit@nodea39 191103]#
-----------
Many Thanks
Andrew
-----Original Message-----
From: Georg Mischler [mailto:[email protected]]
Sent: 27 November 2003 17:02
To: '[email protected]'
Subject: RE: [Radiance-general] Resolution Mismatch in Rpict.
Andrew Bissell wrote:
We have tried a few things here to resolve this but with no luck as yet.
We
have compiled the latest HEAD distribution. When we run rpiece it waits
about 3 minutes before it comes up with the error, but no files are
created
and no other information is written to the command line. We have just
taken
a step back and run an older model on the cluster and this ran fine with
rpiece etc.
Can you provide the exact error message? (copy and paste).
Both rpiece and rpict can complain about "resolution mismatch",
in different contexts, and the messages vary only slightly.
I don't think we established yet which of the three possible
problems that I can see we're actually talking about.
We are going to rebuild the .oct tonight just in case.
"Resolution mismatch" is about pixel resolution in images,
and has nothing to do with your octree.
-schorsch
--
Georg Mischler -- simulations developer -- schorsch at schorsch com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________
From your error message, it looks like rpict is dying before it ever gets going. The partial error message "rpict: signal - " is particularly telling. Rpict apparently dies before it even finishes its error message! The "rpiece: resolution mismatch from rpict" is produced by rpiece after rpict dies - it's not part of the same message.
My guess is that rpict is using more memory than it's allowed on your node(s). In csh/tcsh, the command:
% limit
Tells you your current resource limit settings for individual processes. If "datasize" or "memoryuse" are anything besides "unlimited," try typing:
unlimit datasize
unlimit memoryuse
There are some limits that may be imposed by the system administrator, I think, and you may not be able to increase past a certain size, but the above commands should give you the maximum values for each. While you're at it, do the same for the "cputime" variable. I recommend adding the above lines to your script, so you don't have to remember to do it each time you start the script. Alternatively, you could add the above to your .cshrc file if you are running the C-shell, which is NOT the default shell under Linux. (I don't know how to control limits in the other shells -- I assume you can, though.)
If the resource limit settings are not the problem, it may be that you are running out of swap space on your nodes, and you'll need to increase your swap partition or add another swap device/file to get it to run at all.
-Greg
···
From: Andrew Bissell <[email protected]>
Date: November 27, 2003 9:30:40 AM PST
Georg
Error message below: -
-----------
[bhlit@nodea39 191103]# ./royalmillsrpiece
rpict: signal - rpiece: resolution mismatch from rpict
[bhlit@nodea39 191103]#
-----------
From your error message, it looks like rpict is dying before it ever gets going. The partial error message "rpict: signal - " is particularly telling. Rpict apparently dies before it even finishes its error message! The "rpiece: resolution mismatch from rpict" is produced by rpiece after rpict dies - it's not part of the same message.
My guess is that rpict is using more memory than it's allowed on your node(s). In csh/tcsh, the command:
% limit
Tells you your current resource limit settings for individual processes.
ulimit -a
for bash users
you may want to run 'top' (hit m to get into memory mode) to see what happens with rpict.
xosview is another nice way to check what your systems are doing. You should see swap space or memory diminishing before rpict dies, if that is the scenario.
The 'top' command is indeed top, never knew that existed and it proved to be
a great help in solving this problem. The limits for data and infact most
settings were already 'unlimited' when we checked so we were left with
watching 'top' to see what happened.
The machines we were running on generally had 64MB of RAM with 128MB swap
partitions, we added a couple of machines that had 128MB RAM with 1.2GB
swap, anyway when we ran rpeice on the higher spec machines with 'top'
running we saw that rpict peaked at 640MB of RAM use and rpiece / rpict
completed without failing. When we ran rpiece on the lower spec machines
the error 'mismatch resolution etc' occurred exactly as we ran out of RAM
and SWAP space.
It is interesting to note that a ~300MB .oct requires 640MB or RAM and SWAP
space.
Once again thanks for the help, it's good to have found out more about Linux
as well as Radiance. The only problem we have now is rebuilding our cluster
machines to increase the SWAP space.
Best Regards
Andrew
···
-----Original Message-----
From: Peter Apian-Bennewitz [mailto:[email protected]]
Sent: 27 November 2003 23:37
To: [email protected]
Subject: Re: [Radiance-general] Resolution Mismatch in Rpict.
Greg Ward wrote:
Hi Andrew,
From your error message, it looks like rpict is dying before it ever
gets going. The partial error message "rpict: signal - " is
particularly telling. Rpict apparently dies before it even finishes
its error message! The "rpiece: resolution mismatch from rpict" is
produced by rpiece after rpict dies - it's not part of the same message.
My guess is that rpict is using more memory than it's allowed on your
node(s). In csh/tcsh, the command:
% limit
Tells you your current resource limit settings for individual processes.
ulimit -a
for bash users
you may want to run 'top' (hit m to get into memory mode) to see what
happens with rpict.
xosview is another nice way to check what your systems are doing. You
should see swap space or memory diminishing before rpict dies, if that
is the scenario.
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________
________________________________________________________________________
This e-mail has been scanned for all viruses by Star Internet. The
service is powered by MessageLabs. For more information on a proactive
anti-virus service working around the clock, around the globe, visit: http://www.star.net.uk
________________________________________________________________________