Hi Greg,
thank you so much for your test- so we will install Mavericks on all our
machines now.
yes, I can confirm that Mavericks works fine regarding multiprocessing
(since my desktop machine has it installed as well and I often use
rtrace -n ...).
It is also a wise decision to keep older systems as long as possible -
some simple tools are harder to install on the newer systems. E.g. it
took me several hours to install the netpbm - package (e.g. for
pnmtojpeg ) on Mavericks (but got it finally by the usage of macports).
thanks again!
cheers,
Jan
On 07/24/2015 09:12 PM, Gregory J. Ward wrote:
Hi Jan,
I went ahead and tried it out, and can confirm that as of Yosemite 10.10.4 (the
latest), multi-processing is still very seriously broken. Processes freeze and
you are forced to kill them as the only remaining option.
The good news is that multi-processing seems to work find under Mavericks. I
ran the same test and didn't encounter any problems. I recommend that no one
using Radiance upgrade past 10.9 at this stage.
Cheers,
-Greg
From: "Gregory J. Ward" <gregoryjw...@gmail.com>
Subject: Re: [Radiance-dev] MacOS Yosemite
Date: July 24, 2015 9:04:58 AM PDT
Hi Jan,
It's really difficult to find any information on low-level system calls in Mac OS X.
Unfortunately, searching for "select" comes up with many irrelevant hits, and bug fixes
tend to list applications and behavior rather that root causes. The only really way to tell if the
bug has been fixed is to update to the latest version of Yosemite on one machine and run some tests
using rpiece to see if it hangs. (It's easy enough to test by running "rad -N 4 -v 1
scene.rif" or similar.)
I can perform such a test tomorrow or the next day if no one else has a current
version of Yosemite to try it on. I'm still running Lion, which is the
*oldest* version of Mac OS X that my laptop supports....
Cheers,
-Greg
From: Jan Wienold <jan.wien...@epfl.ch>
Subject: Re: [Radiance-dev] MacOS Yosemite
Date: July 24, 2015 2:36:52 AM PDT
Hi all,
is this problem still existent ? We are currently re-installing several
mac-machines and I'm wondering if I still should go for Maverick to avoid this
problem.
thanks!
Best,
Jan
Am 11/18/14 um 7:15 PM schrieb Andreas Noback:
Hi Greg,
thank you for the quick response. I suspected that it is a bug from Apple and
was hoping that the update today (10.10.1) would change something, but it does
not. In general the Yosemite release seems to me buggier that the last couple
of major updates, so it could be wise to avoid it for a while ...
Best,
Andreas
Am 18.11.2014 um 18:58 schrieb Gregory J. Ward <gregoryjw...@gmail.com>:
Hi Andreas,
A quick follow-up to this. I tried it out on my copy of Yosemite, and
confirmed the problem. It seems to be an intermittent problem with the system
select() call, which means there's little I can do about it except hope Apple
recognizes the issue and posts a patch at some point. The last change I made
in this code related to a hanging condition was in 1997, and it's been working
across Unix implementations since then.
For now, I can only suggest you avoid multiprocessing until the next patch
release. You can try reporting a bug to Apple, but without a simple test case
to reproduce it, they are unlikely to do anything other than register the
complaint. I haven't seen anything on the net about it, yet.
Cheers,
-Greg
From: "Gregory J. Ward" <gregoryjw...@gmail.com>
Subject: Re: [Radiance-dev] MacOS Yosemite
Date: November 18, 2014 8:10:48 AM PST
HI Andreas,
I am sorry to hear about this issue, and it is unlikely that there is any way
to debug it. A hung process doesn't respond to debugging, either!
The best approach is to kill one of the processes using "kill -QUIT" and seeing
if it leaves behind a diagnostic file in $HOME/Library/Logs/CrashReporter/. Then at
least, we may find out what routine it is hanging in.
The only other idea I had is to disable "App Nap" system-wide to see if this is
causing the problem. See:
http://www.defaults-write.com/10-9-disable-app-nap-in-os-x
I don't know why this would affect non-application processes, or why it would
present an issue in 10.10 if it wasn't an issue in 10.9, but it's worth a try.
I will see if I can reproduce this on my copy of Yosemite as well.
Best,
-Greg
From: Andreas Noback <m...@noback.info>
Subject: [Radiance-dev] MacOS Yosemite
Date: November 18, 2014 1:25:27 AM PST
Dear List,
has someone experience with radiance and new MacOS 10.10? I tried it and got
some unpleasant results: If you start rad with the -N option (for example: rad
-N 4 -o x11 -v hem test.rif) it start as usual, but after a few seconds it
seems to stuck, i. e. not responding to input, no further refining, no
processor load. The processes seem to be waiting for something (forever):
1447 s000 S+ 0:00.00 rad -N 4 -o x11 -v hem test.rif
1450 s000 S+ 0:00.03 rvu -vth -vp 0 0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180
-vv 180 -vo 0 -va 0 -vs 0 -vl 0 -dp 512 -
1452 s000 S+ 0:00.25 rvu -vth -vp 0 0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180
-vv 180 -vo 0 -va 0 -vs 0 -vl 0 -dp 512 -
1453 s000 S+ 0:00.24 rvu -vth -vp 0 0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180
-vv 180 -vo 0 -va 0 -vs 0 -vl 0 -dp 512 -
1454 s000 S+ 0:00.24 rvu -vth -vp 0 0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180
-vv 180 -vo 0 -va 0 -vs 0 -vl 0 -dp 512 -
1455 s000 S+ 0:00.28 rvu -vth -vp 0 0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180
-vv 180 -vo 0 -va 0 -vs 0 -vl 0 -dp 512 -
Similar things happen if you use it without -o x11 (rad -N 4 -v hem test.rif):
some process start, parts of the image will be rendered. Than the load goes to
zero and nothing happens any more. Here you can see some zombies:
1315 s000 S+ 0:00.01 rad -N 4 -v hem test.rif
1318 s000 Z+ 0:00.00 (rad)
1325 s000 S+ 0:00.00 rpict -S 1 -PP pfM4wGSj -dp 512 -ar 32 -ms 0.05 -ds
.3 -dt .1 -dc .5 -dr 1 -ss 1 -st .1 -ab
1408 s000 S+ 0:00.00 rad -N 4 -v hem test.rif
1409 s000 S+ 0:00.01 rpiece -F test_hem_rpsync.txt -PP pfM4wGSj -vth -vp 0
0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180 -v
1410 s000 S+ 0:00.00 rpict -S 1 -PP pfM4wGSj -dp 512 -ar 32 -ms 0.05 -ds
.3 -dt .1 -dc .5 -dr 1 -ss 1 -st .1 -ab
1411 s000 Z+ 0:00.00 (rpict)
1412 s000 S+ 0:00.00 rad -N 4 -v hem test.rif
1413 s000 S+ 0:00.01 rpiece -F test_hem_rpsync.txt -PP pfM4wGSj -vth -vp 0
0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180 -v
1414 s000 S+ 0:00.00 rpict -S 1 -PP pfM4wGSj -dp 512 -ar 32 -ms 0.05 -ds
.3 -dt .1 -dc .5 -dr 1 -ss 1 -st .1 -ab
1416 s000 S+ 0:00.01 rpiece -F test_hem_rpsync.txt -PP pfM4wGSj -vth -vp 0
0 0.001 -vd 0 0 1 -vu 0 1 0 -vh 180 -v
1417 s000 S+ 0:00.00 rpict -S 1 -PP pfM4wGSj -dp 512 -ar 32 -ms 0.05 -ds
.3 -dt .1 -dc .5 -dr 1 -ss 1 -st .1 -ab
There is nothing in the logs. I got the same results with the precompiled
binaries and binaries compiled from the head revision under 10.10 with Xcode
6.1.
Any suggestions?
Andreas Noback
_______________________________________________
Radiance-dev mailing list
Radiance-dev@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-dev
_______________________________________________
Radiance-dev mailing list
Radiance-dev@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-dev
--
Dr.-Ing. Jan Wienold
Ecole Polytechnique Fédérale de Lausanne (EPFL)
EPFL ENAC IA LIPID
http://people.epfl.ch/jan.wienold
LE 1 111 (Office)
Phone +41 21 69 30849
_______________________________________________
Radiance-dev mailing list
Radiance-dev@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-dev
_______________________________________________
Radiance-dev mailing list
Radiance-dev@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-dev
--
Dr.-Ing. Jan Wienold
Ecole Polytechnique Fédérale de Lausanne (EPFL)
EPFL ENAC IA LIPID
http://people.epfl.ch/jan.wienold
LE 1 111 (Office)
Phone +41 21 69 30849
_______________________________________________
Radiance-dev mailing list
Radiance-dev@radiance-online.org
http://www.radiance-online.org/mailman/listinfo/radiance-dev
