Problem with avlmemi

Hi all,

we installed the 3.4 RADIANCE and get now in many cases following error message:

rpict: consistency - address not found in avlmemi

We also tried to store an ambient file and to restart the job with it - with the same result.

If we use our "older" version (I think it is 3.1 patchlevel 11), this problem does not occur.

Any idea how to solve that?

Jan

···

--
-------------------------------------------------------------------
   o -
  > \ Jan Wienold
                 Fraunhofer Institute for Solar Energy Systems
                 SOLAR BUILDING DESIGN GROUP - DAYLIGHTING
                 Heidenhofstr. 2
                 79110 Freiburg
                 GERMANY

                 Phone: ++49 - (0) 761 - 4588 -5- 133
                 Fax: ++49 - (0) 761 - 4588 -9- 133
                 Email: [email protected]

                 Internet:
                 http://www.ise.fhg.de

Jan Wienold wrote:

Hi all,

we installed the 3.4 RADIANCE and get now in many cases following error
message:

rpict: consistency - address not found in avlmemi

We also tried to store an ambient file and to restart the job with it - with
the same result. If we use our "older" version (I think it is 3.1 patchlevel
11), this problem does not occur.

Any idea how to solve that?

Thats one of the signs to tell if a mailing list is authoritative
on a topic. It is, when questions can be answered by pointing to the
archives: :wink:

http://www.radiance-online.org/pipermail/radiance-general/2001-July/000017.html

Have fun!

-schorsch

···

--
Georg Mischler -- simulations developer -- schorsch at schorsch.com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/

Jan Wienold wrote:

Hi all,

we installed the 3.4 RADIANCE and get now in many cases following error message:

rpict: consistency - address not found in avlmemi

We also tried to store an ambient file and to restart the job with it - with the same result.

If we use our "older" version (I think it is 3.1 patchlevel 11), this problem does not occur.

Any idea how to solve that?

Howdy Jan -- long time no see! :^)

Oh yeah... the age-old avlmemi bug... the plight of any RADIANCE
developer! This problem popped up on patchlevel 20. I circumvented it by
simply disabling the call to sortambvals() at line #230 in the ambient.c
module:

#if 0
   if (tracktime) /* sort to minimize thrashing
*/ sortambvals(0);
#endif
   
Runs fine without it! Maybe Greg can shed some light on why
sortambvals() is needed. As far as I can see this should be redundant on
modern VM systems.

See ya!

···

--
"Life is too short for core dumps"

Hi:

Ambient value sorting, sortambvals(), can be turned of
at compile time (rather than messing with the source
code) with the following addition to rmake in the
MACH= line:

-Dtracktime=0

for example (for compilation on a linux box):

#!/bin/sh
exec make "SPECIAL=" \
        "OPT=-O2 -DSPEED=40" \
        "MACH=-DBSD -Dlinux -DDCL_ATOF
-L/usr/X11R6/lib -I/usr/include/X11 -DNOSTEREO
-Dtracktime=0 -DMAXOBJBLK=131071 -DBIGMEM" \
        ARCH=IBMPC "COMPAT=bmalloc.o erf.o
getpagesize.o" \
        INSTDIR=/usr/local/bin \
        LIBDIR=/usr/local/lib/ray \
        CC=gcc "$@" -f Rmakefile

Regards,

-Jack de Valpine

Jan Wienold wrote:
>
> Hi all,
>
> we installed the 3.4 RADIANCE and get now in many

cases following error message:

>
> rpict: consistency - address not found in avlmemi
>
> We also tried to store an ambient file and to

restart the job with it - with the same result.

>
> If we use our "older" version (I think it is 3.1

patchlevel 11), this problem does not occur.

>
> Any idea how to solve that?

Howdy Jan -- long time no see! :^)

Oh yeah... the age-old avlmemi bug... the plight of

any RADIANCE

developer! This problem popped up on patchlevel 20.

I circumvented it by

simply disabling the call to sortambvals() at line

#230 in the ambient.c

module:

#if 0
   if (tracktime) /* sort

to minimize thrashing

*/ sortambvals(0);
#endif
   
Runs fine without it! Maybe Greg can shed some light

on why

sortambvals() is needed. As far as I can see this

should be redundant on

modern VM systems.

See ya!

--
"Life is too short for core dumps"
_______________________________________________
Radiance-general mailing list
[email protected]

http://www.radiance-online.org/mailman/listinfo/radiance-general

···

---- [email protected] wrote:

SIGH...

I wish to heaven they would fix the qsort() routine so it would stop haunting me... How can something so basic and necessary remain broken for so many years??

I should really have put the -Dtracktime=0 switch into the makeall script for Linux, but since it shows up in a number of other places that use the GNU libraries, this probably wouldn't fix everyone's problem. I thought the Linux folks had repaired it, to be honest.

It's true that modern VM systems are quite good when it comes to swapping, but also true that highly fragmented memory access is always a performance hit when you run out of RAM, so I'm reluctant just to remove the optimization altogether. It does work on systems that don't use the GNU libraries.... At the very least, I should recompile the Linux binaries on the website.

Have people been getting this error with the compiled binaries, or just ones they compiled themselves from the 3.4 source?

-Greg

···

Jan Wienold wrote:

Hi all,

we installed the 3.4 RADIANCE and get now in many cases following error
message:

rpict: consistency - address not found in avlmemi

We also tried to store an ambient file and to restart the job with it - with
the same result. If we use our "older" version (I think it is 3.1 patchlevel
11), this problem does not occur.

Any idea how to solve that?

Thats one of the signs to tell if a mailing list is authoritative
on a topic. It is, when questions can be answered by pointing to the
archives: :wink:

http://www.radiance-online.org/pipermail/radiance-
general/2001-July/000017.html

Have fun!

-schorsch

--
Georg Mischler -- simulations developer -- schorsch at schorsch.com
+schorsch.com+ -- lighting design tools -- http://www.schorsch.com/

Greg,

FYI - We got this error with binaries we compiled ourselves from the 3.4
source. We use Suse 7.3 if that helps. Thanks everyone for the insight into
this problem! We have everything running correctly now.

~Zack

···

[email protected] wrote:

Send Radiance-general mailing list submissions to
        [email protected]

To subscribe or unsubscribe via the World Wide Web, visit
        http://www.radiance-online.org/mailman/listinfo/radiance-general
or, via email, send a message with subject or body 'help' to
        [email protected]

You can reach the person managing the list at
        [email protected]

When replying, please edit your Subject line so it is more specific
than "Re: Contents of Radiance-general digest..."

Today's Topics:

   1. Re: Problem with avlmemi (Greg Ward)

--__--__--

Message: 1
Date: Fri, 8 Mar 2002 14:30:40 -0800
From: Greg Ward <[email protected]>
To: [email protected]
Subject: [Radiance-general] Re: Problem with avlmemi
Reply-To: [email protected]

SIGH...

I wish to heaven they would fix the qsort() routine so it would stop
haunting me... How can something so basic and necessary remain broken
for so many years??

I should really have put the -Dtracktime=0 switch into the makeall
script for Linux, but since it shows up in a number of other places that
use the GNU libraries, this probably wouldn't fix everyone's problem. I
thought the Linux folks had repaired it, to be honest.

It's true that modern VM systems are quite good when it comes to
swapping, but also true that highly fragmented memory access is always a
performance hit when you run out of RAM, so I'm reluctant just to remove
the optimization altogether. It does work on systems that don't use the
GNU libraries.... At the very least, I should recompile the Linux
binaries on the website.

Have people been getting this error with the compiled binaries, or just
ones they compiled themselves from the 3.4 source?

-Greg

> Jan Wienold wrote:
>
> Hi all,
>
> we installed the 3.4 RADIANCE and get now in many cases following error
> message:
>
> rpict: consistency - address not found in avlmemi
>
> We also tried to store an ambient file and to restart the job with it -
> with
> the same result. If we use our "older" version (I think it is 3.1
> patchlevel
> 11), this problem does not occur.
>
> Any idea how to solve that?
>
>
> Thats one of the signs to tell if a mailing list is authoritative
> on a topic. It is, when questions can be answered by pointing to the
> archives: :wink:
>
> http://www.radiance-online.org/pipermail/radiance-
> general/2001-July/000017.html
>
>
> Have fun!
>
> -schorsch
>
> --
> Georg Mischler -- simulations developer -- schorsch at schorsch.com
> +schorsch.com+ -- lighting design tools -- http://www.schorsch.com/

--__--__--

_______________________________________________
Radiance-general mailing list
[email protected]
http://www.radiance-online.org/mailman/listinfo/radiance-general

End of Radiance-general Digest

--
Celebrating 20 Years of Improving Building Energy Performance

Zack Rogers
Daylighting Designer/Engineer
Architectural Energy Corporation
2540 Frontier Avenue, Suite 201
Boulder, CO 80301 USA

tel (303)444-4149 ext.235
fax (303)444-4304