Issue 602 in fds-smv: Intel Compiler Flag -heap-arrays: Memory Usage Increase
codesite...@google.com
Owner: ----
Labels: Type-Defect Priority-Medium
New issue 602 by pfh...@ameritech.net: Intel Compiler Flag -heap-arrays:
Memory Usage Increase
http://code.google.com/p/fds-smv/issues/detail?id=602
Application Version: FDS 5.3 Serial
SVN Revision Number: 2994
Compile Date: 1-10-09
Operating System: Linux x86_64 (Fedora Core 9)
Intel 11.0-074 C and Fortan Compilers
I have used the -heap-arrays compiler flag to solve a crash problem
involving large meshes (~1.6M cells). However, I have found that using it
with a file that contains a sprinkler results in memory usage steadily
increasing with no apparent limit. This occurs without any sprinkler
droplets being discharged.
I conducted several runs, removing information from the case file to
determine what portion of the file causes the issue. I believe it is the
sprinkler details. I also ran the case with the NIST compiled FDS5.3 and
the problem does not occur.
The test case I used is attached.
Attachments:
HeapArray.fds 4.2 KB
--
You received this message because you are listed in the owner
or CC fields of this issue, or because you starred this issue.
You may adjust your issue notification preferences at:
http://code.google.com/hosting/settings
codesite...@google.com
Status: Accepted
Owner: bryanwklein
Labels: OpSys-Linux Performance
Comment #1 on issue 602 by mcgratta: Intel Compiler Flag -heap-arrays:
Memory Usage Increase
http://code.google.com/p/fds-smv/issues/detail?id=602
Since FDS 5.3 hasn't been released yet, which NIST compiled FDS 5.3 are you
referring to in the second paragraph?
I am running the case compiled for 32 bit linux and it is using 592 MB
(resident
size), which doesn't appear to be increasing. Is this consistent with what
you are
seeing. If so, the addition of -heap-arrays (and only that) makes the RAM
requirement continually grow?
codesite...@google.com
Comment #2 on issue 602 by pfh...@ameritech.net: Intel Compiler Flag
http://code.google.com/p/fds-smv/issues/detail?id=602
My mistake, I meant to indicate I ran the case with the NIST compiled
FDS5.2.5-32 bit
for linux and it is running with no problems. This run uses 605 MB and
remains steady.
I am experiencing the problem with 64 bit linux and the -heap-arrays. When
I compile
with the flags below the problem does not occur. I have been running the
cases on a
Dell with dual core Xeon processors.
intel_linux_64 : FFLAGS = -O3 -m64 -axSSSE3 -static-intel
intel_linux_64 : CFLAGS = -O3 -m64 -Dpp_noappend
intel_linux_64 : FCOMPL = ifort
intel_linux_64 : CCOMPL = icc
If I compile with the flags below the problem does occur. I do not have the
32 bit
Intel compilers loaded so I cannot compile FDS for 32 bit linux.
intel_linux_64 : FFLAGS = -O3 -m64 -heap-arrays -axSSSE3 -static-intel
intel_linux_64 : CFLAGS = -O3 -m64 -Dpp_noappend
intel_linux_64 : FCOMPL = ifort
intel_linux_64 : CCOMPL = icc