Best Machine Spec for simulation used in Window 7
211 views
Skip to first unread message
Nat Ong
May 30, 2012, 3:51:29 AM5/30/12
to FDS and Smokeview Discussions
I have run a FDS model,approximately 3 million cells by using linking
2 Intel Xeon E3 machines of 2.8GHz, 8MB cache,12GB RAM and 4 cores
however it tooks me 4 days to complete it while the speed of
simulation is faster by run it after linking two machines of Intel(R)
Core i5-2400 C...@3.1Ghz,6MB cache, 4GB RAM and 4 cores. i am planning
to buy 4 more machines of higher spec to run my big model,says
11million cells. Have anyone come across or use the following machines
before? My plan is to choose a high spec machine that could run a big
model (say >10mil cells) by itself.
Just copy the computer spec from HP website as the following machines
attract my eye balls. Have anyone know/ have any experience running
big FDS model on a single machine before? Any suggested computer spec?
Intel® Xeon® E3-1280 (3.50 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1270 (3.40 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1245 (3.30 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1240 (3.30 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1230 (3.20 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1225 (3.10 GHz, 6 MB cache, 4 cores)
I am hoping to skip running parallel as the linking of machines part
always causing a big headache to me.
Thanks,
Natalie
2 Intel Xeon E3 machines of 2.8GHz, 8MB cache,12GB RAM and 4 cores
however it tooks me 4 days to complete it while the speed of
simulation is faster by run it after linking two machines of Intel(R)
Core i5-2400 C...@3.1Ghz,6MB cache, 4GB RAM and 4 cores. i am planning
to buy 4 more machines of higher spec to run my big model,says
11million cells. Have anyone come across or use the following machines
before? My plan is to choose a high spec machine that could run a big
model (say >10mil cells) by itself.
Just copy the computer spec from HP website as the following machines
attract my eye balls. Have anyone know/ have any experience running
big FDS model on a single machine before? Any suggested computer spec?
Intel® Xeon® E3-1280 (3.50 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1270 (3.40 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1245 (3.30 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1240 (3.30 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1230 (3.20 GHz, 8 MB cache, 4 cores)
Intel® Xeon® E3-1225 (3.10 GHz, 6 MB cache, 4 cores)
I am hoping to skip running parallel as the linking of machines part
always causing a big headache to me.
Thanks,
Natalie
Xiaoyun Wang
May 30, 2012, 3:57:40 AM5/30/12
to fds...@googlegroups.com
May i ask if you want to skip parallel running͵ then whats the 4 cores for?
Regards͵
--
You received this message because you are subscribed to the Google Groups "FDS and Smokeview Discussions" group.
To post to this group, send email to fds...@googlegroups.com.
To unsubscribe from this group, send email to fds-smv+u...@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/fds-smv?hl=en.
Erichovi
May 31, 2012, 2:10:23 AM5/31/12
to fds...@googlegroups.com
I'm guessing Nat is talking about not running a single model on multiple computers.
To unsubscribe from this group, send email to fds-smv+unsubscribe@googlegroups.com.
Nat Ong
May 31, 2012, 9:41:06 PM5/31/12
to FDS and Smokeview Discussions
Hi Xiao Yun/Erichovi,
Yes, you are right Erichovi. The 4 cores of a single workstations are
fantastic as it allow me to run a big model however it cant support a
11 million cells fds model by only utilizing its 4 cores of 2.8GHz,
8MB cache,12GB RAM each. Understood that this would be possible by
linking up 2 workstations and run it but i am hoping to avoid the
linking. Do you guys have any ideas how to achieve this or any
recommendations on pc spec that i should be looking at?
Thanks,
Nat
On May 31, 2:10 pm, Erichovi <eco.carls...@gmail.com> wrote:
> I'm guessing Nat is talking about not running a single model on multiple
> computers.
>
>
>
> On Wednesday, 30 May 2012 17:57:40 UTC+10, ivy wrote:
>
> > May i ask if you want to skip parallel running͵ then whats the 4 cores
> > for?
> > Regards͵
> >> fds-smv+u...@googlegroups.com.
>
> - Show quoted text -
Yes, you are right Erichovi. The 4 cores of a single workstations are
fantastic as it allow me to run a big model however it cant support a
11 million cells fds model by only utilizing its 4 cores of 2.8GHz,
8MB cache,12GB RAM each. Understood that this would be possible by
linking up 2 workstations and run it but i am hoping to avoid the
linking. Do you guys have any ideas how to achieve this or any
recommendations on pc spec that i should be looking at?
Thanks,
Nat
On May 31, 2:10 pm, Erichovi <eco.carls...@gmail.com> wrote:
> I'm guessing Nat is talking about not running a single model on multiple
> computers.
>
>
>
> On Wednesday, 30 May 2012 17:57:40 UTC+10, ivy wrote:
>
> > May i ask if you want to skip parallel running͵ then whats the 4 cores
> > for?
> > Regards͵
> >> For more options, visit this group at
> >>http://groups.google.com/group/fds-smv?hl=en.- Hide quoted text -
>
> - Show quoted text -
dr_jfloyd
May 31, 2012, 10:46:49 PM5/31/12
to fds...@googlegroups.com
You will need more RAM than 12 Gb if you want to run 10 million or more cell cases. The HP desktops look like they top out at 16 Gb which may not be enough RAM to do 12 million cells.
> >> fds-smv+unsubscribe@googlegroups.com.
brad c
Jun 1, 2012, 1:55:56 PM6/1/12
to FDS and Smokeview Discussions
Natalie,
I believe there is a difference in the brand of RAM also. For example,
two "identical" systems: Win 7 Pro, 64 bit, 16 GB RAM, 4 i5-2500K cpu
@ 3.30 GHz. fds5_openmp_win64, 4.7 million cells. ~3 week run time for
1800 sec sim. On 1 machine resources about 75%, runs to completion no
problem. On other machine, with the identical input file, resources
jump to ~98% and computer restarts after 4-6 hours. All settings on
both machines matched-- the solution was to add 8 GB RAM to "broken"
machine! Well, there was one final problem-- the last of a long list
of 'head-scratchers'-- at ~1230 sec it halted with Numerical
Instability. It was a large car park with impluse fans, exhaust, 2
large entry/exit, fairly steady since about 300 sec. I was dumping
restart files @ 50 sec-- somebody said "try restarting it". I did and
it ran like a champ from 1200 to 1800 :)
I believe there is a difference in the brand of RAM also. For example,
two "identical" systems: Win 7 Pro, 64 bit, 16 GB RAM, 4 i5-2500K cpu
@ 3.30 GHz. fds5_openmp_win64, 4.7 million cells. ~3 week run time for
1800 sec sim. On 1 machine resources about 75%, runs to completion no
problem. On other machine, with the identical input file, resources
jump to ~98% and computer restarts after 4-6 hours. All settings on
both machines matched-- the solution was to add 8 GB RAM to "broken"
machine! Well, there was one final problem-- the last of a long list
of 'head-scratchers'-- at ~1230 sec it halted with Numerical
Instability. It was a large car park with impluse fans, exhaust, 2
large entry/exit, fairly steady since about 300 sec. I was dumping
restart files @ 50 sec-- somebody said "try restarting it". I did and
it ran like a champ from 1200 to 1800 :)
Nat Ong
Jun 1, 2012, 10:21:21 PM6/1/12
to FDS and Smokeview Discussions
my 11mil cell model was die in the high spec workstation intel Xeon E3
of 4 cores with the error message numerical instable.The activity
shown that its nemory is reaching 90% of my 12GB RAM and i added
another 6GB RAM to it and start the run again.Will update again on
Monday.
Thk.
of 4 cores with the error message numerical instable.The activity
shown that its nemory is reaching 90% of my 12GB RAM and i added
another 6GB RAM to it and start the run again.Will update again on
Monday.
Thk.
Reply all
Reply to author
Forward
0 new messages