[TRNSYS-users] trnsys processing speed benchmark

Gorissen, R. R.Gorissen at student.tue.nl
Mon Nov 13 01:53:51 PST 2006


In addition to Rémi question and Janne answer I ran the same project on 3 different systems; one older athlon 1800+ with 512 mb ddr, an amd athlon 64 X2 dual core 3800+ with 1Gb ddr and on a newer core Duo system (ibm t60 notebook) T2400 with 1.5 Gb ddr2. Unfortunately I do not have a pentium 4 system at hand. 

 

The trnsys project that was used for the test consists of a large heating facility, non standard types, weather and dhw data readers, type 56, various simulation summary types,... (but no external routines)

 

Trnsys 15 was used with the successive solver method.

 

processing time:

 

amd Athlon XP 1800+ (1.53 GHz):  64.23 sec.

amd Athlon X2 3800+ (1 GHz):  57.64 sec.  

amd Athlon X2 3800+ (2 GHz):  29.38 sec.

intel Core Duo T2400 (1.83GHz): 33.67 sec.

 

Even at 1 GHz the newer athon (A64) architecture is faster than the previous generation athlon (XP) at 1.53 GHz. The second core is not used during trnsys simulations !!

 

Ram usage stays well below 20 mb on each run.

 

My guess (based on various processor architectures benchmarks I've seen on the net) is that later generation Pentium 4's (>3GHz)  are comparable to the newer atlhon 64 generation at 2 GHz in non multiple threading applications  (although at a much higher energy consumption!). Newer intel core duo and core 2 duo systems will be quicker than comparable athlon 64 systems (at the same clock speed) and will require a lot less power under load than a Pentium 4 system (less heat -> less noise)

 

The benefit of dual core here is not the decrease in calculation time but in the usability of the system (when a trnsys calculation is running I can still use other applications like matlab or excel without slowing down the trnsys simulation)

 

Regards,

raf

 

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/trnsys-users-onebuilding.org/attachments/20061113/2994101b/attachment-0001.htm>


More information about the TRNSYS-users mailing list