here are the results of my benchmark tests with the new algorithm
B R L - C A D B E N C H M A R K
=================================
Running /usr/code/brlcad-build/bin/benchmark on Thu Jul 25 10:14:19 WAT 2013
Logging output to run-20835-benchmark.log
Linux localhost.localdomain 2.6.32-220.el6.i686 #1 SMP Sat Dec 10 17:47:51
EST 2011 i686 i686 i386 GNU/Linux
Using [/usr/code/brlcad-build/bin/rt] for RT
Using [/usr/code/brlcad-build/bin/../share/db] for DB
Using [/usr/code/brlcad-build/bin/../share/pix] for PIX
Using [/usr/code/brlcad-build/bin/../share/pix] for LOG
Using [/usr/code/brlcad-build/bin/pixcmp] for CMP
Using [/usr/code/brlcad-build/bin/elapsed.sh] for ELP
Using [32] for TIMEFRAME
Using [300] for MAXTIME
Using [3] for DEVIATION
Using [3] for AVERAGE
RT reports the following version information:
BRL-CAD Release 7.24.1 The BRL-CAD Raytracer RT
BRL-CAD Release 7.24.1 The BRL-CAD Ray-Tracing Library
BRL-CAD Release 7.24.1 The BRL-CAD Numerical Computation Library
BRL-CAD Release 7.24.1 The BRL-CAD Utility Library
Minimum run time is 3 minutes, 12 seconds
Maximum run time is 30 minutes
Estimated time is 9 minutes, 36 seconds
Running the BRL-CAD Benchmark tests... please wait ...
+++++ moss
Frame 0: 357882 rays in 0.50 sec = 708785.30 rays/sec (RTFM)
Frame 1: 715764 rays in 0.96 sec = 748038.49 rays/sec (RTFM)
Frame 4: 5726112 rays in 7.41 sec = 772350.97 rays/sec (RTFM)
Frame 5: 11452224 rays in 14.88 sec = 769419.60 rays/sec (RTFM)
Frame 6: 22904448 rays in 29.79 sec = 768928.91 rays/sec (RTFM)
pixcmp pixels: 262144 matching, 0 off by 1, 0 off by many
moss.pix: answers are RIGHT
+++++ world
Frame 0: 488787 rays in 1.52 sec = 322149.10 rays/sec (RTFM)
Frame 1: 977574 rays in 3.00 sec = 325365.23 rays/sec (RTFM)
Frame 3: 3910296 rays in 11.95 sec = 327093.25 rays/sec (RTFM)
Frame 4: 7820592 rays in 23.41 sec = 334064.27 rays/sec (RTFM)
Frame 5: 15641184 rays in 46.53 sec = 336135.15 rays/sec (RTFM)
pixcmp pixels: 262144 matching, 0 off by 1, 0 off by many
world.pix: answers are RIGHT
+++++ star
Frame 0: 312053 rays in 1.04 sec = 300529.93 rays/sec (RTFM)
Frame 1: 624106 rays in 1.98 sec = 315412.22 rays/sec (RTFM)
Frame 4: 4992848 rays in 15.39 sec = 324365.51 rays/sec (RTFM)
Frame 5: 9985696 rays in 30.59 sec = 326443.56 rays/sec (RTFM)
pixcmp pixels: 262144 matching, 0 off by 1, 0 off by many
star.pix: answers are RIGHT
+++++ bldg391
Frame 0: 375758 rays in 1.48 sec = 253244.68 rays/sec (RTFM)
Frame 1: 751516 rays in 2.96 sec = 254229.71 rays/sec (RTFM)
Frame 3: 3006064 rays in 11.42 sec = 263314.15 rays/sec (RTFM)
Frame 4: 6012128 rays in 22.94 sec = 262034.73 rays/sec (RTFM)
Frame 5: 12024256 rays in 46.15 sec = 260561.46 rays/sec (RTFM)
pixcmp pixels: 262144 matching, 0 off by 1, 0 off by many
bldg391.pix: answers are RIGHT
+++++ m35
Frame 0: 656362 rays in 2.28 sec = 288491.18 rays/sec (RTFM)
Frame 1: 1312724 rays in 4.50 sec = 291501.67 rays/sec (RTFM)
Frame 3: 5250896 rays in 17.31 sec = 303259.37 rays/sec (RTFM)
Frame 4: 10501792 rays in 34.33 sec = 305873.45 rays/sec (RTFM)
pixcmp pixels: 262070 matching, 74 off by 1, 0 off by many
m35.pix: answers are RIGHT
+++++ sphflake
Frame 0: 1889266 rays in 4.89 sec = 386095.88 rays/sec (RTFM)
Frame 1: 3778532 rays in 9.65 sec = 391698.43 rays/sec (RTFM)
Frame 2: 7557064 rays in 19.16 sec = 394437.58 rays/sec (RTFM)
Frame 3: 15114128 rays in 38.28 sec = 394787.82 rays/sec (RTFM)
pixcmp pixels: 262144 matching, 0 off by 1, 0 off by many
sphflake.pix: answers are RIGHT
... Done.
Total testing time elapsed: 8 minutes, 48 seconds
The following files have been generated and/or modified:
*.log ..... final log files for each individual raytrace test
*.pix ..... final pix image files for each individual raytrace test
*.log.* ... log files for previous frames and raytrace tests
*.pix.* ... pix image files for previous frames and raytrace tests
summary ... performance results summary, 2 lines per run
Run '/usr/code/brlcad-build/bin/benchmark clean' to remove generated pix
files.
Summary:
Abs localhost.localdomain 768928.91 336135.15 326443.56
260561.46 305873.45 394787.82 398788.39 Thu Jul 25 10:23:08 WAT
2013
*vgr localhost.localdomain 5612.21 5012.45 5822.07 4883.08
4326.96 26.63 4280.56
#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#
Benchmark results indicate an approximate VGR performance metric of 4281
Logarithmic VGR metric is 3.63 (natural logarithm is 8.36)
#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#*#
These numbers seem to indicate that this machine is approximately 4281 times
faster than the reference machine being used for comparison, a VAX 11/780
running 4.3 BSD named VGR. These results are in fact approximately 3.63
orders of magnitude faster than the reference.
Here are some other approximated VGR results for perspective:
120 on a 200MHz R5000 running IRIX 6.5
250 on a 500 MHz Pentium III running RedHat 7.1
550 on a dual 450 MHz UltraSPARC II running SunOS 5.8
1000 on a dual 500 MHz G4 PowerPC running Mac OS X 10.2
1500 on a dual 1.66 GHz Athlon MP 2000+ running RedHat 7.3
52000 on a 4x4 CPU 2.93 GHz Xeon running RHEL Server 5.4
65000 on a 512 CPU 400 MHz R12000 Running IRIX 6.5
You are encouraged to submit your benchmark results and system
configuration information to [email protected]
~~~~~~~~~~~~~~~~~~~~
Including additional kernel and hardware information in the log.
Collecting system state information (via /sbin/sysctl)
Collecting system CPU information (via /proc/cpuinfo)
Read the benchmark.tr document for more details on the BRL-CAD Benchmark.
The document should be available in the 'doc' directory of any source
or complete binary distribution of BRL-CAD.
Output was saved to run-20835-benchmark.log from
/usr/code/brlcad-build/bench
Benchmark testing complete.
---
Cheers!
Nyah
On Thu, Jul 25, 2013 at 1:38 PM, Check Nyah <[email protected]> wrote:
> >ease do share your test and summarize your testing results here or on the
> patch itself when you have it. If you need help setting up the test
> harness, just ask. We >seem to end up writing a quick test harness for
> something ever week. :)
>
> I have already implemented and tested the algorithm with patch already on
> sf [https://sourceforge.net/p/brlcad/patches/216/ here]. Well i need some
> assistance in setting up the test and returning the results for both. But
> patched in the code and performed a benchmark test which returned all
> results correct.
>
> Cheers!
> Nyah
>
>
> On Thu, Jul 25, 2013 at 1:37 PM, Christopher Sean Morrison <[email protected]
> > wrote:
>
>>
>> On Jul 25, 2013, at 4:24 AM, check.nyah wrote:
>>
>> > Sorry, Sean for the mistake I was talking about the
>> bn_mat_determinant() routine(src/libbn/mat.c:1051). I will do the tests on
>> my proposed routines and do a comparism to see it its better; but i know
>> it'll be.
>>
>> Ah, that makes a lot more sense then. I don't doubt that you've made
>> them better, and indeed your patch 216 summary with operation counts sounds
>> very promising that it'll be faster. The main question becomes merely
>> verification that it's correct, which can also be attained by comparison
>> testing.
>>
>> Please do share your test and summarize your testing results here or on
>> the patch itself when you have it. If you need help setting up the test
>> harness, just ask. We seem to end up writing a quick test harness for
>> something ever week. :)
>>
>> Cheers!
>> Sean
>>
>>
>>
>> ------------------------------------------------------------------------------
>> See everything from the browser to the database with AppDynamics
>> Get end-to-end visibility with application monitoring from AppDynamics
>> Isolate bottlenecks and diagnose root cause in seconds.
>> Start your free trial of AppDynamics Pro today!
>>
>> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
>> _______________________________________________
>> BRL-CAD Developer mailing list
>> [email protected]
>> https://lists.sourceforge.net/lists/listinfo/brlcad-devel
>>
>
>
------------------------------------------------------------------------------
See everything from the browser to the database with AppDynamics
Get end-to-end visibility with application monitoring from AppDynamics
Isolate bottlenecks and diagnose root cause in seconds.
Start your free trial of AppDynamics Pro today!
http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk
_______________________________________________
BRL-CAD Developer mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/brlcad-devel
