Author: Dann Corbit
Date: 13:46:44 01/30/03
Go up one level in this thread
On January 30, 2003 at 16:38:11, Dann Corbit wrote: >On January 30, 2003 at 15:51:12, Andreas Guettinger wrote: > >>On January 30, 2003 at 08:30:51, Russell Reagan wrote: >> >>>On January 30, 2003 at 02:00:57, Dann Corbit wrote: >>> >>>>On January 30, 2003 at 01:34:19, Russell Reagan wrote: >>>> >>>>>On January 30, 2003 at 00:05:35, Dann Corbit wrote: >>>>> >>>>>>Try the following with your compiler inside of your profiler: >>>>> >>>>>With MSVC 6 Pro... >>>>> >>>>> Func Func+Child Hit >>>>> Time % Time % Count Function >>>>>--------------------------------------------------------- >>>>> 3819.452 34.0 3819.452 34.0 1000000 _swi (test.obj) >>>>> 3718.852 33.1 3718.852 33.1 1000000 _poi (test.obj) >>>>> 3710.973 33.0 3710.973 33.0 1000000 _arr (test.obj) >>>>> 0.000 0.0 11213.196 99.7 1 _main (test.obj) >>>> >>>>Interesting. MS VC++ usually has huge differences for the different methods. >>>> >>>>What optimization settings did you use? >>> >>>I set the active configuration to Release, and enabled profiling. When I compile >>>for Pentium Pro, I get: >>> >>> Func Func+Child Hit >>> Time % Time % Count Function >>>--------------------------------------------------------- >>> 3806.877 33.9 3806.877 33.9 1000000 _swi (test.obj) >>> 3714.811 33.1 3714.811 33.1 1000000 _poi (test.obj) >>> 3702.155 33.0 3702.155 33.0 1000000 _arr (test.obj) >>> 0.000 0.0 11217.521 99.9 1 _main (test.obj) >> >> >>I tried this on gcc with gcc -O2 -pg -Wall hallo.c, and the results of gprof: >> >> >> called/total parents >>index %time self descendents called+self name index >> called/total children >> >> 0.00 0.00 1/1 __start (788) >>[1] 100.0 0.01 1.24 1 _main [1] >> 0.57 0.00 1000000/1000000 _swi [2] >> 0.42 0.00 1000000/1000000 _arr [3] >> 0.25 0.00 1000000/1000000 _poi [4] >> >>----------------------------------------------- >> >> 0.57 0.00 1000000/1000000 _main [1] >>[2] 45.6 0.57 0.00 1000000 _swi [2] >> >>----------------------------------------------- >> >> 0.42 0.00 1000000/1000000 _main [1] >>[3] 33.6 0.42 0.00 1000000 _arr [3] >> >>----------------------------------------------- >> >> 0.25 0.00 1000000/1000000 _main [1] >>[4] 20.0 0.25 0.00 1000000 _poi [4] >> >>----------------------------------------------- >> >>regards >>Andreas > >Intel Profiler with Intel C++ compiler: >Full Name Timer Samples Timer events Timer % >tanh.A 783 783000 15.5388 >sinh.A 527 527000 10.45842 >cosh.A 521 521000 10.33935 >sin.A 446 446000 8.850962 >exp.A 441 441000 8.751736 >cos.A 423 423000 8.394523 >tan.A 367 367000 7.283191 >log10.A 362 362000 7.183965 >log.A 317 317000 6.290931 >swi 260 260000 5.159754 <<<<<<<< * LAME * <<<<<<<<<<<< >arr 95 95000 1.885295 <<<<<<<< * GOOD * <<<<<<<<<<<< >sqrt.A 92 92000 1.825759 >poi 75 75000 1.488391 <<<<<<<< * BEST * <<<<<<<<<<<< >log 40 40000 0.793808 >sin 36 36000 0.714427 >log10 35 35000 0.694582 >tanh 35 35000 0.694582 >sqrt 33 33000 0.654892 >cosh 32 32000 0.635047 >tan 32 32000 0.635047 >exp 31 31000 0.615201 >cos 30 30000 0.595356 >sinh 22 22000 0.436595 Intel Profiler with Microsoft C++ .NET compiler: Full Name Timer Samples Timer events Timer % _ffexpm1 1141 1141000 16.17981 _fFCOSH 1064 1064000 15.08792 _trandisp1 552 552000 7.827567 _CItan 440 440000 6.239365 _CIsin 400 400000 5.67215 cos 394 394000 5.587067 _CIlog10_default 357 357000 5.062394 _fFSINH 355 355000 5.034033 _CIlog_default 348 348000 4.93477 _fload 295 295000 4.18321 _ctrandisp2 245 245000 3.474192 swi 231 231000 3.275666 <<<<<<<< * LAME * <<<<<<<<<<<< _fFEXP 214 214000 3.0346 _ctrandisp1 192 192000 2.722632 _fload_withFB 178 178000 2.524107 sqrt 140 140000 1.985252 arr 117 117000 1.659104 <<<<<<<< * GOOD * <<<<<<<<<<<< _math_exit 94 94000 1.332955 poi 91 91000 1.290414 <<<<<<<< * BEST * <<<<<<<<<<<< log 35 35000 0.496313 sinh 33 33000 0.467952 exp 25 25000 0.354509 cosh 23 23000 0.326149 _rttospop 22 22000 0.311968 log10 22 22000 0.311968 tanh 22 22000 0.311968 _rtnospop 11 11000 0.155984 _exp_default 6 6000 0.085082 main 5 5000 0.070902
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.