delorie.com/archives/browse.cgi   search  
Mail Archives: djgpp-workers/1999/06/28/10:56:40

Message-ID: <19990628165640.A197@tabor.ta.jcu.cz>
Date: Mon, 28 Jun 1999 16:56:40 +0200
From: Jan Hubicka <hubicka AT ta DOT jcu DOT cz>
To: djgpp-workers AT delorie DOT com
Subject: regparm... byte benchmarks
Mime-Version: 1.0
X-Mailer: Mutt 0.93i
Reply-To: djgpp-workers AT delorie DOT com

Hi
I've found the byte benchmarks on my hdd, so here are the results:
(-fomit-frame-pointer -O3 -mregparm=x)
regparm                 0            1             2               3
size                100kb        98kb          97kb           96kb
integer             1.865239     1.867126      1.878186       1.892956
fp                  1.421428     1.503115      1.493591       1.494438

so here is approx 4% difference in executable size and similar difference
in running speed. So overall mregparm=3 seems to do best job
There are slight regresions in some tests between regparm0 and regparm3.
The largest slowdown is stringsort 2.06 -> 2.04 (about 1%, so it can be statistical
error as well, as the slowdown between mregparm=1 to mregparm=3 in fp tests)
Largest speedups are in the fourier (1.62 -> 1.87 and huffman 1.86->2.04)
so mregparm seems to be win.

Full results follows
Honza

mregparm=0

BYTEmark (tm) Native Mode Benchmark ver. 2 (3/95)
NUMERIC SORT:  Iterations/sec.: 73.232475  Index: 1.892557
STRING SORT:  Iterations/sec.: 4.696991  Index: 2.064612
BITFIELD:  Iterations/sec.: 10111774.312300  Index: 1.734489
FP EMULATION:  Iterations/sec.: 3.783849  Index: 1.819158
FOURIER:  Iterations/sec.: 1434.591546  Index: 1.624385
ASSIGNMENT:  Iterations/sec.: 0.587097  Index: 2.236815
IDEA:  Iterations/sec.: 99.565217  Index: 1.523336
HUFFMAN:  Iterations/sec.: 67.283724  Index: 1.869771
NEURAL NET:  Iterations/sec.: 0.779465  Index: 1.318891
LU DECOMPOSITION:  Iterations/sec.: 22.706172  Index: 1.340546
...done...
===========OVERALL============
INTEGER INDEX: 1.865239
FLOATING-POINT INDEX: 1.421428
 (90 MHz Dell Pentium = 1.00)
 ==============================

 mregparm=1

 BYTEmark (tm) Native Mode Benchmark ver. 2 (3/95)
 NUMERIC SORT:  Iterations/sec.: 72.944158  Index: 1.885106
 STRING SORT:  Iterations/sec.: 4.697209  Index: 2.064707
 BITFIELD:  Iterations/sec.: 10112833.383773  Index: 1.734671
 FP EMULATION:  Iterations/sec.: 3.783849  Index: 1.819158
 FOURIER:  Iterations/sec.: 1678.600000  Index: 1.900675
 ASSIGNMENT:  Iterations/sec.: 0.587097  Index: 2.236815
 IDEA:  Iterations/sec.: 99.347826  Index: 1.520010
 HUFFMAN:  Iterations/sec.: 68.168133  Index: 1.894349
 NEURAL NET:  Iterations/sec.: 0.786199  Index: 1.330286
 LU DECOMPOSITION:  Iterations/sec.: 22.750520  Index: 1.343164
 ...done...
 ===========OVERALL============
 INTEGER INDEX: 1.867126
 FLOATING-POINT INDEX: 1.503115
  (90 MHz Dell Pentium = 1.00)
  ==============================

  mregparm=2
 BYTEmark (tm) Native Mode Benchmark ver. 2 (3/95)
 NUMERIC SORT:  Iterations/sec.: 72.944158  Index: 1.885106
 STRING SORT:  Iterations/sec.: 4.676788  Index: 2.055731
 BITFIELD:  Iterations/sec.: 10111774.312300  Index: 1.734489
 FP EMULATION:  Iterations/sec.: 3.768213  Index: 1.811641
 FOURIER:  Iterations/sec.: 1653.213039  Index: 1.871929
 ASSIGNMENT:  Iterations/sec.: 0.585848  Index: 2.232056
 IDEA:  Iterations/sec.: 99.565217  Index: 1.523336
 HUFFMAN:  Iterations/sec.: 71.655213  Index: 1.991252
 NEURAL NET:  Iterations/sec.: 0.786199  Index: 1.330286
 LU DECOMPOSITION:  Iterations/sec.: 22.663529  Index: 1.338029
 ...done...
 ===========OVERALL============
 INTEGER INDEX: 1.878186
 FLOATING-POINT INDEX: 1.493591
  (90 MHz Dell Pentium = 1.00)
  ==============================

  mregparm=3

  BYTEmark (tm) Native Mode Benchmark ver. 2 (3/95)
  NUMERIC SORT:  Iterations/sec.: 75.112301  Index: 1.941137
  STRING SORT:  Iterations/sec.: 4.608295  Index: 2.025624
  BITFIELD:  Iterations/sec.: 10170863.260280  Index: 1.744625
  FP EMULATION:  Iterations/sec.: 3.768213  Index: 1.811641
  FOURIER:  Iterations/sec.: 1653.213039  Index: 1.871929
  ASSIGNMENT:  Iterations/sec.: 0.587097  Index: 2.236815
  IDEA:  Iterations/sec.: 100.000000  Index: 1.529988
  HUFFMAN:  Iterations/sec.: 73.688427  Index: 2.047754
  NEURAL NET:  Iterations/sec.: 0.784497  Index: 1.327407
  LU DECOMPOSITION:  Iterations/sec.: 22.751373  Index: 1.343215
  ...done...
  ===========OVERALL============
  INTEGER INDEX: 1.892956
  FLOATING-POINT INDEX: 1.494438
   (90 MHz Dell Pentium = 1.00)
   ==============================

- Raw text -


  webmaster     delorie software   privacy  
  Copyright © 2019   by DJ Delorie     Updated Jul 2019