GCC 3.4.0 20040114 + GCC 3.3.2 compile-time performance comparisonon MICO project sources.

[Karel Gardas <kgardas at objectsecurity dot com>]

Hello,

I have just build MICO 2.3.11's ORB core directory (http://www.mico.org)
with both compilers from subject with -time -O0 -Wall -DPIC -fPIC -c
options and the comparison table looks:

[1] File
[2] GCC 3.3.2
[3] GCC 3.4.0 20040114
[4] Delta %

[1]		[2]	[3]	[4]
os-unix.cc	4.91	5.33	-7.88
dii.cc		17.45	18.58	-6.08
typecode.cc	10.2	10.51	-2.95
any.cc		7.51	7.65	-1.83
codec.cc	6.57	6.68	-1.65
buffer.cc	4.13	3.94	4.82
context.cc	4.22	4.15	1.69
except.cc	5.29	5.67	-6.7
dispatch.cc	5.23	5.62	-6.94
string.cc	4.09	3.95	3.54
object.cc	5.62	5.81	-3.27
address.cc	5.73	6.05	-5.29
ior.cc		16.98	18.27	-7.06
orb.cc		22.46	23.08	-2.69
boa.cc		10.32	10.98	-6.01
dsi.cc		14.83	16.01	-7.37
transport.cc	5.01	5.25	-4.57
t..port/tcp.cc	4.73	5.17	-8.51
t..port/udp.cc	4.73	5.24	-9.73
t..port/unix.cc	4.79	5.19	-7.71
iop.cc		20.54	22.42	-8.39
util.cc		6.5	7.02	-7.41
basic_seq.cc	4.34	4.32	0.46
fast_array.cc	4.7	4.67	0.64
ssl.cc		13.66	15.15	-9.83
fixed.cc	4.43	4.34	2.07
intercept.cc	14.66	15.75	-6.92
codeset.cc	6.44	6.91	-6.8
queue.cc	5.14	5.51	-6.72
static.cc	25.59	26.58	-3.72
current.cc	13.19	14.84	-11.12
policy_impl.cc	17.3	18.8	-7.98
service_info.cc	13.18	14.63	-9.91
ioptypes.cc	15.01	16.51	-9.09
ssliop.cc	13.45	14.74	-8.75
value.cc	15.72	16.66	-5.64
valuetype.cc	14.31	15.88	-9.89
v..type_impl.cc	17.02	18.64	-8.69
dynany_impl.cc	11.02	11.74	-6.13
policy2.cc	13.37	14.77	-9.48
tckind.cc	13.17	14.87	-11.43
orb_excepts.cc	13.26	14.89	-10.95
policy.cc	13.29	14.52	-8.47
poa.cc		17.16	19.07	-10.02
poa_base.cc	14.43	15.95	-9.53
poa_impl.cc	22.55	23.81	-5.29
dynany.cc	14.52	15.97	-9.08
uni_base64.cc	0.1	0.12	-16.67
uni_unicode.cc	0.17	0.21	-19.05
uni_fromuni.cc	0.35	0.41	-14.63
uni_touni.cc	0.38	0.47	-19.15
except2.cc	7.06	7.47	-5.49
pi.cc		15.61	17.53	-10.95
pi_impl.cc	23.57	24.79	-4.92
typecode_seq.cc	13.58	14.99	-9.41
timebase.cc	13.24	14.44	-8.31
ir.cc		59.49	53.89	10.39
ir_base.cc	16.19	17.64	-8.22
imr.cc		19.07	23.93	-20.31
mtdebug.cc	4.6	4.61	-0.22

Sum		696.16	742.59	-6.25


So as you can see, we have about six percents slowdown with the GCC 3.4.0
in comparison with GCC 3.3.2.

In case anybody is interested, I would be happy to provide him/her with
some interesting preprocessed files for more detailed testing.

Thanks for your work on GCC!

Karel

PS: GCC 3.4.0 was configured with:
thinkpad:~$ c++ -v
Reading specs from
/home/karel/usr/local/gcc-main/lib/gcc/i686-pc-linux-gnu/3.4.0/specs
Configured with: ../gcc-main/configure
--prefix=/home/karel/usr/local/gcc-main --enable-shared --enable-threads
--enable-languages=c++ --disable-checking --enable-__cxa_atexit
Thread model: posix
gcc version 3.4.0 20040114 (experimental)


GCC 3.3.2 was configured with:
thinkpad:~$ c++ -v
Reading specs from
/home/karel/usr/local/gcc3.3.2/lib/gcc-lib/i686-pc-linux-gnu/3.3.2/specs
Configured with: ../gcc-3_3-branch/configure
--prefix=/home/karel/usr/local/gcc3.3.2 --enable-shared --enable-threads
--enable-languages=c++ --disable-checking --enable-__cxa_atexit
Thread model: posix
gcc version 3.3.2
thinkpad:/mnt/karel/sl3-for-atlas/demo/security/atlas-sl3$

PPS: Are you also interested in optimized (-O2) build?
--
Karel Gardas                  kgardas@objectsecurity.com
ObjectSecurity Ltd.           http://www.objectsecurity.com