[MPlayer-users] gcc 2.95.3 vs. gcc 3.2

[Arpi]

Hi,

I've done some quick comparson between 2.95.3 and brand-new 3.2 version.
Note, that 3.2 supports pentium4 target/optimization, so it should result
better code - in theory.

System: 1.8ghz p4, 512mb rd800 ram, i850, abit th7-ii raid mb, ibm 80g hdd
MPlayer: CVS version, almost every auto-detectable option present and detected

2.95.3 compilation time:  (./configure;time make)

real    2m0.673s
user    1m51.150s
sys     0m8.550s

(repeated to verify results - error under 1%)

3.2 time: (./configure --cc=gcc-3.2;time make)

real    3m1.728s
user    2m49.810s
sys     0m10.930s

so, about 50% slower. ok, we hope the result will be faster so it does
worth it. (gcc 2.95.3 -march & -mcpu was i686, 3.2's was pentium4,
both auto-selected by ./configure)

File:  /3a/MPlayer/benchmark/testsuite2/Star_Wars_Episode_1.avi
VIDEO:  [DIV3]  720x576  24bpp  25.00 fps  1858.5 kbps (226.9 kbyte/s)
gcc 2.93.5 : 22.070 : 21.985 : 22.038
gcc 3.2    : 22.328 : 22.354 : 22.370

-> slower at high bitrate

File:  /3a/MPlayer/benchmark/testsuite2/405divx_sm_v2[1].avi
VIDEO:  [DIV3]  356x240  24bpp  30.00 fps  343.0 kbps (41.9 kbyte/s)
gcc 2.93.5 : 4.054 : 4.052 : 4.072
gcc 3.2    : 4.062 : 4.042 : 4.086

-> ~ same at low bitrate

File:  /3a/MPlayer/benchmark/testsuite2/sample.light.it.up.avi
VIDEO:  [DIV3]  640x352  24bpp  23.98 fps  823.5 kbps (100.5 kbyte/s)
gcc 2.93.5 : : 3.983 : 3.953 : 3.949
gcc 3.2    : : 4.086 : 4.045 : 4.050

-> bit slower at middle bitrate

I'm just wondering why do people prefer teh slow buggy 3.x series over the
good old, working 2.95.3 ?

Oh, i've found it!!!

-rwxr-xr-x    1 root     root      2408836 Aug 30 16:47 mplayer.295
-rwxr-xr-x    1 root     root      2336740 Aug 30 16:47 mplayer.32

3.2 produces 3% smaller (stripped size) binary :)


A'rpi / Astral & ESP-team

--
Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu