For me the problem is the -s or silent option in the time-compile-mplayer command. Just running the same commands that script does without timing gives me wild results sometimes works, sometimes dont. It is never silent. Running the script time-compile-mplayer without the -s option is silent anyway. I would guess that may be a make bug? Anyway silent or -s seems uneeded and since Michael assumed the -jX was the problem he fell back to -s which is problem no matter with or without the -j option. But just running with make -j5 gives me more real results. Seems to end with encoder or mencoder right?
Anyway there's my 2 cents.

Quote Originally Posted by deanjo View Post
OK, something really odd here.

Whenever the mplayer compile test is ran from a suite such as

./phoronix-test-suite run universe
or
./phoronix-test-suite run universe-cli
or
./phoronix-test-suite run multicore

I get the mplayer build not completing and giving false results like

Code:

======================================
Timed MPlayer Compilation (Run 1 of 1)
======================================

MPlayer Build Time: 10.51 Seconds

====================================
Timed MPlayer Compilation:
Time To Compile

10.51 Seconds

Average: 10.51 Seconds
====================================
If I run just the mplayer compilation test on it's own or in some other suites I get more realistic results.

./phoronix-test-suite run build-mplayer
or
./phoronix-test-suite run compilation


Code:
======================================
Timed MPlayer Compilation (Run 1 of 1)
======================================

MPlayer Build Time: 85.05 Seconds

====================================
Timed MPlayer Compilation:

85.05 Seconds

Average: 85.05 Seconds
====================================

dean@linux:~/phoronix-test-suite>
This is with using a clean install of the suite .7 version. All other tests complete fine and give realistic results. It's just mplayer in certain suites.