Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 24 Sep 2014 20:13:18 +0100] rev 215
Fix in BenchmarkRunnerAdapterStX: the adapter itself must expand --script/--arguments.
Otherwise it won't be possible to use Smalltalk/X specific options in script file.
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 25 Aug 2014 10:06:56 +0100] rev 214
Merged 67b35d0a0ee3 and 918f0c6191c2
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 12 Mar 2014 18:42:12 +0000] rev 213
Some more work on St/X benchmarks.
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 10 Mar 2014 11:40:06 +0000] rev 212
Initial commit of Smalltalk/X benchmark set.
Not all benchmarks are updated for CalipeL. Benchmarks taken from
package stx:goodies/benchmarks & stx:goodies/benchmarks/misc
Jan Vrany <jan.vrany@fit.cvut.cz> [Thu, 22 May 2014 12:26:59 +0100] rev 211
Added options to disable JIT compilation.
--disable-jit disabled both Smalltalk and Java JIT.
--disable-jit-smalltalk and --disable-jit-java disables selectivelt JIT for the language.
Note, that --disable-jit-smalltalk disables JIT for all languages compiled to Smalltalk
bytecodes such as JavaScript or Ruby.
Jan Vrany <jan.vrany@fit.cvut.cz> [Thu, 22 May 2014 12:07:48 +0100] rev 210
Removed dependency on stx:libprofiler.
Calipel still can be run under callgrind (by means of --callgrind option), but packages
are no longer dependent on it so Calipel will load into a system with no stx:libprofiler loaded.
However, then running --callgrind has no meaning and won't produce any output.
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 21 May 2014 13:11:23 +0100] rev 209
Fix: setup for performance only when running using standalone runner.
Do not override JIT-compiler settings and other values for performance when running
in IDE.
This was confusing as settings were silently changed when Calipel was loaded
into image.
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 21 May 2014 12:33:44 +0100] rev 208
CLosing branch mh_highcharts_spike
Jan Vrany <jan.vrany@fit.cvut.cz> [Wed, 21 May 2014 12:32:17 +0100] rev 207
Merged 29602f0696d8 and 05be338e59fe
Jan Vrany <jan.vrany@fit.cvut.cz> [Mon, 10 Mar 2014 11:29:09 +0000] rev 206
Refactoring and cleanup.
* Support for multiple per-benchmark setups.
* Fix for overridden parameter definitions.
* Fix in parameter combinator. When parameter has no default value but possible values
are listed and no value is explicitly defined (using -D), then run benchmark with all values
listed in parameter definition.
* Run setup/teardown before and after **EACH** benchmark run and before/after warmup.