Intel C v8.1 benchmark results for ruby interpreter

[Andrew Snow]

Hello,  I use the Ruby language for work and pleasure and recently 
decided to test Intel C 8.1 as per the lang/icc port for the ruby 
interpreter, which has been shown to be quite sluggish sometimes.

I have run a test on my workstation which showed a significant 
improvement which I wanted to share with you.

The benchmark in question is a small one which generates a mandelbrot, 
you can find it here: 
http://www.timestretch.com/FractalBenchmark.html#346e82c069dbc3bc26fa841434a43ed5





1. Machine specs

OS: FreeBSD 6.2-STABLE (Apr 29, 2007) i386 (32 bit)
Kernel: SMP + cpu I686_CPU only + SCHED_ULE
CPU: Intel(R) Core(TM)2 CPU 4300  @ 1.80GHz (2383.99-MHz 686-class CPU)
RAM: 2GB DDR1 @ 400
Motherboard: Asrock
Chipset: i865 @ 1066 FSB (yes, I know)

/etc/make.conf:
CFLAGS=-Os
COPTFLAGS=-Os
CPUTYPE=prescott
(One CPU core was 100% busy running some other stuff the whole time.)


2. Ruby compiled with in-built GCC

gcc version 3.4.6
ruby 1.8.6 (2007-03-13 patchlevel 0) [i386-freebsd6]

Benchmark times (secs): 7.725779, 7.627775, 6.990523, 7.599726, 7.629552
Avg = 7.514671


3. Ruby compiled with ports/lang/icc 8.1

Compile options: -g -O2
ruby 1.8.6 (2007-03-13 patchlevel 0) [i386-freebsd6.2]

Benchmark times: 5.870923, 6.201353, 6.271936, 6.293150, 6.369043
Avg = 6.201281


Improvement = 17.5%!