Mac OS X 10.6 Brings Serious Performance Gains

Written by Michael Larabel in Operating Systems on 28 August 2009 at 11:56 PM EDT. Page 3 of 13. 130 Comments.

The hardware contained in the older Mac Mini "MM1" was an Intel Core 2 Duo T5600 processor clocked at 1.83GHz, an Intel 945 + ICH7-M motherboard with Intel 945G integrated graphics, 1GB of DDR2 memory, and an 80GB Hitachi HTS542580K9SA00 SATA HDD. Found in the newer Mac Mini "MM2" is an Intel Core 2 Duo P7350 processor clocked at 2.00GHz, an MCP79 motherboard with NVIDIA GeForce 9400M 128MB (shared memory) graphics, 1GB of DDR3 memory, and a 120GB FUJITSU MHZ2120BH G1 Serial ATA 2.0 5400RPM hard drive.

Prior to carrying out or Mac OS X 10.5.8 and Mac OS X 10.6.0 benchmarking we performed clean installations of the operating system each time to avoid any cruft on the hard drive and all system settings were left at their defaults during testing. Some of the software items worth reporting for Mac OS X 10.5.8 are a 9.8.0 i386 kernel, X Server 1.4.2-apple33, OpenGL 1.2 APPLE-1.5.48 on MM1, 2.0 NVIDIA-1.5.48 on MM2, GCC 4.2.1 (provided by Xcode 3.1), Python 2.5.1, and a Journaled HFS+ file-system. Worth noting for Mac OS X 10.6.0 is the 10.0.0 kernel, Server 1.4.2-apple45, GCC 4.2.1 (provided by Xcode 3.2), Python 2.6.1, OpenGL 2.1 NVIDIA-1.6.0 on MM2, OpenGL 1.4 APPLE-1.6.0 on MM1, and a Journaled HFS+ file-system.

With the Phoronix Test Suite the tests we used for this Snow Leopard benchmarking were Nexuiz, Urban Terror, OpenArena, Java 2D Microbenchmarks, LAME MP3 encoding, dcraw, GraphicsMagick, timed apache compilation, timed PHP compilation, 7-Zip compression, Gzip compression, Bork file encryption, OpenSSL signing, BYTE Unix Benchmarks, SciMark 2.0, Threaded I/O Tester, Stream, John The Ripper, timed MAFFT alignment, PostgreSQL, Sudokut, PyBench, Crafty, TSCP, Tachyon, C-Ray, and Sunflow Rendering System. Worth noting for other PTS users is that FFmpeg, GnuPG, MPlayer, GraphicsMagick, and X-Plane 9.00 currently is not compatible with Mac OS X 10.6.0. With some of these test profiles there were multiple options that we ran the tests with, which in total leads to 60 different tests that we had run for this Leopard vs. Snow Leopard benchmarking. With the Phoronix Test Suite most test profiles are automatically ran multiple times to ensure the numbers are statistically significant and accurate, which puts the number of actual runs even many times higher. Of course, then multiply that number by two for our two Mac Mini systems to get the total number.

Without further ado, on the following pages are our benchmarks comparing Mac OS X 10.5.8 "Leopard" and Mac OS X 10.6.0 "Snow Leopard" for your viewing pleasure from the older Mac Mini with Intel graphics (MM1) and the newer Mac Mini with NVIDIA graphics and a faster processor (MM2).


Related Articles