My Hacintosh Experiment

March 1st, 2011 by Museman

Previous: Performance Benchmarking...

Stress Testing:

Although I'm sure all the components are well-built (and guaranteed!), I'm also curious to check out how the MicroMini behaves under stress. Two separate tests were run to quantify its performance limits: One subjective, the other objective.

Test #1: Chess.app

In this test, the Chess program included with Mac OS X is launched and a computer-vs-computer game is run repeatedly to see how many games the computer can complete without stalling or freezing. Before each new game, the level of "Strength" in the Preferences, corresponding to the number of moves ahead the computer considers before making its' next move, is increased: The higher the strength, the more CPU-intensive the game becomes. As an added bonus, you can watch the computer play against itself.

MicroMini Hackintosh: Chess.app test results
Computer Plays: Fastest Default 75% Strongest
Completion Ratio: 4/4 4/4 4/4 1/1
%-Completed: 100% 100% 100% 100%

In general, the "Fastest" games finished in about ten minutes, while the "Strongest" game required about five minutes per move, or over five hours to complete the game. I think I'll save any further testing of the "Strongest" setting for later!

Test #2: MaxCPU

NOTE: The Mac OSX version of MaxCPU is no longer available. The Windows version is still available here.
This software, written by Chris Hynes, simultaneously and continually executes a collection of OS X programs and Terminal commands intended to run a Mac's CPU, video card, memory, and hard drive near or at their performance limits. The test continues until it is stopped by the user. During the test a number of active windows run simultaneously, providing some rather impressive eye-candy.

For this test, after powering on the MicroMini:
  1. The computer was left on for an hour to establish baseline temperature levels.
  2. Preferences for Screen Saver and Enery Saver were set to "Never" to prevent Sleep.
  3. The applications "Activity Moniitor" and "Temperature Monitor" were launched.
  4. The "Start MaxCPU" Automator Action was launched.
  5. After about 37 minutes, the test was stopped by launching "Stop MaxCPU".
  6. When the CPU temperatures returned to the baseline, temperature data was exported from "Temperature Monitor" & plotted.
MicroMini Hackintosh Temp Plot

To make comparisons a little easier, the results were normalized by calculating temperature ratio for each core (T/Tmax) and time-shifting all cores to start at the same time:
MicroMini: Normalized Temp Plot
Having no other data for comparison, and not knowing how OS X delegates work among multiple cores, I can only make some subjective observations:
  • It took eight minutes for all of the cores to reach maximum temperature. During this time Activity Monitor reported all cores near or at 100% utilization, so the lag shown among the cores is curious.
  • Core temperatures ranged between 78 and 82 degrees Centigrade (or 172 to 180 Farenheit). With the case's four fans run at full speed during use, I thought it might run a little cooler, but I suspect the major influence is the CPU fan.
  • After the test was stopped at the 43-minute mark, Core 1 took less time to cool down than the others (4 minutes vs. 10 or more), despite having the highest temperature. During this time Activity Monitor reported all cores essentially at 0% utilization. Once again, the lag shown among the cores is curious.
  • The hard drive seems to be running very cool. If true, it's expected, as the drive is installed just behind the front fan at the fan's centerline height.
Next: User Experience...