Upgrading Memory and Hard Drive in your Mac mini

by OWC Michael

With the optimization of Apple's Operating System "Snow Leopard," comes the optimization of our memory and hard drive upgrade benchmark testing. All of our tesing shown here has been completed using Apple OS X version 10.6.x in both 64 bit and 32 bit mode where applicable.

Why update and standardize the testing? Our prior benchmark tests were run across different versions of the operating system which could potentially cause slight variations within results. Some of the tests were also re-designed, such as our 21 Step Photoshop action test and its successor the Large-scale 47 Step Photoshop Action Test. The 21 Step test was simply completed too quickly to show significant returns in time once machines were able to accept more than 2GB of total memory.

As a final result, you can now directly compare the performance of contrasting Apple machines such as a Core Solo Mac mini with 2GB of RAM installed versus a Core 2 Duo MacBook Pro with 4GB of memory allowing you to make the most informed decision about upgrading your existing system or the purchase of a new machine.

"Adobe Photoshop CS4 Action Test" Results

This test measures the time (in seconds) it takes to execute a custom action script comprising of 47 commonly-used, memory-intensive processes on a 21 megapixel image using Adobe Photoshop CS4. It is designed to represent an average photo editing session of a graphic artist. No other processes are running in the background besides those that keep the machine itself running. This is pure, linear computing where every available resource is on hand for the one test running. Photoshop CS4 is a 32 bit program that has a limit of 3GB of memory that can be used by the program itself so timing should not differentiate greatly with additional memory added to the system.

When CS5 is released from Adobe as a true 64 bit program, we are expecting that any additional memory added to the system is going to speed up a single process like this, but once you have all the memory in the system that the single process can use, and you're running no other processes, the differences in time are very slight. Lower times are better.

You can download the photoshop action script here: OWC Photoshop Bench.atn.zip

Test Conclusion: As Photoshop itself uses up to 3GB of memory, once the 3GB of memory is installed in the machine, the results flatten out. However, check the tests below to see what adding additional processes at the same time does to overall computation time.

Back To Top

"Stress Test" Results

Our Stress Test is a variation on the Adobe Photoshop CS4 Action Test where test is performed while iTunes is playing music with the visualizer running. Really, who doesn't have at least a browser, or a mail client, or music playing when using another program on their computer. By their very nature we expect our computers to be able to keep up with our busy lives. While waiting for some code to compile, or your movie to encode, or a file to download... let's face it anytime that bar moves from left to right telling you the computer is busy with something - we all tend to try doing something else. Surf the web, play a quick game, chat online with friends..while we pass the time waiting we use more resources. We just don't want to take more resources away from our main process or we'll slow it down.

This test adds one additional program to represent that additional light computing we do on the side in real life. With the Visualizer on, the resource draw on the machine is not limited to processor power and memory but to the video card and memory allocated for video as well. Lower times are better.

Test Conclusion: Photoshop itself uses up to 3GB of memory and iTunes uses about 1GB for this test. Once the 4GB of memory is installed in the machine, the results flatten out. It is interesting to note on the machines running 2GB of memory alone that the same Photoshop process takes over an additional minute to complete.

Back To Top

"RAM Hog I Test" Results

This test times how long it takes to run the Adobe Photoshop CS4 Action Test once again when 50% of the installed RAM is allocated to a custom RAM Hog I program. This is designed to replicate steady usage of the machine, such as web surfing, email, streaming media, etc. while waiting for the action to render.

Just because your computer is working on one process, doesn't mean you stop using the computer. It is this multi-tasking ability that make computers such an intregal part of our daily lives and such a powerful tool. This massive ability to multitask is where additional memory added to your computer system does the most good. The more OWC memory you have, the more effective and efficient your overall performance will be. Lower times are better.

Test Conclusion: With additional processes running, the need for additional memory is clear. Once the maximum memory is installed on the machine do we get back to roughly the same computing time of the linear test itself.

Back To Top

Adobe After Effects Render Test

We have also run the Total Benchmark created by Brian Maffitt. Adobe After Effects CS4 spawns subprocesses -- one for each core -- when 'multiprocessing' is enabled in After Effects Preferences. Each of those subprocesses can grab up to 3GB of RAM. To adjust the preferences for Multiprocessing go to After Effects > Preferences > Memory & Multiprocessing. We ran our tests with the option to "Render Multiple Frames Simultaneously" checked, "Maximum allocation per CPU" set to 2.09GB, and the "Longer RAM Preview" to "Faster Rendering" slider was left in the middle.

The first comp mostly focuses on raw processing power as it generates a fractal sequence along with a heavily feathered mask, with Colorama, Extract, Wave Warp, Channel Blur, Curves, and Extract all in the mix. The second comp imports the fractal noise movie, and adds several layers and many effects to generate a fun, complex 3D scene. We are comparing the time results of the second test.

Times are given in minutes : seconds. Lower results are better.

Test Conclusion: While it makes little difference on this test whether running the machine with a 5400RPM drive or an SSD (mainky because the rendering is done with memory and processor without much disk action) additional memory does make a difference. This is again a very linear test with nothing else run in the background. It is safe to assume that with other processes running, the times would be longer through 6GB and that maxing the memory at 8GB would be the best bet for quickest computing.

Back To Top

"HALO Frame Rate Test" Results

Recognized as one of the greatest achievements in video game entertainment, the Xbox version of Halo has already sold over 3 million units and continues to be a top seller. The Macintosh version adds 16-person online multiplayer, new vehicles, new weapons, new maps, customizable multiplayer games, the precision of playing with a mouse and keyboard and more.

For our testing, we ran the Timedemo Benchmark from the Intel-native version of the popular game, Halo v2.0.3
When Halo is launched in Timedemo mode, it will run through four fixed cinematic sequences from the game. Once it's finished, it generates a file called Timedemo.txt in the ~Documents\Halo directory. The contents of the file can be examined to see what settings were used in the timedemo, including what the average framerate for the demo was.

Results are the average frame rate (in frames per second) achieved. Higher results are better.

Test Conclusion: Across the board the Mac mini gives great frame rates - more than adequate for smooth gaming experiences. But when trying to squeeze every bit of performance out of your machine, installing a Solid State Drive can boost the framerate by as much as 12fps.

Back To Top

Want to check your setup against our numbers?
System preferecnces were also all left as close to factory defaults as possible with few exceptions. Under the Energy Saver pane, sleep has been turned off for both display and hard disk.

When a specific program is used outside its factory default, the change is noted at the beginning of each test.

No network connection, airport or bluetooth settings are enabled in order to eliminate outside influences to the tests.