Upgrading Memory in your
iMac Intel (Mid 2007) or Mac mini (Mid 2007)

"To Pair or Not to Pair" - part in a continuing series (Updated August 16th, 2007)

In December of 2006, we did testing to establish why Apple had recommended Pairs for MacBook, Mac mini, and certain iMac models. The models Apple recommends paired memory for all have the Intel GMA950 integrated video chipset in common. Of note is that rather than having its own dedicated Video Memory, the GMA950 shares its 64MB VRAM allocation for the main system memory.

All of Apple's current Core 2 Duo Models (same is true between Core Duo equipped Mac models) share the same processor bus and memory controller. It is only the models with GMA950 which Apple recommends the pairs for and from that we'd infer the recommendation is related to impact on video performance.

The bottom line? The results of our testing did not support Apple's recommendation and rather make it very clear that in addition to the potential economic advantages, there are real performance advantages to having more memory - even if it means not having a pair. It is our recommendation to upgrade memory, be it one at a time or with a pair, or to 4.0GB - based on what you need and what your budget allows.

*See original December 2006 article*

The tests that follow show the performance of different paired and non-paired memory configurations.

NOTE: A 4.0GB (2GB x 2 Pair) configuration was also tested by 'popular demand' in the Mac mini. Although the Core 2 Duo Mac mini is limited to 3.0GB addressed, the 4GB Set (2GB x 2 Pair) does appear enable 128 Bit addressing. While there does seem to be a small performance benefit gained with the 4.0GB vs. 3.0GB config, we'd recommend sticking with 3.0GBs max (in Core 2 Duos) unless you absolutely are seeking every last trickle of performance boost possible.

Apple iMac Intel (Mid 2007) and Mac mini (mid 2007) Memory Testing of 1.0GB, 2.0GB, 3.0GB and "4.0GB*" configurations by Other World Computing.

Photoshop CS3 Results

Adobe Photoshop CS3 is the newest, Intel-native iteration of the popular image editing program.

This test measures the time (in seconds) it takes to execute a custom 21-step action script using Adobe Photoshop CS3. Lower times are better.

"Stress Test" Results

This test also times how long it takes Adobe Photoshop CS3 to run a series of 21 actions. However, in addition to Photoshop, the iTunes Visualizer is also set to run. This generally taxes both the video card and the processor. Lower times are better.

"RAM Hog" Results

This test times how long it takes to run our standard 21-step Photoshop action in Photoshop CS3 when 50% othe installed RAM is taken up with another program. Lower times are better.

Halo Results

In order to heavily tax the video card, we ran the time demo of the Intel-native version of the popular game, Halo.

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

XBench Results

XBench is a standardized utility for measuring the performance of your computer, as compared to a "baseline" computer. (More information and the application, itself, can be found here.)

For this test, we checked all the options except for the Disk Test, as it was the only one that did did not either touch System memory or Video Memory. Higher Point results are better.

In conclusion

The numbers speak for themselves. Increasing the RAM in your computer is by far the best performance boost you can do to your system. Even increasing the RAM in your computer to just 2GB makes an incredible difference over the factory stock amount of RAM, especially when running more than one application.