Bury the Megabyte Myth, a 1.25 GHz Mac mini with 512MB RAM--How Fast Is It?
We ordered a 512MB PC2700 DDR 333MHz RAM module on line and received it Friday. Unfortunately, the price of PC2700 DDR RAM jumped up in the past week from the recent mid-$60's so our module cost $84, plus $8.62 overnight delivery by DHL. We actually received a faster 512MB PC3200 DDR 400MHz module so perhaps a shortage of the slower PC2700 RAM may explain the recent price increase. To install the RAM in our Mac mini, we had to "crack" open its case again with our 4" putty knife. Thankfully, the second time we opened our Mac mini was much easier than the first.
Here are the results using our standard suite of speed tests using both "canned benchmarks" and "real world" tests. (see our speed test FAQ):
| Test | Mac mini |
Mac mini
|
|---|---|---|
| Benchmarks: |
w/256MB
|
w/512MB
|
| Altivec Fractal 1.1.3 (GFLOPS) |
4.5
|
4.3
|
|
Cinebench 2003 - Rendering
|
113
|
113
|
|
- OpenGL Hardware Lighting
|
441
|
433
|
| Let1kWindowsBloom 1.0 (sec) |
34
|
34
|
| Xbench 1.1.3 - CPU |
152
|
150
|
| - Threading |
111
|
109
|
| - Memory |
129
|
125
|
| - Quartz Graphics |
161
|
158
|
| - OpenGL Graphics |
111
|
108
|
| - User Interface Graphics |
208
|
206
|
| - Hard Disk |
57
|
58
|
| "Real World" Tests: | ||
| Startup (sec) |
50
|
50
|
| Shutdown (sec) |
19
|
14
|
| Launch Classic Mode (sec) |
13
|
14
|
| Dup. 700 MB Folder 3600 Files (sec) |
94
|
99
|
| Scroll 142p Acrobat Reader File (sec) |
25
|
25
|
| Encode MP3 in iTunes 4.7 (sec) |
53
|
53
|
| Export QT Movie in iMovie 4.0.1 (sec) |
46
|
45
|
| Quake III Arena 1.32 (frames/sec) |
89
|
88
|
*Faster = higher scores and lower seconds. Average of 3 tests.
The results are virtually identical within normal statistical variation for our Mac mini using 256MB or 512MB of RAM. There is no evidence that the performance of the 1.25GHz Mac mini is in anyway impaired at 256MB of RAM.
However, two weeks ago we devised an additional test that would show more clearly the effects, if any, of different amounts of RAM. Our RAM test using Xbench and open applications is adequate to demonstrate the utility of adding RAM for general use but we decided to kick it up a notch to be more convincing. We added concurrent QuickTime Streaming of Steve Job's Macworld Expo keynote address and iTunes playing Beethoven's Sixth to the five open applications. Of couse, we also had the test application Xbench and Activity Monitor open to examine RAM usage.
Here is a table of RAM usage at 256MB and 512MB without the applications open (No Apps--white column), with them open (Apps--yellow column) and with QuickTime streaming and iTunes playing (Apps+--orange column):
|
256MB
|
512MB
|
|||||
|
RAM Allocation
|
No Apps
|
Apps
|
Apps+
|
No Apps
|
Apps
|
Apps+
|
|
Wired
|
37
|
42
|
42
|
49
|
54
|
57
|
|
Active
|
97
|
138
|
140
|
80
|
165
|
240
|
|
Inactive
|
48
|
69
|
70
|
47
|
175
|
153
|
|
Subtotal Used
|
183
|
249
|
252
|
176
|
394
|
462
|
|
Free
|
73
|
7
|
4
|
336
|
118
|
62
|
|
Total
|
256
|
256
|
256
|
512
|
512
|
512
|
At 256MB, there is a small change in the amount fo RAM assigned by Mac OS X to all categories when receiving QuickTime streaming and with iTunes playing (compare the orange vs yellow columns). The free RAM drops from the already small 7MB to 4MB. With 512MB, the RAM use allocation by category is substantially increased when receiving QuickTime streaming and iTunes is playing a song. Does this affect the performance of the 1.25GHz Mac mini? More importantly, does using 512MB ameliorate any of the performance hits?
Here are the speed test results with no applications open, five applications open and with QuickTime streaming and iTunes playing:
|
256MB RAM
|
512MB RAM
|
|||||
|
Xbench Test
|
No Apps
|
Apps
|
Apps+
|
No Apps
|
Apps
|
Apps+
|
|
CPU
|
152
|
142
|
103
|
150
|
143
|
114
|
|
Thread
|
111
|
104
|
72
|
109
|
101
|
76
|
|
Memory
|
129
|
123
|
93
|
125
|
123
|
94
|
|
Quartz Graphics
|
161
|
157
|
103
|
158
|
156
|
104
|
|
OpenGL Graphics
|
111
|
103
|
64
|
108
|
102
|
64
|
|
User Interface
|
208
|
191
|
107
|
206
|
194
|
111
|
If one compares the Xbench scores with no applications open (white columns) with those for the five additionals apps open (yellow columns), there is a perfomance hit of 8 per cent or less, mostly less, whether using 256MB or 512MB of RAM. Opening a QuickTime stream plus playing a song on iTunes (orange columns) results in a further and significant performance hit that totals up to 49 per cent for the Mac mini using 256MB of RAM. However, the salient conclusion from comparing between yellow columns and between orange columns is that the additional 256MB RAM in the 512MB RAM module does not ameliorate the Xbench performance hit, at least not appreciably and certainly not to the point where it justifies an investment in the 512MB RAM module.
Do our tests throw conventional wisdom--more RAM is better--on its ear? Well, not exactly. For some, more RAM is still always better. It would be easy to construct a test that would clearly justify the expense of additional RAM. For example, loading a very large graphics file into Adobe Photoshop and then applying numerous filters to the image might even justify the near $200 expense of a 1GB RAM module for the Mac mini. But graphics professionals are not the target market for the Mac mini and such a test would be misleading to the people in Apple's Mac mini target market--entry-level or highly price-conscious users.
There was a day not too long ago when the convention "more RAM is better" applied to most people using Macs. Classic Mac OS required one to specifically allocate RAM to each application and RAM was very expensive so most people had too little RAM. It also applied in the early days of Mac OS X when new Macs came with a mere 64MB of RAM. But times have changed and convention has not yet caught up. With Mac OS X, 256MB of RAM in a low-end consumer Mac is more than adequate.
The RAM megabyte myth needs to be buried alongside the megahertz myth.
Saw something? Send a tip
The archive ran on reader tips. What did you see, where, and do you want the credit?