The XBench test results for eight Mac OS X computer configurations are shown below. Six of the configurations are Mac mini computers with different amounts of RAM. Two are high-end PowerBook G4 computers, one running under typical, unoptimized conditions, the other optimized for maximum benchmark performance. All tests except “PowerBook 1.5Ghz 1GB RAM” were run at 1024×768 resolution. The 1.5Ghz 1GB RAM test results were taken from the peak performing 1.5GHz PowerBook listed on the comparisons page at XBench.com.
Higher scores are better. A score of 100 on any test shows equivalent performance to a dual-processor G4 Macintosh that was touted by Apple as 83 percent faster on Adobe Photoshop than a 1.7GHz Pentium 4 Windows PC. In every case except disk speed, the lowest-priced Mac mini configuration was over 11% faster than that reference machine, and frequently upwards of 30% faster. More expensive Mac mini configurations show 25% to 75% performance increases. As such, even the cheapest Mac mini will likely represent a performance improvement for low-end PC users who have not upgraded over the last several years.
The results also broadly suggest that an increase of Mac mini RAM from 256MB to 512MB yields a noticeable performance increase for both CPU-intensive tasks and regular use of a Mac mini’s user interface, regardless of clock speed. Mac minis use slower hard disks than the PowerBooks and therefore had lower disk speeds.
In many cases, the Mac minis outperformed the unoptimized PowerBook 1.5Ghz machine in the benchmarks despite lower CPU clock speeds; however, an ideally optimized PowerBook can outperform all competitors by some margin. For reference, a well-oiled but not perfect PowerBook will score closer to the 150 mark overall, and will not see the dramatic performance jumps indicated in the User Interface and Disk Test sections below.
Overall Score reflects the cumulative results of seven major tests of system performance (and their associated sub-tests), as broken down below.
||182.43 (best measured)
CPU Test consists of five tests that collectively measure one processor running one application with different sorts of math processes.
Thread Test uses two tests to simulate multiple applications or a single application running multiple processes at once. Shows dramatic benefits on a machine with multiple processors, which none of the test machines here have.
Memory Test includes seven tests to show computer’s ability to perform memory operations, measuring memory bandwidth.
Quartz Graphics Test
Five tests of the Quartz Graphics system challenge the machine’s graphics card, memory bandwidth, CPU, and floating point capabilities at once.
OpenGL Graphics Test
A single test uses the graphics card and CPU together to determine 3D performance.
User Interface Test
The single UI Test shows a system’s performance in drawing "standard system controls." Again, the graphics card and CPU are tested.
Four disk tests together show typical throughput to the hard disk and seek time of the drive. A less fragmented, empty drive does better than a full and/or fragmented one.