Tests - iTunes CD rip & QT Pro 7 video conversions
Here we tried to ensure that we were running the same versions of iTunes and QuickTime Pro 7 under Mac OS X and Windows but these aren't always exactly in synch. In each case, iTunes was 6.0.4 but QT Pro 7 was 7.1.1 under OS X and 7.1 under Windows.
iTunes CD ripping
A stopwatch was required for our ripping tests. Tracks were ripped one at a time in iTunes' default AAC format and we stopped timing when the ripped version of the track began to play - something that happens automagically in iTunes when the rip finishes.
We ripped from two CDs - David Gray's latest, Life in Slow Motion, and the first disc in a six-CD Django Reinhardt compilation, Djangology. The same tracks were ripped under each OS and in the same order.
Our selection from each CD included tracks from the start, the middle and the end. That way we had a representative selection - tracks are ripped quicker the further out they're located on the disc because the effective speed of the laser reading head over the disc gets faster the further it's positioned from the centre of the disc.
We made no attempt to judge the quality of the ripped tracks, we merely timed how long the ripping took.
As you can see below, there are no results for XP running under Parallels - iTunes simply refused to rip CDs when we ran it under Parallels, though, remember, we were using a beta of that virtual PC app.
The results obtained in the other two configurations show that Apple has a big job to do with CD ripping under OS X on the mini (and perhaps on Intel Macs generally). The speed was abysmal compared to rips carried out under XP running on its own partition. Our suspicion is that the real problem is caused by OS X simply not spinning CDs as fast as XP does or not being able to read them accurately at higher speeds. So, this might just be a CD/DVD drivers issue.
QuickTime Pro 7 - video conversion
For our QT Pro 7 video-conversion tests, we used as the source a 116MB PAL DV AVI file. This was of 720x576 resolution, 32.06 seconds in duration and had 16-bit (48kHz) PCM stereo sound and a 25fps frame rate.
Again a stopwatch was required and we made no attempt to judge the quality of the video being produced, merely timing how long the conversions took.
In one set of tests, the AVI was converted to iPod video format (320x240, 25fps) and produced a file - encoded in H.264 format - that, in every case, was 3.19MB in size.
In the other tests, the AVI was converted to Sorenson 3 video and the file that was produced was 31.16MB in size in each case.
The settings we used for the Sorenson 3 conversion were:
General
* Fast-start for internet streaming
* Little-Endian
* Little-Endian
Video
* Restrict to 6400Kbit/sec
* Optimised for de-interlace
* Same frame rate (25fps) and size (720x576)
* Quality - medium
* Restrict to 6400Kbit/sec
* Optimised for de-interlace
* Same frame rate (25fps) and size (720x576)
* Quality - medium
Audio
* 48kHz/Linear CPM
* Sample size - 16 bits
* Quality - normal
* 48kHz/Linear CPM
* Sample size - 16 bits
* Quality - normal
QT7 Pro - convert AVI to iPod
Performance under OS X was respectable for the conversion to iPod format. However, QT 7 under XP running in its own partition bested it by a decent margin. Rather surprisingly, although the conversion carried out under Parallels was the slowest of the three, the differential between it and Mac OS X was the same as between OS X and Windows running in its own partition.
This suggests to us that Apple has some work to do with the creation of iPod video under OS X on Intel Macs - and that's a bit worrying if, as we believe is the case, QT 7 is running natively on Intel-based Macs.
QT7 Pro - convert AVI to iPod
Here, OS X got well and truly dumped on - perhaps because there has been little or no work done by Apple or Sorenson to get Sorenson's encoder working well on Intel-based Macs. The OS X performance was truly dreadful, and conversion took more than twice as long as under XP running from its own partition and almost twice as long as XP under Parallels.
Our overall conclusion from the iTunes and QT Pro tests are that Apple's OS X programmers needs to work harder to speed things up under OS X on Intel-based minis (and, perhaps, all Intel-based Macs).
We do understand that their colleagues who program for XP versions have had a head start with Intel CPUs but, as a company, Apple should be embarrassed that Tunes and QT Pro - two of its headline apps - work so much slower under OS X than Windows XP.
