So I work for a computer hardware review website and I’d like to use Audacity as a benchmark in future reviews. The issue that I am having is finding a task in Audacity that would be limited by the performance of the CPU, rather than by memory size or I/O bandwidth of the system.
I should mention that I am using a Windows 7 x64 SP1 system and running Audacity version 2.0.2.
The processed sound quality is not really affected by hardware. With slow hardware processing tracks just takes longer than with a fast machine, but the end result is identical.
Hardware, drivers and hardware configuration can have a big effect on playback quality. Trying to play a lot of tracks on a slow machine can cause the sound to break up and stutter. The problem tends to be worst when doing “overdub recording” (recording one track while playing back others). Measuring this could be difficult on a modern computer unless it is seriously under powered or only just meets minimum system requirements. You may need a lot of audio tracks before it starts stuttering. A data bottleneck anywhere in the system may cause stuttering and skips (see here: http://manual.audacityteam.org/man/FAQ:Recording_-_Troubleshooting#skips)
Spectrograms use a fair bit of CPU time when the FFT window size is large. I think this is all done in memory, but it might need to access the disk to access the next bit of audio.
You could display an audio file with, say, 2 minutes showing, as a spectrogram with the window set to 32768, and see how long it takes to refresh the screen when you press Page Up to scroll a page to the right. I don’t know if the contents
If it’s still not slow enough, duplicate the track a few times and display them all as spectrograms.
The spectrogram parameters are in the Preferences, in the Spectrograms section.