Thanks very much Frank for such a quick response and for taking the time to investigate the issue I raised.
Your reply is very interesting, as you appear to have used a very different set-up to me, and obtained very different (opposite!) results...
The biggest difference in the setup is that I am on Windows XP, not Mac. It would be intersting to investigate further whether the different results can be related to this. Does most Ableton testing get done on Macs as a first port of call, and could this explain the paradox?
I must also state up-front that my tests were done prior to the recent update, i.e. using Live 4.0.4. I realise that various bugs have been fixed and was pleased to read in the forum here that audio click/pops aren't happening as before. This is good news. Might the improvements in v.4.1 also account for the improved CPU results you acheived?
Regarding Tracktion, Jules (the developer) and Mackie always stress the cross-platform nature of the product but I suspect that the Mac version may perform less well than on Windows, based on the figures you give. I am surprised by your Tracktion results given the power of your G5. On my WinXP laptop I can take the CPU meter up to around 85% without any audio drop-out, which seems significantly better than the results on Mac that you report. Curious...! Also as previously stated I can load around 50% more VSTs for similar results. I need to quantify this more precisely as you have, and will try the good-old Ambience Test!
Of course Tracktion also has a freeze (which I wasn't using) with which I can load up loads more stuff. Perhaps Live will have freeze in a later update?
Regarding latency, I started out with a latency of around 288 samples, which is mid-range on my Edirol UA-20 UBS 1.1 interface. This setting had been acceptable when previously using Tracktion and Reason, although it's worth noting that my typical workflow is to start off by scoring in Sibelius 3 and then importing MIDI rather than recording in via a keyboard (which I tend mostly to use only for audio recording) so latency is rarely a big problem for me.
Because of the issues I was having I increased the audio-buffer to a latency of 1440 samples to see if this would free up CPU. The results were actually not as significant as I thought they would be.
I can reduce latency to 144 samples minimum according to Live's options menu.
What I will now try is to use this minimum latency, with version 4.1, and see if this produces a noticeable improvement in overall performance. I will report back in due course and let you know how I get on!
Thanks again
