Digi Core Audio Driver and Logic 8

[shadowbox]

Noticed a few things about the Digi Core Audio Driver in Logic 8 and thought others might be interested. I have the latest version of the Digi Core Audio driver that came with PT TDM 7.3.

First off, sometimes without warning, the buffer setting in Logic won't actually set Digi driver with the correct setting. This can be verified by opening up the Digi Core Audio Manager and seeing what it lists as the buffer. Mine was stuck on 512 and once I quit Logic and quit the Digi Core Audio manager, then re-opened the Digi Core Audio manager I was able to select any setting between 64 and 1024. 32 samples is not an option.

When I reopened Logic it then correctly switched the actual buffer setting when I changed the setting in Logic's audio preferences.

When I tried to switch the buffer settings to 32 it actually switched the Digi Core Audio Driver to 64 as it can not go any lower than that. This got me curious so I did a latency test.

I recorded a click and lined it up exactly on the beat. (audio not Midi)
I then recorded the output of the click to another track. Here is what I found:

The Core Audio driver is supposed to report to Logic what its approx. latency is and Logic will automatically compensate for the delay. But the Digi Driver does not do this, at least not correctly. So even though you can not turn this feature off, I was still able to see the relative latencies between the different buffer sizes.

The first surprising thing was that the latency for the 64 sample setting was higher than it was for 128. I checked it 3 times and got the same result. In my test the 64 sample buffer resulted in the click being delayed by 290 samples. Here are the other results:
128 samples= 228 sample delay
256 samples= 356 sample delay
512 samples = 612 sample delay

Do you see a pattern?
and the 1024 buffer resulted in a 1124 sample delay.

So it would seem that the best buffer setting possible is 128. 64 is just a waste of CPU. I would appreciate if someone could verify this for me. But it does explain why some people have said on this board in latency threads that they can still hear a delay or phasing when they are using the 32 sample buffer with the Digi driver. Doesn't it?