[Dean Roddey]

One issue in these discussions is what 'the DAW' is. 'The DAW' is really just the summing engine. It has nothing to do with plugins, which are completely separate things. If you limit yourself to just the DAW and its summing engine, then the differences are going to be very small.

There will be small differences between a floating point and fixed point summing engine of course, but they won't be very large. They have different round off issues. Different floating point engines will be basically the same for the most part, despite claims above. All the machines any of us are going to use will have floating point processing that complies with the IEEE floating point specification, though there can be slightly differences in terms of the options used by compiler generated floating point code. But I'd imagine that there's lots of hand tuned assembler in the core parts of the DAW as well.

Testing of the DAW itself and it ssumming engine is very definitely legitimately done by way of the null test by running any audio track through two of them.

As mentioned, something like pan laws (which are always part of the DAW proper) and dither algorithms when you mix down (sometimes part of the DAW, sometimes done by plugs) can create subtle differences. So you should use equiv pan laws and third party SRC/dither if you are really testing the DAW itself and not their implementations of those algorithms.

As soon as you bring plugs into the mix, unless they are the same plugs on each DAW, then you will obviously start getting differences just as you would if you replaced one hardware EQ or Comp with another.

If you use any sort of time based effects, thos can cause lots of problems with null tests because the cycle of the time based effect may start at a different point in the cycle. So you need to be careful when you do comparative mixes even if you use the same effect plugin, and be sure to do the mix down in such a way that the plugin always starts with the effect at the same point, i.e. do the mix all the way from the start to end, or always from exactly the same point. Otherwise the sweep of a chorus or flanger or modulator can cause seeming differences where there really isn't one.