Agree on the whole that summing is summing is summing. Most DAWs these
days will use floating point (either 32bit or double precision) and
A+B=C no matter what.
There may be dithering at the end but you are talking a single bit
possible difference between renders at points, which I don't believe
anybody is going to hear. I personally find claims that one dither
algorythm sounds better than another, especially at 24bits, can be
nothing but salesman talk.
Now how the DAW loads and converts your samples (eg mixing 44.1kHz and
48kHz samples within the same project, or using a different project same
rate to all your samples) or if you are using an internal sampler to
play at different frequencies, does make a difference. For example
compare Cubic vs Sinc (SinX/X) Interpolation and you very well might
start to hear differences (there are more methods but these are a couple
of the most common.) You may also find that one software uses Cubic on
play and Sinc on rendering (as cubic is less processor intensive) and
thus there are subtle difference between them whereas another may use
Sinc at all times. While a lot of people disregard this as it's not part
of the summing itself I argue it is part of the audio engine.
But every time I have ever come across the X sounds great, Y sounds
pants claims by people and then requests for examples to illustrate this
they have either never materialised or the protagonist has ended up
backing down from their claim. Especially on the rendered waveforms, due
to the fact I mentioned above of some DAWs only using the better
quality interpolation at that point.
As mentioned the characteristics of the internal effects and DSPs used
are going to have a lot more affect on your sound than any of this ever
will!
On the whole any differences that do exist are so small I really wouldn't worry about them.
Dale.