A bit further testing with truepianos seems to indicate that the only module that has a noticeable difference (at least to me) is the Atlantis module which is a preview for the next version. The other modules I tested I could not tell a difference.
 
I should also note that where I hear the most difference is in the attack of the note especially on the higher velocities. The upsample seems to have a more rounded attack and the regular seems harsh to me. Of course depending on what you want out of the track may determine weather you prefer the upsample or not.
 
YMMV
 
 

RSMCGUITAR
Thanks Craig.
I'm pretty new and I think that example was great for a newb like me to get the idea 'ears on.'

Cool. If you listen on phones, you can hear the foldover distortion in the background of the first half. One reason I chose that sound is it didn't have a lot of lows to mask the distortion.
 
In fact, a lot of time the distortion will be masked by the audio, so you can't hear it per se...but if you mixed with the distortion present, the sound might seem weird if it's gone.

Tom Riggs
A bit further testing with truepianos seems to indicate that the only module that has a noticeable difference (at least to me) is the Atlantis module which is a preview for the next version. The other modules I tested I could not tell a difference.
 
I should also note that where I hear the most difference is in the attack of the note especially on the higher velocities. The upsample seems to have a more rounded attack and the regular seems harsh to me. Of course depending on what you want out of the track may determine weather you prefer the upsample or not.
 
YMMV
 
 

It's interesting you mention the next version as I contacted them about 6 months ago and apparently new versions are still on the drawing board just taking longer than expected.

Anderton
 
Been there, done that. Here's an example of a file I played at my address to the New Music Seminar last week in New York. This download link is good for 7 days.
 
The file is at 44.1 kHz. The second half was upsampled to 96 kHz then downsampled back to 44.1 kHz. If you can't hear a difference between the first and second sections, especially with headphones, you should probably consider a career path that involves something other than audio. 

I can certainly hear a difference. The second half has a much fuller richer sound especially in the highs. Is that because of the removal of foldover distortion? Assuming that something has been removed, the upsampled then downsampled again version should have less power at audible frequencies. Conversely if foldover or another form of distortion is introduced by the operation, there should be new frequencies or more power at existing frequencies introduced by the procedure. 
 

 
The above is a sonogram of your sample file. Notice the striking addition of new power in the top of the audible spectrum. Unless I am misunderstanding your description this second half is of the processed (upsampled/downsampled) signal. If so the original signal is the one which should contain only the original frequencies. Where did the new frequencies come from? If they are not present in the original and are present in the processed signal are they not artifacts? 

slartabartfast

Anderton
 
Been there, done that. Here's an example of a file I played at my address to the New Music Seminar last week in New York. This download link is good for 7 days.
 
The file is at 44.1 kHz. The second half was upsampled to 96 kHz then downsampled back to 44.1 kHz. If you can't hear a difference between the first and second sections, especially with headphones, you should probably consider a career path that involves something other than audio. 

I can certainly hear a difference. The second half has a much fuller richer sound especially in the highs. Is that because of the removal of foldover distortion? Assuming that something has been removed, the upsampled then downsampled again version should have less power at audible frequencies. Conversely if foldover or another form of distortion is introduced by the operation, there should be new frequencies or more power at existing frequencies introduced by the procedure. 
 

 
The above is a sonogram of your sample file. Notice the striking addition of new power in the top of the audible spectrum. Unless I am misunderstanding your description this second half is of the processed (upsampled/downsampled) signal. If so the original signal is the one which should contain only the original frequencies. Where did the new frequencies come from? If they are not present in the original and are present in the processed signal are they not artifacts? 

Yes, seeing it here too w/ SPAN. The oversampling is adding distortion at 19.4k and above.

Note that any distortion that might occur is not caused by SONAR's oversampling. The changes caused by SONAR's upsampling and downsampling will be very minor. You can verify that by using a sine wave as input and upsampling a plugin that does no processing or is bypassed. You can also test that plugin directly at the higher sample rate (without resampling) and bounce it to compare. The results should be very close to when resampled.
 
Plugin behavior at different sample rates can vary that's why the only way is to test it for yourself. Its possible that a plugin can cause distortion at higher sample rates. For example an instrument might have its samples stored at 44.1K and when run at 88.2 it will may resample to that rate. Depending on the quality of the SRC in the plugin you could get artifacts. Most plugins won't use the same high quality SRC that SONAR uses since its CPU expensive... There could be any number of other reasons for distortion as well depending on the DSP used in the plugin. There is no one size fits all answer. This is one of the main reasons for all the controversy about differences at high sample rates. It has more to do with variations in DSP processing at different sample rates than the higher sample rate.

For those who are wondering you can see graphs on Sonar's SRC at http://src.infinitewave.ca/ they don't have a sample from Platinum yet but they do from X3 which is likely pretty similar. It does have a tiny bit of aliasing, but at like -100dB so totally inaudible, and it may be on purpose to get a slightly cleaner impulse response.

Doktor
Well I don't know of any global settings that can get changed in Sonar, that can alter sound, other than via preferences and plugin manager. Correct me if I'm wrong? Plenty of studios share DAW computers as you know...

 

Noel Borthwick
If you have a multi user setup and care about someone changing your settings, you shouldn't be sharing the same user profile. That said all the resampling settings are stored in aud.ini (which is per user) so you can easily backup and restore the config settings if you need to. And even latency settings can "alter sound". There are some plugins that sound different at different buffer sizes. 

 

Doktor
 
And then run Sonar as Administrator ;)

I've never seen a recording studio set up additional user accounts. I agree they should but if they are using old plugins such as pentagon which does not store settings in the correct area of the registry, you know life isn't that simple.

Point is up until now I knew that only that altering in preferences and plugin manager will change global settings. Changing global setting in preferences or a setting section has been standard for some time in almost every app I can think of (no doubt somebody will find something obscure to prove me wrong).

Otherwise if a global setting gets changed elsewhere it generally comes with a warning... 'All settings for this plugin will be changed are you sure?'

Notecrusher

slartabartfast
 
I can certainly hear a difference. The second half has a much fuller richer sound especially in the highs. Is that because of the removal of foldover distortion? Assuming that something has been removed, the upsampled then downsampled again version should have less power at audible frequencies. Conversely if foldover or another form of distortion is introduced by the operation, there should be new frequencies or more power at existing frequencies introduced by the procedure. 
 

 
The above is a sonogram of your sample file. Notice the striking addition of new power in the top of the audible spectrum. Unless I am misunderstanding your description this second half is of the processed (upsampled/downsampled) signal. If so the original signal is the one which should contain only the original frequencies. Where did the new frequencies come from? If they are not present in the original and are present in the processed signal are they not artifacts? 

Yes, seeing it here too w/ SPAN. The oversampling is adding distortion at 19.4k and above.

The examples are NOT upsampled audio files, but rendered virtual instruments. In other words, the first example was a virtual instrument sitting in a 44.1 kHz project. It was never recorded at 44.1 kHz, it was rendered at 44.1 kHz. The second example has the same instrument, same track, same MIDI data feeding it, etc., and was also never recorded at 44.1 kHz. However, it was rendered at 96 kHz via upsampling, then downsampled to 44.1 kHz.
 
What you're seeing in the graph is what's so cool about the process; what is in the audio range is reproduced accurately when downsampled. 

I like Noel's last comments (labeled HELPFUL). Seems to sum it up nicely.