I pulled up an AWM2 performance "Fat Sine" and took a look. I can hear some aliasing, but using a parameter like "Legato Slope" dialed up to max reduces the effect such that I doubt it will be apparent in "the mix" for a live situation. If I was playing a solo piece without any other instrumentation - then maybe it would be noticeable. Even a studio mix I think would bury the aliasing.

I was able to "glide" up and down pressing with a palm instead of finger to minimize the sound of keys - I could have used earphones as a means to isolate as well (instead). What I produced was something similar to the ipad youtube video (expected "proper" glide of a sine).

As the 10/10/2010 post on the Motif XF - legato slope is available on the Montage as in the Motif XF. I doubt there have been any changes to portamento given the highly evolutionary nature of most of the sound-producing building blocks.

That said, I believe tests can be still be produced (using Montage) which demonstrate less than desirable outcomes. However, I also believe most of the "issues" can be mitigated and satisfy the needs of most playing situations.

Well how about the octave issue? That sounds unacceptable. Here is how to simulate it:

-create a voice with two elements (with two string samples)
-coarse on of them to 1 or 2 octaves higher
-set the portamento to on, time1,2 or rate1,2 so it sounds as desired

now try to perform a trill. If the time between two notes are short enough MOTIF XF cannot create a correct portamento. It goes crazy and plays a note from one or two octaves higher/lower depending on the setting.

I can mitigate this by using PEG but the sample will sound synthesized

I also disagree with this not being audible in the mix. May be for a weak and bad sounding strings or synths but if you build a powerful voluminous sound it is definitely no what I want. Can someone explain a bit more? and confirm if both issues still exist?
1- First issue, trill and wrong portamento as in my original post
2- non-smooth transition

Really appreciate it if anyone can help me fixing it.

My concern is the issue #1 at the moment. Because as I said, it's wrong. #2 is not wrong it's bad and unpleasant and I can live with it.

"Not audible in the mix" depends on the sample of course. I tested "Fat Sine" and not anything else. Depending on the loop points and phase relationship on a keybank-to-keybank comparison - your mileage may vary. As I mentioned - the performance of Montage is not going to significantly outclass Motif XF since I believe the portamento implementation is the same. I do not know which IC is responsible for the portamento. Vs the XF I believe the tone generator is upgraded. But I believe the portamento "core" logic is lifted verbatim. My main feedback was in response to GospelMusician's demo of an iPad doing what he had issues getting Yamaha keyboards to do - and using "Fat Sine" - one can get results similar to the iPad demo on Montage.

I didn't do further testing because it was a little difficult to discern what issue #1 and issue #2 were since I didn't see labels for each issue in the post.

Testing the "string theory" - I believe the Montage performs in this category as the Motif did. I hear a synth, not string samples during the portamento glide time of strings setup as you described and oscillating between two keys at extremes.

I get better results if I decrease the time around 40 and "run up" or "run down" the notes vs using two discrete keys that are far apart to let Montage play the entire portamento range. This is easiest going from a white to white key - a bit harder going black to black - and a bit harder still to "pull off" going from a white key to a black key.

It would probably be best to spell out:

Mode, Time (value), Time Mode, Legato Slope (value) you want tested. The values are going to transfer direct from Motif XF to Montage.