I guess some independent and verifiable testing may be in order. I fired up Cubase to do just that - but my Cubase novice-ness is somewhat holding up the train.

What I did was setup two single-waveform PARTs. PART 1 and PART 2. PART 1 was panned hard left and PART 2 panned hard right. I only turn on the left speaker so only PART 1 is heard. My assumption is that polyphony dropouts are FIFO meaning the oldest note gets the noose.

So I start with a C3 for a couple beats on MIDI Channel 1 (PART 1) then add 127 notes on PART2. I actually use all different notes since MIDI happens to handle 128 unique notes (0-127). The theory is that no dropouts should occur if all is setup correctly.

What I found is I had to subtract out 18 notes before my C3 stops getting cut off.

I don't think this is a valid test because I believe there is likely an error in my setup - maybe duplicate notes.

I know I can ask Cubase to remove duplicates - but I find it surprising I'm having such a hard time finding any statistics on Cubase MIDI. I'd like to hilight all and find a statistics window which shows how many notes I have selected. This would help me verify if there is a problem with my "piano roll".

Also, Cubase is kind of a poor match in the MIDI department for Montage because the Polyphony limiter function maxes out at 100. Montage needs to max this out at something more sensible. I didn't see a configuration setting to change the max polyphony - but who knows.

Anyhow - it seems like I have something approximating 128 polyphony if I can ever straighten out gathering Cubase statistics. I could also play the MIDI file to MIDI-OX and use that tool - but kind of wanted to learn something about Cubase since it's bundled.

I'm going to ignore this for now, NOT delete the 18 notes testing AWM2 and add those back as PART3 - an FM-X operator. I start a D#4 note on MIDI channel 3 to invoke the FM-X. In this case, I'm not going to add any more FM-X notes since I am testing the theory about 128 - uhh ... sounds. I'm not going to say "waveforms" because only AWM2 has waveforms (samples) and the terms are slightly different in FM-X.

So with this configuration, I have the first C3 (AWM2) cut off - but the D#4 (FM-X) does not cut off.

Conclusion: FM-X polyphony is not tied to AWM2 polyphony.

Ok - lets flip the script. I'm going to take those 127 later notes (AWM2 since they are channel 1) and make them channel 3 instead. I pan FM-X now to the right so I don't hear it muddling up my C3 (AWM2). The hypothesis is that AWM2 will not cut off since it is only playing one note within its engine.

So I play this - notes that previously made AWM2 drop a note - now they are assigned to FM-X and do not drop the AWM2 note.

Conclusion: Last conclusion above is even more solid.

I think the 128 shared theory is not holding true. I'd have to learn more about cubase to get more surgical, but this is a reasonable test as-is.

I do think what you say about 128 WAVEFORMS is true. AWM2 only uses waveforms - so there are only 128 total at a time. FM-X uses zero waveforms - so makeing FM-X sound notes has its own pool of polyphony which does not invoke WAVEFORMS. I haven't settled it in my own mind if every carrier with a non-zero level is considered a unit of polyphony or if the entire algorithm with how many carriers are sounding is a single polyphony unit. I thought the latter was stated. This is easy to test. Add more carriers and see if the polyphony decreases.

FM-X is 128 Mono not stereo.

Would have been better with the 256, dynamically divided between AWM and FMx.

Then it would have been perfect for DAW use.

Jason wrote:
I think the 128 shared theory is not holding true. I'd have to learn more about cubase to get more surgical, but this is a reasonable test as-is.

Yamaha clearly states that the Montage can play 128 [notes|waveforms|voices|elements] (pick your word for that, I use "voice") with AWM2 at the same time. In addition it can play 128 FM-X voices according to Yamaha. I would trust that claim. Also all experiments I did were pointing in that direction.

According to the description of the innards at http://sandsoftwaresound.net the FM-X is probably on a different DSP, actually.

Note that each FM-X voice comprises 8 operators, so in theory you should be able to play 128 * 8 saw waves (the FM-X saw wave) in parallel from the FM-X engine.

@Stefan,

I do trust the statements from Yamaha that explain you get 128+128 however you slice it. At the same time, if someone is not convinced - then you come up with a test to verify the claim that will provide the evidence.

I was trying to also outline that perhaps the claim of only 128 Waveforms at all times may have misled Phil's own thinking - providing a possibility for making that statement true while also showing it does not mean FM-X shares.

Although driving Cubase is not exactly my thing - I gave a fairly "ok" outline of how to setup your own test for verification. Not a horrible thing to push the hardware and see it drop notes in-line with the spec.

Also, my recollection of what a "unit of polyphony" for FM-X matches your description - that it's whatever those 8 operators of an algorithm are doing is one unit as you've described. Previous to the post, I tried to dig up the post from BM that outlined this but couldn't dig it up.

Certainly rounding out the picture is the bit about activity. Seems like common sense - but I suppose as you're building a sound you could lose the fact that polyphony defines atomic sub-parts in action (not at rest). So, for AWM2 for example, any element not triggered at any given time does not subtract from the polyphony count.

I hope bad mister or Yamaha will join-in here and clearly tell us the spec and the rules on how note pool works and what constituents in use , not in use relative to the AMP ENV's completion times, triggering, etc
I seriously doubt FMX gets added to equal 256. All the ad copy would have a field day with that since the standard 128 mono/64 stereo polyphony feature has grown very old

128 stereo.

https://www.yamahasynth.com/forum/intro-and-a-few-questions

Understanding how polyphony works is an important subject to understand. It is not as simple as 8 Elements use 8 notes, that rarely happens (I explain below). An Element only uses polyphony when it is actually sounding. When an Element is sitting in reserve (which is how most are used), it does not use up polyphony. The advantage of having access to 64 oscillators (8 Parts, 8 Elements) is awesome, and works because of this "dynamic allocation" of resources.

An acoustic piano that is a three way velocity switching program, like the "Full Concert Grand" - well known from the Motif XS/XF and MOX/MOXF series - although it has a soft, a medium and a hard strike sample set and has key off noise, and a separate undamped sample set... All eight Elements are used, but when a Key is triggered on this program only two notes of polyphony are used. This is because "an Element only uses polyphony when it is actually sounding". When the velocity value is low only the soft sample is recalled and plays, the others are "in reserve". It uses two notes of polyphony because to recreate the Full Concert Grand in stereo there are two separate samples, one representing the left channel, the other the right channel.

In the Montage, the Full Concert Grand uses only a single note of polyphony for each Key pressed. The Montage is 128 notes of stereo polyphony on the AWM2 engine side.

The Montage has a separate 128 notes of polyphony for the FM-X engine. I stress: this is separate polyphony. Separate, being the key word here.

OPERATORS refer to FM-X Oscillators* (*The Operator, in the short explanations of FM, can be equated to an OSCILLATOR although its structure and role is quite different... an Operator includes an AEG (amplitude envelope generator) and is an interactive block... As Sam mentions some make sound others are modifiers. FM-X uses polyphony differently than the sampled-based ELEMENTS of the AMW2 engine. Whether you use one Operator or all 8 Operators to create your FM-X PART the polyphony will be either 128 (when you select poly) or 1 (if you select 'mono').

I would imagine "in use" is from the note-on (which results in a response) until the level of release equals zero. Yes, I recognize I'm answering counter to your request.

Jason, Bad Mister's comments are explaining a different subject regarding the practical implementation of polyphony relative to programming and how smart programming makes efficient use of the note pool and other related topics. He then says:
..."The Montage is 128 notes of stereo polyphony on the AWM2 engine side. The Montage has a separate 128 notes of polyphony for the FM-X engine. I stress: this is separate polyphony. Separate, being the key word here."
I see where there is STILL a lack of clarity here, ...Separate being key here could easily mean they retain full independence as in separate thus 128 from awm +128 from fmx. People are still entering into lengthy discussion when there should not be a need for this.
This is what I think: the max polyphony is 128 period doesn't matter if waveform is mono/stereo or an FMX note so, an example would be: 64 FMX notes plus 32 AWM2 mono waveforms plus 32 AWM2 stereo waveforms. I highly doubt I'm incorrect.
Constantly saying you have 128 in engine A + 128 in engine B and that they are completely separate leads one to think 256

People are still entering into lengthy discussion when there should not be a need for this

That you are correct about.
The AWM2 usage of polyphony is its own. The FM-X usage of polyphony is its own.

If you have an AWM2 Part that is triggering a stereo Waveform you can't play that sound with 128 notes of polyphony. If you attempt to layer this with a second AWM2 Waveform you are out of polyphony. You've used the 128 stereo AWM2 notes of polyphony of that engine.
If you, however, layer the original AWM2 program with an FM-X Part, you will notice it does sound, because the AWM2 usage of polyphony is a totally separate thing from the FM-X engine. The FM-X engine has 128 notes still available.

Now that's almost as silly as this discussion. Because that's not how you use it. Running out of polyphony - say you did a setup with all AWM2 sounds and you are mindlessly using up polyphony... and you have this huge hit where you obviously have exceeded the polyphony, substituting an FM-X sound would solve your issue... and vice versa, say you are doing a DAW setup where it's all FM-X sounds and again you hit the maximum polyphony, substituting an AWM2 sound will solve your issue. Two separate pools.

FM-X is not audio samples, "stereo" does not apply. Bert was mistaken if he said that. I've been interviewed at shows and it's easy to reverse it: but it is and remains 128 notes of stereo polyphony AWM2, plus 128 notes polyphony FM-X.

And although 128 + 128 = 256, the only correct way to say it is (128 stereo plus 128)

Stereo FM-X (sic) is considered pseudo-stereo and is accomplished by using two Parts panned wide.
FM-X is pure synthesis, you'd no more call it 'stereo' than you would an analog synth.

So if you had a DAW based MIDI recording that used all 128 notes of AWM2 polyphony, you could play along with full 128 notes from the FM-X engine.

What on earth are you attempting to do?

Bad Mister wrote... "What on earth are you attempting to do? " Attempting to get a straight answer on this matter. As you know, this was on the minds of many besides me.
FINALLY! Thank you Bad Mister. As I had wrote earlier Bad Mister will be able to answer this correctly. You clearly explained anything I need to know. Thank you

Bad Mister wrote:

People are still entering into lengthy discussion when there should not be a need for this

That you are correct about.
The AWM2 usage of polyphony is its own. The FM-X usage of polyphony is its own.

If you have an AWM2 Part that is triggering a stereo Waveform you can't play that sound with 128 notes of polyphony. If you attempt to layer this with a second AWM2 Waveform you are out of polyphony. You've used the 128 stereo AWM2 notes of polyphony of that engine.
If you, however, layer the original AWM2 program with an FM-X Part, you will notice it does sound, because the AWM2 usage of polyphony is a totally separate thing from the FM-X engine. The FM-X engine has 128 notes still available.

Now that's almost as silly as this discussion. Because that's not how you use it. Running out of polyphony - say you did a setup with all AWM2 sounds and you are mindlessly using up polyphony... and you have this huge hit where you obviously have exceeded the polyphony, substituting an FM-X sound would solve your issue... and vice versa, say you are doing a DAW setup where it's all FM-X sounds and again you hit the maximum polyphony, substituting an AWM2 sound will solve your issue. Two separate pools.

FM-X is not audio samples, "stereo" does not apply. Bert was mistaken if he said that. I've been interviewed at shows and it's easy to reverse it: but it is and remains 128 notes of stereo polyphony AWM2, plus 128 notes polyphony FM-X.

And although 128 + 128 = 256, the only correct way to say it is (128 stereo plus 128)

Stereo FM-X (sic) is considered pseudo-stereo and is accomplished by using two Parts panned wide.
FM-X is pure synthesis, you'd no more call it 'stereo' than you would an analog synth.

So if you had a DAW based MIDI recording that used all 128 notes of AWM2 polyphony, you could play along with full 128 notes from the FM-X engine.

What on earth are you attempting to do?

so, why there are 2 version of the samples ? stereo and mono L/R ? if the polyphony is always 128 I prefer to use a stereo waveform always....

so, why there are 2 version of the samples ? stereo and mono L/R ? if the polyphony is always 128 I prefer to use a stereo waveform always....

I don't understand the question? Nothing prevents you from choosing always to use a Stereo Waveform. Compatibility with previous data formats found in earlier Motif series Libraries (where a separate audio piece was kept for the left and right Channels) is the reason you see various left and right Waveforms. There are also often mono versions of instrument Waveforms.

Not all people who use today's modern synths set them up in stereo. We've been discussing creating a bread & butter sound set optimized for those who still live in a mono world. Rather than just scattering the Mono versions of things in the list, put them together. You'd be amazed at how many MOXF's, Motifs and Montage's are played through mono setups... does stereo suffer when played back in mono? No doubt.

Bad Mister wrote:

If you have an AWM2 Part that is triggering a stereo Waveform you can't play that sound with 128 notes of polyphony. If you attempt to layer this with a second AWM2 Waveform you are out of polyphony. You've used the 128 stereo AWM2 notes of polyphony of that engine.

Can you please reword or explain this a bit clearer BM. What do you mean that you can't trigger a stereo Waveform with 128 notes of polyphony? Can we assume this is a typo (based on contrary things you've said) and that you meant "can" instead of "can't"?

Bad Mister wrote:
What on earth are you attempting to do?

Personally, I am making totally new sounds/Performances that are unique and special and amazing and I specifically need/want to use lots of Elements for this and I want to know why I'm not able to procure 128 polyphony as is advertised when trying this (see current thread: HERE)

Bad Mr.
Your last answer. Truly confusing. Typo's?
Please re-post As It is very contradictory in places.