UVI Falcon 3

VST, AU, AAX, CLAP, etc. Plugin Virtual Instruments Discussion
RELATED
PRODUCTS
Alive - UVI Falcon Expansion Massive X

Post

IvyBirds wrote: Sun May 12, 2024 12:38 amYou say adding more instances would make it run better than a single instance with layers
Nope. You’re over-simplifying, and reading things you want from responses which they aren’t actually saying (not just mine).

The missing context is there’s also a host involved and presumably it’s running other processes. Say you have 2 plugins.. Each using 70% of a 2 core CPU, leaving 30% free on each core. Now, what will your host be able to fit more easily into the remaining resources? One plugin using 60% CPU, or two plugins using 30% CPU each?

Because HALion allows control over the amount of cores, in theory, you can set it to 2 cores and it might still fit in the remaining 30% for each core. But not only might this approach conflict with a host (especially a non-Steinberg one ;) ), it also assumes perfect scaling Vs single core - which is never the case. That’s why, when you see benchmarks, their multi-core results aren’t just a simple multiple of the single core score times the number of cores. Additional cores carry additional overhead. One of many things you appear not to be grasping. :shrug:
You also keep on bringing up Diva which is silly
Apparently it wasn’t as silly when you brought it up first, whilst complaining Falcon won’t do multi-core. ;) What you likely didn’t expect is that it serves as an excellent example of a feature you’d want to avoid using unless it was the only way to stop a single core overloading, which is basically the only situation where you could reasonably argue a NEED for multi-core support anyway.

This doesn’t really apply a whole lot to Falcon, since you’d normally only ever get to that point by layering, for which you can also use extra instances. :party: Anyway, we're clearly going in circles at this point, so I'll step off and wish you luck.. :wheee: :tu:

Post

PAK wrote: Sun May 12, 2024 6:25 am ...
What you likely didn’t expect is that it serves as an excellent example of a feature you’d want to avoid using unless it was the only way to stop a single core overloading, which is basically the only situation where you could reasonably argue a NEED for multi-core support anyway.
...
While we are both obviously wrong (and silly) according to ivybird, it seems our reasonings are carbon-copy.
Anyway I think this quote can be the final conclusion of the discussion.

Post

PAK wrote: Sun May 12, 2024 6:25 am
This doesn’t really apply a whole lot to Falcon, since you’d normally only ever get to that point by layering, for which you can also use extra instances. :party: Anyway, we're clearly going in circles at this point, so I'll step off and wish you luck.. :wheee: :tu:
And that right there is my entire point

For layered patches you get better performance with Falcon if instead of a layer you put each element on it's on track with multiple instances

The reason for this is that the DAW will assign a new thread to each instance.

So why not instead of having to run six instances of Falcon to get six threads for six elements, you could just run one with six elements and then Falcon could just assign a thread to each element?

That's exactly how HALion7 does it

Post

Jac459 wrote: Sun May 12, 2024 7:04 am
While we are both obviously wrong (and silly) according to ivybird, it seems our reasonings are carbon-copy.
Anyway I think this quote can be the final conclusion of the discussion.
Since your reasonings are a carbon copy maybe you can answer why he recommended instead of creating a layered patch in Falcon, I just open a new instance to run each element?

Thanks in advance

And remember according to your viewpoint that should hold no advantage since the DAW itself is just managing the data flow

So according to you one patch with 6 elements or 6 instances with one element is the same right?

Post

IvyBirds wrote: Sun May 12, 2024 7:47 am
Jac459 wrote: Sun May 12, 2024 7:04 am
While we are both obviously wrong (and silly) according to ivybird, it seems our reasonings are carbon-copy.
Anyway I think this quote can be the final conclusion of the discussion.
Since your reasonings are a carbon copy maybe you can answer why he recommended instead of creating a layered patch in Falcon, I just open a new instance to run each element?

Thanks in advance

And remember according to your viewpoint that should hold no advantage since the DAW itself is just managing the data flow

So according to you one patch with 6 elements or 6 instances with one element is the same right?
Oops,
1 - you missed more elements that I thought in the discussion.
2 - I don't even know what you mean ... "remember according to your viewpoint that should hold no advantage since the DAW itself is just managing the data flow". I don't even understand what you mean by the DAW is managing the data flow. Which data flow ? What are you talking about man?

Edit: The fact that you don't take the time to read and understand what others are saying is a bit upsetting to be honest. I (and I am not the only one) have spent time to try to explain things to you and you come with a total misunderstanding of the things said... If you don't get it and you think I am not clear, just ask, don't make stupid ideas yourself...

Post

IvyBirds wrote: Sun May 12, 2024 7:39 am ...
So why not instead of having to run six instances of Falcon to get six threads for six elements, you could just run one with six elements and then Falcon could just assign a thread to each element?

That's exactly how HALion7 does it
1. Multithreading is actually done by the OS...
2. The host is splitting it´s workload as much as it can into seperate threads and reports these threads to the OS...
3. When now additional components are interfering with this order like plugins doing multithreading on their own of which the host doesn´t know anything it can lead to trouble respectively in a worse overall performance... "Too many cooks spoil the broth" ever heard of it??

So even if Multithreading helps you at first glance to put more and more stuff into a single instance it can lead to exact the opposite when everything comes together in a full blown project...

I remember well that Urs said when he implemented Multithreading into Diva that actually one should avoid using it as it can/will cause trouble...

This might be a good reason why some developers willingly avoid to make their plugins multicore...
Perhaps this will change in future with CLAP but for now it´s perhaps more of a bad habbit to do too much in serial /single instance and starting to use your system more efficiently by spreading the workload in parallel/ multiple instances as much as you can...

4. With this "workaround" which is at least for VST/VST3 actually the better method you have the benefit you can do everything you want right now and don´t have to claim and wait for things which probably will never happen...

As long as I can remember it was always the case that the user has more to adopt to the rules of the computer than the other way round...

5. Last but not least getting used to spread your own work out to seperate tracks or in some DAWs at least to parallel on the same track it can have many benefits on top/ more freedom how to trigger or process the individual layer which can be tough to do with a single instance...

This is what i.e. PAK and Jac459 are trying to tell you...
No offense but simply denying and throwing yourself on the floor and drumming with your fists doesn't really get you anywhere... it doesn´t do with children and you will be no exception...

Sorry, but this is simply the truth...

Post

Jac459 wrote: Sat May 11, 2024 1:28 pm
Lerian wrote: Sat May 11, 2024 11:10 am I hope someone from UVI reads this topic, as both the UI and the multicore processing are manageable to fix. Would be a shame for a product this good to have these artificial limitations.
Again, you will have to explain me what is the actual need for multicore in an instrument which is to run among 50 others VST at the same time, carefully distributed by the OS or the DAW among your 6,8 or 12 cores...
Do you really use Falcon standalone and maxout your CPU ? Maybe just buy a computer of this century then (just teasing here).

For the UI, it is I think a far more serious issue as it is the change of a paradigm of ergonomic (and not just a coat of pait) while keeping the compatibility with their hundreds of soundwares... A colossal work IMHO...
I'm not an expert on multi threading and how cpus balance load, but there are for sure other plugins doing that successfully so I don't see why not.

On the UI/UX part I can speak with some authority, and there is no paradigm to change. The only thing to change is how the same paradigm is presented to the user - in a more intuitive manner.

Let me give you some minor examples of friction and confusion inducing design choices to show you how bad it was designed from an UX persepective:

Image

- As you can see above, those arrows imply a scroll through the tabs, while its preset scrolling. Also the hamburger menu implies that I would see the whole list of tabs. Nope, again, its a presets menu.

- Some graphical representations have an eye to zoom the LFO or whatever. That eye is not positioned inside the graphic itself, but way right and above, as its in the presets menu.

- Sometimes you scroll and the EQ points move, sometimes the effects list move. No consistency.

- The FX/Event/Mod vertical positioning is also very confusing. There is no intuitive view, to see exactly the signal path, you have to guess a lot.

And there are many more inconsistencies, logical faults, misplacements, and so on, that adds friction and confusion, and scare users away. All can be fixed with a bit of work, a bit of testing, and some more work. Again, its not the paradigm that's faulty, is the UI/UX itself.

Post

Lerian wrote: Sun May 12, 2024 9:08 am
Jac459 wrote: Sat May 11, 2024 1:28 pm
Lerian wrote: Sat May 11, 2024 11:10 am I hope someone from UVI reads this topic, as both the UI and the multicore processing are manageable to fix. Would be a shame for a product this good to have these artificial limitations.
Again, you will have to explain me what is the actual need for multicore in an instrument which is to run among 50 others VST at the same time, carefully distributed by the OS or the DAW among your 6,8 or 12 cores...
Do you really use Falcon standalone and maxout your CPU ? Maybe just buy a computer of this century then (just teasing here).

For the UI, it is I think a far more serious issue as it is the change of a paradigm of ergonomic (and not just a coat of pait) while keeping the compatibility with their hundreds of soundwares... A colossal work IMHO...
I'm not an expert on multi threading and how cpus balance load, but there are for sure other plugins doing that successfully so I don't see why not.

On the UI/UX part I can speak with some authority, and there is no paradigm to change. The only thing to change is how the same paradigm is presented to the user - in a more intuitive manner.

Let me give you some minor examples of friction and confusion inducing design choices to show you how bad it was designed from an UX persepective:

Image

- As you can see above, those arrows imply a scroll through the tabs, while its preset scrolling. Also the hamburger menu implies that I would see the whole list of tabs. Nope, again, its a presets menu.

- Some graphical representations have an eye to zoom the LFO or whatever. That eye is not positioned inside the graphic itself, but way right and above, as its in the presets menu.

- Sometimes you scroll and the EQ points move, sometimes the effects list move. No consistency.

- The FX/Event/Mod vertical positioning is also very confusing. There is no intuitive view, to see exactly the signal path, you have to guess a lot.

And there are many more inconsistencies, logical faults, misplacements, and so on, that adds friction and confusion, and scare users away. All can be fixed with a bit of work, a bit of testing, and some more work. Again, its not the paradigm that's faulty, is the UI/UX itself.
Thanks for clarifying the UX part... Maybe it can be simpler that I am thinking and your examples are indeed very logical, but will that be enough ? I am not so sure.

As for the balance of resources (multithreading and multi-core are 2 different concepts), the point is that you are already balancing your resources very efficiently with the OS/Host as you are using at once a ton of plugins in a session (I said 50 but can be much more). The interest of doing that at plugin level is the question... (I vote against :-)).

Post

Jac459 wrote: Sun May 12, 2024 9:21 am
Thanks for clarifying the UX part... Maybe it can be simpler that I am thinking and your examples are indeed very logical, but will that be enough ? I am not so sure.
There is a bit of work to be done, as there are many logical issues to be dealt with, but it can be done. I have some experience in designing complex interfaces (including 2 reaper themes) and there is nothing that can't be fixed in Falcon without changing the paradigm. It just needs to lay things down in a more logical intuitive manner so you don't have to think when you look at it - everything can make sense at the first look.

The hardest part is for the makers to realize that they invented the thing, and to them everything looks logical because they know everything inside-out about it. Once they realize that, they should hire an UX designer to make it easy to grasp even to the most novice users.

Many developers underestimate the power of UX in the success of a product. UX can make or break a product, and history is full of examples. Especially in the music producing world where you have to move fast to put ideas into sound, before losing inspiration due to increasing frustration.

Post

Lerian wrote: Sun May 12, 2024 9:50 am
Jac459 wrote: Sun May 12, 2024 9:21 am
Thanks for clarifying the UX part... Maybe it can be simpler that I am thinking and your examples are indeed very logical, but will that be enough ? I am not so sure.
There is a bit of work to be done, as there are many logical issues to be dealt with, but it can be done. I have some experience in designing complex interfaces (including 2 reaper themes) and there is nothing that can't be fixed in Falcon without changing the paradigm. It just needs to lay things down in a more logical intuitive manner so you don't have to think when you look at it - everything can make sense at the first look.

The hardest part is for the makers to realize that they invented the thing, and to them everything looks logical because they know everything inside-out about it. Once they realize that, they should hire an UX designer to make it easy to grasp even to the most novice users.

Many developers underestimate the power of UX in the success of a product. UX can make or break a product, and history is full of examples. Especially in the music producing world where you have to move fast to put ideas into sound, before losing inspiration due to increasing frustration.
Absolutely agree, and as an ex-developer I can tell you that I was beyond shit in UX yet I was the one in charge of designing.
For me I can tell when I like to use it and when I don't like. Proper UX is something complex yet it doesn't interest me at all, that is why I 1000% agree with you.
I don't know what are your thoughts on Rob Papen synths. For a UX designer I guess it must be painful to watch... I like the synths but oh my.... What a pain to watch...

Post

Jac459 wrote: Sun May 12, 2024 7:58 am : The fact that you don't take the time to read and understand what others are saying is a bit upsetting to be honest. I (and I am not the only one) have spent time to try to explain things to you and you come with a total misunderstanding of the things said... If you don't get it and you think I am not clear, just ask, don't make stupid ideas yourself...
Hilarious dodge, thanks for the laugh, I'll just take the win and move on thanks for playing

Again you think breaking up a complex layered patch into individual instances for each element gives a performance boost

But having the plugin break up each element doesn't

Your hilarious

Post

IvyBirds wrote: Sun May 12, 2024 12:46 am
Jac459 wrote: Sat May 11, 2024 10:54 pm
IvyBirds wrote: Sat May 11, 2024 6:11 pm
Jac459 wrote: Sat May 11, 2024 1:28 pm
Again, you will have to explain me what is the actual need for multicore in an instrument which is to run among 50 others VST at the same time, carefully distributed by the OS or the DAW among your 6,8 or 12 cores...
Even if you want to dismiss the real world experiences of others, the fact remains that compared to other software Falcon is a real pig when it comes to CPU usage on complex layered patches relative to other software that offers similar results. If you never do that, awesome but Falcon offers that for a reason

The reality many people perceive is that multicore support makes difference and that feature is offered by its competition

As for me I can make a complex layered patch in HALion7 using multiple cores and that will use considerably less CPU than a similar complex layered in Falcon. Why is that? If I go into the settings in HALion7 and limit it to a single core it also makes a difference in weaker performances compared to bumping up cores to 8. Why is that?

It happens in the real world using Cubase and Studio One. Maybe it's the way those DAWs handle multi core versus single core? All I know is it makes a difference

That won't make a difference on a modern CPU when only using a few plugins but will make a difference when using 50

It's awesome in this thread that we have been offered opinions from bank software programmers on why it won't matter if they add it, but it does very obviously make a difference in Cubase and Studio One which are two very popular DAWs. Maybe they can chime in and explain why that is happening?

We can debate all day long on why it shouldn't happen but it does happen and that's is really all that matters
You seems very upset.
But you are talking about Falcon not being efficient which is different that saying it needs multi-core support. I never commented on this point.
Second, I am just trying to understand how it works behind and I always take precautions to say I am not an expert on the particular case of DAW vs VST. But your conclusion that UVI sucks and they should have multi-core supports is either because But my whole point (and some other people) is to say that in a DAW season with multiple other synths it won't give you any benefice (and I do believe that it is what matters for many).
you continue to say that but ignore the fact that it does and I have seen and experienced it doing just that in multiple DAWs sessions with multiple DAWs

To make that suggestion you have to make the assumption that a DAW is going to do a perfect job of allocating resources and that will not introduce unwanted artifacts

In a perfect world a DAW would be 100% accurate and efficient at allocating resources but we don't live in a perfect world do we?

So offering multi core support by treating each element in a layered patch as basically a separate instance of the program gives the DAW guidance on how to allocate those resources in a logical way
We get it by now. Falcon is not made by your beloved Behringer so it does everything wrong.

Post

IvyBirds wrote: Sun May 12, 2024 1:02 pm
Jac459 wrote: Sun May 12, 2024 7:58 am : The fact that you don't take the time to read and understand what others are saying is a bit upsetting to be honest. I (and I am not the only one) have spent time to try to explain things to you and you come with a total misunderstanding of the things said... If you don't get it and you think I am not clear, just ask, don't make stupid ideas yourself...
Hilarious dodge, thanks for the laugh, I'll just take the win and move on thanks for playing

Again you think breaking up a complex layered patch into individual instances for each element gives a performance boost

But having the plugin break up each element doesn't

Your hilarious
I am happy to be hilarious. Unfortunately I don't have such strong feelings about you. I just see mediocrity and boredom.

The fact that you see this discussion as a battle to win says a lot about you though....

I am happy to let you this "win" though as it seems to be important to you.
but please don't try again to reinterpret my words, you don't seems smart enough to do so.

Post

Trancit wrote: Sun May 12, 2024 8:52 am 1. Multithreading is actually done by the OS...
2. The host is splitting it´s workload as much as it can into seperate threads and reports these threads to the OS...
3. When now additional components are interfering with this order like plugins doing multithreading on their own of which the host doesn´t know anything it can lead to trouble respectively in a worse overall performance... "Too many cooks spoil the broth" ever heard of it??

So even if Multithreading helps you at first glance to put more and more stuff into a single instance it can lead to exact the opposite when everything comes together in a full blown project...

I remember well that Urs said when he implemented Multithreading into Diva that actually one should avoid using it as it can/will cause trouble...

This might be a good reason why some developers willingly avoid to make their plugins multicore...
Perhaps this will change in future with CLAP but for now it´s perhaps more of a bad habbit to do too much in serial /single instance and starting to use your system more efficiently by spreading the workload in parallel/ multiple instances as much as you can...

4. With this "workaround" which is at least for VST/VST3 actually the better method you have the benefit you can do everything you want right now and don´t have to claim and wait for things which probably will never happen...

As long as I can remember it was always the case that the user has more to adopt to the rules of the computer than the other way round...

5. Last but not least getting used to spread your own work out to seperate tracks or in some DAWs at least to parallel on the same track it can have many benefits on top/ more freedom how to trigger or process the individual layer which can be tough to do with a single instance...

This is what i.e. PAK and Jac459 are trying to tell you...
No offense but simply denying and throwing yourself on the floor and drumming with your fists doesn't really get you anywhere... it doesn´t do with children and you will be no exception...

Sorry, but this is simply the truth...
Again, the DAW reports threads to the OS it assigns a new thread to each track

With layered patches if the plugin assigns each element to a new thread the host DAW can see that just like if each element was running in a different instance

That is not the way Diva is implemented because it is not creating layered patches

If you get a performance boost by giving each element it's own instance and thus a new thread you will also achieve a performance boost by giving each element it's own thread at the plug-in level

In either case the DAW is aware of it

I am also fully aware of the pros and cons of using individual instances versus layers from a performance, composition, and MIDI triggering standpoint

And again I have seen first hand from a CPU usage standpoint the advantages of multi threading as I describe.

No offense but simply denying the reality of others in things you have no real world experience in, as i.e. PAK and Jac459 are doing... and throwing yourself on the floor and drumming with your fists doesn't really get you anywhere... it doesn´t do with children and you will be no exception...

Sorry, but this is simply the truth...

They keep on explaining how they think things should work, while ignoring how things actually work

They keep on banging the drum that a DAW perfectly distributes resources while also telling you the work around of opening new instances instead of making layers because a DAW doesn't distribute resources perfectly

Post

Jac459 wrote: Sun May 12, 2024 1:12 pm
I am happy to be hilarious. Unfortunately I don't have such strong feelings about you. I just see mediocrity and boredom.

The fact that you see this discussion as a battle to win says a lot about you though....

I am happy to let you this "win" though as it seems to be important to you.
but please don't try again to reinterpret my words, you don't seems smart enough to do so.
[/quote]
Again with the dodge and the insults amazing how engaged you are in something you pretend is boring

Again still waiting for the explanation of why you acknowledge you get a performance boost by splitting each layer in a layered patch into its own instance, but having the plugin do the same internally has no effect

Again not talking about Diva talking about layered patches in Falcon

Post Reply

Return to “Instruments”