Thorn: Dmitry Sches' new synth!

[] Peter:H []

BUT Thorn does import a folder.. so try loading it with 16 WavForms.. yeah >> open import and I arrive at my desktop... well I sorted that with a few shortcuts.. not very intuitive and now I got a Win_Desktop FULL of shortcuts

You are kidding me...sigh...many many and I mean many clicks later the slow dude get's the folder trick finally ...my own stupidity is way to frustrating. And yes it's wavosaur by the way, but not only because it allows to split a wav into 16 regions.

[toonertik]

And yes it's wavosaur by the way, but not only because it allows to split a wav into 16 regions.

Yes, I know.. I have been using it for a few years myself.. It is one of the little gems out there. While so many mention Audacity( which is very good, OFC) I like the Wavosaur workflow.. very simple with most of the tools I need.. and so small >> I love that.. I always think that tells how good a programmer is, no reliance on bloated libraries and maybe lots of assembler code )))))

[AstroCastro]

There is a (cpu) tax on quality. My preferred synths are all quite taxing. There is no free lunch.

You sound like a 14 year old who is bragging how he's playing CoD on max settings.
The main problem we are having is that a bunch of developers who wrote magnificent pieces of code showed or keep showing small or non-existing interest in developing more synths and that's why all those synths are seen as exceptions and not the standard.
A bunch of developers and companies couldn't beat one single Sylenth1 throughout years by shoving synth after synth and let's imagine if Lennard kept making regularly synths by now we would have 5 "Sylenths", maybe one of them could have been a modular synth, other one additive, third one a wavetable synth, etc.
If developers like Lennard kept making synths nowadays we would have over 20 synths with exceptionally good ratio of sound quality and CPU usage, his 5-6 synths and 10-15 from other developers who also made exceptionally optimized synths. Synths like that would become a standard, not CPU hogs.
Urs & Co. wouldn't be able to sell shit if things were different.

This is the main reason why your perception is skewed, you are simply misguided by developers with poor optimization skills that only a CPU hog can produce a high quality sound or that high quality sound goes hand in hand with CPU hogging.

Btw, exactly mentioned Dune (1 and 2) really light for my cpu. Though i should admit that unisson in Thorn i liked more. As well, Largo the same lightweight for me (but what regarding his unisson quality, here ambiguous moment - I would not call it no better and no worse, just other, with his own character).

That's because every single properly optimized synth uses around 50-80MHz of CPU power when you take a dry sawtooth and place a chord.
There's no difference between Synth1, Charlatan, Sylenth1, Dune1, Dune2, Harmor, Helix, etc. when you do this basic test.
All properly optimized synths will give you similar results with this test.

Thorn is using 9 TIMES more of CPU power for this basic task when compared to all properly optimized synths. Of course that things just get multiplied from that point. You can't expect from a synth with a poorly optimized oscillator algorithms to behave like properly optimized synths when its core, the base, starting point which is oscillator is screwed up.

Hive is using 6 times more of CPU power when compared to Dune2 when you do this basic test and Dune2 was released in 2014 and the beta of Hive was released in 2014.
It only shows you who has what work ethics, who will sell you a properly coded synth and who keeps relying on technology progress to swallow his spaghetti coding.

Of course that both Urs and Dmitry will market Hive and Thorn as light on CPU synths, but in reality both of those synths are CPU hogs when compared to properly optimized synths, not to mention their standard CPU hogs.

The whole point is that even though CPU can handle 20 instances of those synths, it doesn't change the fact that they are poorly optimized. If their oscillators were properly optimized they would use at least 5 times less of CPU power.

[Taron]

AstroCastro, I got your back, know and feel where you're coming from.
One trouble with the saw test, though, is that it's an entirely different synthesis! It doesn't just produce a saw wave, but it has to create one based on the frequency bands applied and then treat the result, already doing plenty of times more things than just a simple saw wave routine. Many synths even use stored waveforms, really just having to read them properly with minimal filtering. I believe that makes the SAW test not adequate.
But it's more and deeper than just that. There are certain temptations a developer can't easily withdraw from. Don't know how else to formulate that. Once you got one cool thing working, another cool idea could fit so well and it only takes a little more processing...and then another quick little thing really can't hurt and so on and so on and at the end you've got one big pile of great sounding cpu murderer.

My only real beef with Thorn is the bad idle behavior, which sounds ridiculous of me to say, but it's actually what makes or breaks a pure Thorn composition. I'm a musician, not a construction worker, so freezing tracks never works for me as I let a song keep evolving.

I do sense that there are good ways left for Dmitry to optimize what's going on, including the oscillators. With the basic and beautiful concept of Thorn in place, reaching in there and making things perfect piece by piece will not only evolve Thorn, but actually Dmitry himself. I love the concept of it, I really dig the general feeling of the GUI- except lack of mouse over and some sloppy mouse wheel step sizes. I can see how the modulation section wants to get more power, too, including options to switch from "x" to "+", for example, and possibly setting ranges and/or certain interpolation options/operators.
It is a commercial synth! It's not a free synth. So I can't encourage Dmitry to ignore such things, because it would hurt him and his already very excellent visions. It's a whole lot of work and I absolutely know it. I can't even tell you, if it's worth it...because I haven't tried making a commercial product myself, yet. But I believe it's worth it for the instrument you're creating and the music it makes all of us create with it!

[toonertik]

AstroCastro, I got your back, know and feel where you're coming from.
One trouble with the saw test, though, is that it's an entirely different synthesis! It doesn't just produce a saw wave, but it has to create one based on the frequency bands applied and then treat the result, already doing plenty of times more things than just a simple saw wave routine. Many synths even use stored waveforms, really just having to read them properly with minimal filtering. I believe that makes the SAW test not adequate.
But it's more and deeper than just that. There are certain temptations a developer can't easily withdraw from. Don't know how else to formulate that. Once you got one cool thing working, another cool idea could fit so well and it only takes a little more processing...and then another quick little thing really can't hurt and so on and so on and at the end you've got one big pile of great sounding cpu murderer.

My only real beef with Thorn is the bad idle behavior, which sounds ridiculous of me to say, but it's actually what makes or breaks a pure Thorn composition. I'm a musician, not a construction worker, so freezing tracks never works for me as I let a song keep evolving.

hahaa... you beat me to it... I was just composing a suitable reply in my head when I re-freshed to see this...
So many variables need to be taken into account.. also I tend to think of modern synths having "Sound_Generators" rather than Oscillators, which I more think of in analog hardware

[Taron]

Yep, the only way I role myself! I must let the computer generate the sounds. I somehow deeply dislike semi-romplers, so to say, haha.
There's a much different sound coming out that way, which is almost organic automatically. It doesn't suffer the "chipmunk" phenomenon and can respond in a "natural" way to modulations and therefore playing styles. But we're not in a thread for general synthesis discussion, so I'll stop right now. (...cause I could go on for ever about it, possibly.)

[ZaBong69]

It only shows you who has what work ethics, who will sell you a properly coded synth and who keeps relying on technology progress to swallow his spaghetti coding.

You are really accusing people like Urs and Dmitry of not having a proper work ethics bc. they program synths that are more CPU taxing than you expect them to be?

We all can, very easily, imagine a Sylenth 1 semi-modular with the same CPU requirements as the original one. And a reverb that does not sound shit, has modern analog-modelled rich filter models etc. The reason there currently is not may be either because (a) that all those genious synth programmers with proper work ethics of old just retired after they made their first big hit, like you claim, or (b) it may not be a easy as lay people like us believe once you actually write the damn code. What do you think is more believable?

And btw., it is not only the sound engine that has to work - the GUI also has to be the best nowadays, hirez and fully scalable, and the preset management must be gorgeous, and add at least 1000 presets from professional sound programmers on top of that, that thing has to run on all DAWs, PC and Mac, support MPE etc, and be tested, tested, tested. And what do you get at the end? Some anonymous dud on the internet says that you do not have a proper work ethics.

Seriously.

[ZaBong69]

Deleted.

[Taron]

Sometimes you have to allow for some frustrations to be heard, especially when they're meant to move something forward. The only thing I dislike, is when stuff simply gets insulting for reasons of taste or lack of understanding. But this is pure frustration for reasons I can relate to, too. I would never put it that way, of course, because I think Dmitry is a passionate and very able creator, well on his way to become a solid beacon in the synth dev world, while his lights are already shining. The only way to really push forward is by doing, and he does.
A tiny bit of spiky feedback can help to look into the right direction. I, as I said before, also feel like people tend to lose sight of optimization and really nifty coding. Innovation takes bold steps, but we have to watch out to chisel back in those smaller steps so people can actually walk up them stairs into the future.
Our anonymity is something that allows us to be outspoken, for better and for worse. To me, I never feel anonymous, because I feel like I'm with you all, which can get pretty scary at times, haha. ...however, it allows people to say what they think, and every serious thought, clumsily expressed or not, represents the best of our qualities as a group. Just stay focused on the topic and Thorn doesn't have a thread here to simply advertise (I like to believe), but to help and inspire Dmitry to take it to the next level and make it more and more attractive from the inside and outside.
Man, do I wish we were already in a different age, when financial aspects are a distant memory of barbarism in mankind's history.
Anyway...no need to argue about manners too much, though, I'm sure your love and understanding is being felt, ZaBong!

[ZaBong69]

Our anonymity is something that allows us to be outspoken, for better and for worse. To me, I never feel anonymous, because I feel like I'm with you all, which can get pretty scary at times, haha. ...however, it allows people to say what they think, and every serious thought, clumsily expressed or not, represents the best of our qualities as a group. Just stay focused on the topic and Thorn doesn't have a thread here to simply advertise (I like to believe), but to help and inspire Dmitry to take it to the next level and make it more and more attractive from the inside and outside.

I too, want Dmitry to be inspired. Calling him out on his work ethics is not going to help, it is demotivating and insulting. A guy with that skill set would have no problem landing a gig coding some embedded software we all would never see and hear of, earning way more money than self-publishing software synths for what is, let's call it like it is, a very small and unimportant market of nerds.

I really do love what these guys are doing, be it Steve Duda, Urs Heckmann, Richard from Synapse or our friend Dmitry. Sending them bug reports and teasing them a little bit about CPU usage to get them to push the limits even further is very much ok, but sorry, no excuse for being impolite unless you really are a 14 year old with weapon grade testosterone levels.

[Taron]

Don't do it, ZaBong, don't put gasoline on a little camp fire. It's not that bad. Saying "work ethics" doesn't have to mean anything more than the vanishing pursuit of highly optimized coding. And "optimized coding" doesn't only mean taking any code and just picking the faster instructions. It means looking for a different approach to achieve the same or a similar thing. But that may be impossible, when the actual kicker of the whole engine's concept can't be any faster, while the result of it as it is makes it so attractive for its qualities.
We cannot decide that, since we can't see the code. Every now and then, though, the "wrong" kind of poking can somehow inspire the right kind of dev... well, I said it already.

So, please, let's just move forward. Our positions are clear and I find them all great, actually, not like it was up to me, but if you really respect Dmitry and this thread, let's just move forward. You know how easily things can get out of hand.

[ZaBong69]

So, please, let's just move forward. Our positions are clear and I find them all great, actually, not like it was up to me, but if you really respect Dmitry and this thread, let's just move forward. You know how easily things can get out of hand.

You are right. Thank you.

[Taron]

(too much? )

[S0lo]

There is a (cpu) tax on quality. My preferred synths are all quite taxing. There is no free lunch.

You sound like a 14 year old who is bragging how he's playing CoD on max settings.
The main problem we are having is that a bunch of developers who wrote magnificent pieces of code showed or keep showing small or non-existing interest in developing more synths and that's why all those synths are seen as exceptions and not the standard.
A bunch of developers and companies couldn't beat one single Sylenth1 throughout years by shoving synth after synth and let's imagine if Lennard kept making regularly synths by now we would have 5 "Sylenths", maybe one of them could have been a modular synth, other one additive, third one a wavetable synth, etc.
If developers like Lennard kept making synths nowadays we would have over 20 synths with exceptionally good ratio of sound quality and CPU usage, his 5-6 synths and 10-15 from other developers who also made exceptionally optimized synths. Synths like that would become a standard, not CPU hogs.
Urs & Co. wouldn't be able to sell shit if things were different.

This is the main reason why your perception is skewed, you are simply misguided by developers with poor optimization skills that only a CPU hog can produce a high quality sound or that high quality sound goes hand in hand with CPU hogging.

Btw, exactly mentioned Dune (1 and 2) really light for my cpu. Though i should admit that unisson in Thorn i liked more. As well, Largo the same lightweight for me (but what regarding his unisson quality, here ambiguous moment - I would not call it no better and no worse, just other, with his own character).

That's because every single properly optimized synth uses around 50-80MHz of CPU power when you take a dry sawtooth and place a chord.
There's no difference between Synth1, Charlatan, Sylenth1, Dune1, Dune2, Harmor, Helix, etc. when you do this basic test.
All properly optimized synths will give you similar results with this test.

Thorn is using 9 TIMES more of CPU power for this basic task when compared to all properly optimized synths. Of course that things just get multiplied from that point. You can't expect from a synth with a poorly optimized oscillator algorithms to behave like properly optimized synths when its core, the base, starting point which is oscillator is screwed up.

Hive is using 6 times more of CPU power when compared to Dune2 when you do this basic test and Dune2 was released in 2014 and the beta of Hive was released in 2014.
It only shows you who has what work ethics, who will sell you a properly coded synth and who keeps relying on technology progress to swallow his spaghetti coding.

Of course that both Urs and Dmitry will market Hive and Thorn as light on CPU synths, but in reality both of those synths are CPU hogs when compared to properly optimized synths, not to mention their standard CPU hogs.

The whole point is that even though CPU can handle 20 instances of those synths, it doesn't change the fact that they are poorly optimized. If their oscillators were properly optimized they would use at least 5 times less of CPU power.

There is a highly deceiving factor for even the best developers out there regarding CPU usage. Which is, "developing on a high CPU spec machine".

It is not easy (infact very difficult) to spot brief CPU spikes that happen continuously and silently because of poor coding or say denormals...etc when your development is done on a high spec CPU. The dev would contentiously think that his code is running very low on CPU while coding because he only sees a 3% or 4% usage. But as more code is added over the months, with hundreds of lines of code, more and more CPU is taken silently. By the end of a year, the CPU has grown to like 10% to 15% on a very basic patch and the dev would not dream of delving back into every thing to optimize it properly. And even if he did, it won't be that much of an improvement at this point.

Old developers had the advantage of "low spec CPU" at their hands!!. They would be able to spot CPU problems easily and fix and come with better coding methods on a line by line basis every time they add a piece of code. Thats why I personally do half of my own development on an old low spec machine.

Off-course there are other factors too to improve CPU optimization, like using SSE2 or even assembly language, not using over bloated libraries, GUI hardware acceleration etc.. But the one above is of importance.

[Taron]

I'm just waiting for the day, when absolutely everybody has GPUs that can be used for more expensive computations. Too bad especially musicians tend to rely on their old hardware in fear of losing what they were creating before, I suppose, or because they are poor musicians, hahaha, which seems understandable to me. ...anyway, one day.

It's not necessarily how you instruct the CPU, but conceptual choices or possibilities that can lead to leaps in improvement. You could strip down all your computations to the fastest possible code, if the concept causes the trouble, there ain't no help. There are however certain compiler instructions that help with denormals, making that part much easier. And then there is the careful consideration, which operations can be done in the parameter setting portion, rather than in the processing routines. This can shave off a whole lot of time!
Somethings may have to be trade-offs, but others actually are not, but it may take some rethinking.

Then, today, memory access is often slower than actual computation, as well as having CPU buffers in mind, of course. So much to learn.