Poor PC Performance

[normen]

The big problem is, no matter how tricky you get with synth routing, theres a VST synth that sounds incredible out of the box without all that complicated wiring...Omnisphere/Sylenth/Massive/Serum....the list goes on.

Theres whole DAWs without all that complicated wiring Are you trying to make my point for me?

[antic604]

If you compare for example to advanced automation routing systems in other DAWs - they always work at the buffer size rate. Meaning if you modulate a synths parameter it will be "quantized" to the buffer size. If you compare to advanced modular systems like Reaktor then you don't have the single modules in the context of the DAW, i.e. you can't route the output of a module in Reaktor directly to some synth that sits in a DAW insert.

Can you please stop posting this, as if it was true?

Many DAWs long time ago surpassed Reason it terms of modular flexibility and - in particual - ease of managing the complex setups, and its strength lies in graphical representation of it, which is not to everyone's liking.

Here's an example of Bitwig's LFO going from 0.02Hz to 50kHz (yes, kilohertz):

Since v2.2 it even has a Audio Rate modulator, which you can use to modulate one signal with exact waveform of another, giving you a FM-like modulation.

[normen]

If you compare for example to advanced automation routing systems in other DAWs - they always work at the buffer size rate. Meaning if you modulate a synths parameter it will be "quantized" to the buffer size. If you compare to advanced modular systems like Reaktor then you don't have the single modules in the context of the DAW, i.e. you can't route the output of a module in Reaktor directly to some synth that sits in a DAW insert.

Can you please stop posting this, as if it was true?

Many DAWs long time ago surpassed Reason it terms of modular flexibility and - in particual - ease of managing the complex setups, and its strength lies in graphical representation of it, which is not to everyone's liking.

Here's an example of Bitwig's LFO going from 0.02Hz to 50kHz (yes, kilohertz):

Since v2.2 it even has a Audio Rate modulator, which you can use to modulate one signal with exact waveform of another, giving you a FM-like modulation.

You can use the signal of this to modulate a parameter of another random VST synth? Could you please have posted this sooner including an example?

[Rason]

I'm on 9.2 because I don't want VST atm. I only have a few I use to mix vocals in Reaper.

I want 10, is it slower even with no VST?

10 compared to 8.2 is just a tiny little bit slower if it helps (comparing tracks with no VST). Try urself download trial, it keeps both versions separate on the disk. I wanted to gear up before upgrade but realized the decrease in performance is neglectable.

[antic604]

You can use the signal of this to modulate a parameter of another random VST synth? Could you please have posted this sooner including an example?

Sure you can:

The last option I chose ("pitch") is keyboard tracking (e.g. A4 = 440Hz)

[normen]

You can use the signal of this to modulate a parameter of another random VST synth? Could you please have posted this sooner including an example?

Sure you can:

The last option I chose ("pitch") is keyboard tracking (e.g. A4 = 440Hz)

Cool. Can’t listen right now but I believe you when you say it’s actually modulating at high frequency. But I still wouldn‘t call Bitwig studio a generic DAW - it puts itself as the „modularity king“ so dealing with that issue could be expected? And nobody made elaborate instance comparisons like people do here between Reaper and Reason. I guess Bitwig won‘t exactly excel in Kontakt instance count either (if you modulate). So yeah, it‘s probably not correct saying „in other DAWs“ but I should have said „in the major other DAWs“.

The simple fact is that normal VSTs only read their parameter values once per buffer so you have to work around that one way or the other.

[Smee]

Let's stay on topic - this isn't an anti-Reason rant, it's about Reasons raw performance power when it comes to running large orchestration templates, something that affects probably 5% of its users.

I think much of the problem could be fixed by Propellorheads introducing multiple MIDI sends, this way one instance of a sample can be drawn on for up to 16 instruments. Of course, this would mean folder tracks would be necessary too - which is about time, another thing that makes me shake my head as to why it's not implemented.

Im wondering how much CPU processing is reserved for that SSL mixing desk? If you replicated that in any other DAW, you'd probably lose about 20% resources to just apply EQ's and compressors to every channel, even though they're mostly inactive.

[normen]

Im wondering how much CPU processing is reserved for that SSL mixing desk? If you replicated that in any other DAW, you'd probably lose about 20% resources to just apply EQ's and compressors to every channel, even though they're mostly inactive.

No, if you don't enable EQ or compressor then the mixer takes the same amount of CPU as in every other DAW, a simple summing operation for all buffers from all channels. There isn't any other "analog magic" or something being emulated - not in the EQ or compressor either, they're basically "ideal" as designed by the SSL engineers, without the limitations of the analog hardware used to build them

[swamptooth]

Thanks for the advice guys. I'm really bummed about Reasons performance issues with multiple VSTi instances. For me, it's not just a 'Kontakt' problem, even running standard synths, I'm lucky to get maybe 10-12 instances of Diva or RP Blue 2 running. I'm also not convinced its a PC hardware issue, as Reaper runs easily double these figures without hiccuping. Also, its inability to input multiple channels of MIDI to a VI meaning you have to run individual instances of Kontakt (or any sampler) for every single MIDI channel just blows my mind.

Admittedly, my situation is unusual, most people won't run stupid numbers of VI's like I do, but I'm a composer and need to have often upwards of 100-200 tracks running, often with 20 or 30 instances of live VIs or VST effects.

I love Reason's interface and workflow, but it looks like I'll need to retire it. Damn.

100-200 tracks lol?? Then you need serious power for that. Like Junkie XL power computing. We're talking probably 50k of power and computing. I don't believe Junkie XL uses Reason either. He writes with Cubase I believe, I could be wrong. Not that it matters. He has all the real synths anyways.

This is a pretty simple template in cubase using only Albion 1 and a smattering of pianos and synths and other spitfire audio libraries. 5 year old i7 with 16gb ram. It's really all about project organization and routing which really limits the headaches involved.
Then again this is why I use Cubase AND Reason.

ALBION TEMPLATE.jpg (716.48 KiB) Viewed 2424 times

[househoppin09]

Can someone explain why it's apparently so common for people to assign a separate track to every articulation? I've always found that to be a nightmarishly cumbersome way to work. It depends on the library I guess, but if I were using Albion One I'd think the obvious approach would just have a track for each major group of articulations and keyswitch/expression-map them in the performance (with later editing and tweaking as necessary). If you need two or three different articulations from the same patch playing on top of each other, add instances one by one as necessary. What exactly am I missing? Why do people insist on creating these over-the-top templates?

[swamptooth]

Can someone explain why it's apparently so common for people to assign a separate track to every articulation? I've always found that to be a nightmarishly cumbersome way to work. It depends on the library I guess, but if I were using Albion One I'd think the obvious approach would just have a track for each major group of articulations and keyswitch them in the performance (with later editing and tweaking as necessary). If you need two or three different articulations from the same patch playing on top of each other, add instances one by one as necessary. What exactly am I missing? Why do people insist on creating these over-the-top templates?

Ease of use. First you don't have to remember the keyswitch. Second layering is already set up so you don't have to stop and edit the project when you're in the middle of writing.
I use a 49 key keyboard controller most of the time and the keyswitches are outside the range I'm playing in.
I use a keyswitched template sometimes as well if I'm transcribing in staff view or doing step input but that's increasingly rare.
I thought this method was a bit over the top as well but once I started using it I found it to save hours of unnecessary setup and work.
My generic reason template is only 65 tracks though.

[househoppin09]

Ah, the 49-key controller explains a lot of it. I can't imagine doing orchestral work without an 88-key controller. I also have a very performance-centric compositional style--i.e. if I can't at least approximately play the part I want in real time (allowing for some generous editing later, of course), I'm not going to be satisfied in any case. So, between those two factors, I haven't found the track-per-articulation approach to be of much use.

[swamptooth]

The person who convinced me that this was the way to go is Guy Rowland. If you haven't seen his videos on think space education or his own channel I highly recommend them.

[chimp_spanner]

If you're using cubase you don't need to remember the keyswitches. Create an expression map for each instrument (takes a bit of time initially). And within that map you create articulations that reference the key switch. Once it's done, inside a single part you can highlight an individual note and say that's spiccato, or tremolo, or marcanto or whatever. It's really great. The only downside is that you can't have more than one articulation playing at a time, although it'd be rare to have violas doing trem and pizzicato at the same time. You'd just use two different string sections. Or if you really need to, duplicate tracks as and when.

Could cut the track count down immensely!

[swamptooth]

If you're using cubase you don't need to remember the keyswitches. Create an expression map for each instrument (takes a bit of time initially). And within that map you create articulations that reference the key switch. Once it's done, inside a single part you can highlight an individual note and say that's spiccato, or tremolo, or marcanto or whatever. It's really great. The only downside is that you can't have more than one articulation playing at a time, although it'd be rare to have violas doing trem and pizzicato at the same time. You'd just use two different string sections. Or if you really need to, duplicate tracks as and when.

Could cut the track count down immensely!

Yeah I've invested about 80 hours of my life into expression maps. Certainly fine for some uses and I have templates that use this method. I typically use them if I do score layouts.
As an aside multiple concurrent articulations are possible depending on the library. Spitfire libraries -- many of them -- offer this functionality, though the expression map implementation is convoluted.

[Smee]

This is all really cool stuff, but I'm presuming no one has managed to get Reason to perform anywhere near a well as Cubase, Reaper or pretty much any other DAW when it comes to large orchestral templates?

Oh, and BTW - by 'orchestral', that doesn't' mean I only violins and cellos on every track. The vast majority of my tracks are synths or samples, they're just called orchestral templates because they pre-load all the shit you plan on using on the project. But even still, I doubt Reason would cope well with 100 instances + of Thor running or any other of its instruments, might be wrong, haven't tried it. I remember running a project a while back with about 60 or so stereo wave files and it hickuped like mad even with that.

[normen]

This is all really cool stuff, but I'm presuming no one has managed to get Reason to perform anywhere near a well as Cubase, Reaper or pretty much any other DAW when it comes to large orchestral templates?

Oh, and BTW - by 'orchestral', that doesn't' mean I only violins and cellos on every track. The vast majority of my tracks are synths or samples, they're just called orchestral templates because they pre-load all the shit you plan on using on the project. But even still, I doubt Reason would cope well with 100 instances + of Thor running or any other of its instruments, might be wrong, haven't tried it. I remember running a project a while back with about 60 or so stereo wave files and it hickuped like mad even with that.

Reason easily runs hundreds of Thors, the internal instruments and effects are VERY efficient also because they don't over-do it on the "emulating Alan Parsons favorite XLR cable" department. This also goes to show that Reason isn't flawed in some way but that VST and Reason are simply not a very good match. Reason was especially known for it's CPU efficiency!

[antic604]

Reason easily runs hundreds of Thors, the internal instruments and effects are VERY efficient

I think that was true until Europa & Grain, both of which seem to be much more taxing than the "old" stuff. Wonder, if it's because of the RE format or ar they really that much better.

[normen]

I think that was true until Europa & Grain, both of which seem to be much more taxing than the "old" stuff. Wonder, if it's because of the RE format or ar they really that much better.

It's more about what they do really I suppose. Like Parsec is also relatively taxing as a synth simply because it's additive synthesis. For an additive synth it's relatively light on CPU though

[esselfortium]

I think Reason's poor VST performance is just growing pains with a new implementation. From what I understand, they're aware of the issue and are treating it seriously as something to improve. I'm pretty eager and impatient to see such a fix actually happen, but in the meantime, categorizing it as an intrinsic and unchangeable trait of Reason is just spreading misinformation.

[EnochLight]

I think Reason's poor VST performance is just growing pains with a new implementation. From what I understand, they're aware of the issue and are treating it seriously as something to improve. I'm pretty eager and impatient to see such a fix actually happen, but in the meantime, categorizing it as an intrinsic and unchangeable trait of Reason is just spreading misinformation.

^^ THIS ^^

x1000000.

[antic604]

I think Reason's poor VST performance is just growing pains with a new implementation. From what I understand, they're aware of the issue and are treating it seriously as something to improve. I'm pretty eager and impatient to see such a fix actually happen, but in the meantime, categorizing it as an intrinsic and unchangeable trait of Reason is just spreading misinformation.

^^ THIS ^^

x1000000.

Unless it proves to be unfixable, as in rooted deeper into Reason's inherent architecture.

I mean the same discussions (to slightly lesser degree) happen with regards to Live or Bitwig, because people compare them to Cubase or Reaper, that use double buffers and other tricks to "pre-calculate" playback on all but record-enabled tracks, whereas Live or Bitwig actually encourage you to swap devices & presets on the fly, completely change the arrangement (with their clip launchers), automate everything live and at once, use random modulators, facilitate complex inter-track routing of audio & MIDI, etc. It's apples to oranges and it applies to Reason as well, which - despite being a linear DAW - is much closer to Live & Bitwig in terms of modularity.

Actually, when I tested some plugins in Reason, Live and Bitwig the differences in performance were really small for the most part (1-3 plugin instances), although there was a small group of plugins where the difference was very substantial, like 5 vs. 30 instances and I suspect those were the ones not accepting Reason's fixed 64 samples buffer. Hope this can be fixed.

[normen]

I gave some examples on how it could be "fixed" in Reason, like making a larger buffer for VST plugins etc. but its STILL shoehorning x100000

[EnochLight]

^^ THIS ^^

x1000000.

Unless it proves to be unfixable, as in rooted deeper into Reason's inherent architecture.

Don't worry...

[Smee]

Well, the native synths should play better.

Just did a quick and nasty benchmark on both platforms native synths - just loaded them up and played a single chord until something broke....

In Reaper with its native synth: 1260 record monitoring (before it started getting totally unrecognisable as sound), with a further 1260 record enabled instances - thats 2520 single oscillator VST synths on a single project. The image below shows about 40% CPU on idle, jumps to about 90% on playing - but it played, and technically I could record a 1260 oscillator synth.

Reason, I got to 130 instances of Thor before my system basically nosedived and refused to work. Had to reboot. These were single oscillator patches with everything bypassed, so it was basically the same as Reaper. Of course, Reason doesn't allow you to play multiple instruments at once, and it would be unfair to put them in a combinator as you'd need a 130 track mixer and all the routing to get them all working, so I couldn't measure the performance actually playing 130 instances of Thor - I would imagine very poor. So I recorded a single chord and included a copy of the MIDI on each track, in essence doing the same job - basically refused to work. I got 118 tracks going for about 10 seconds before the 'Your system is too slow' message.

I had more luck with Subtractor, getting 280 tracks before the dreaded 'Your system is too slow' message. Maelstrom, not so good - only 176 tracks. And, as a side note, all of this took an eternity to load and unload, very painful to sit through.

What I did find rather concerning is I also tested wave playback. I know Reaper does over 1000 tracks of mono wave playback easily, I tested that ages ago. But in Reason the system glitched at 420.

I am convinced you pay a heavy penalty for Reasons GUI, its the major difference between the two platforms. Reapers VST instruments are nearly DOS level basic in their operation and interface, yet it sounds the same as Thor and runs rings around it for performance.

Of course, a more accurate test should be performed than me just loading up instances until it dies, but its a fairly good indication there is some serious optimisation issues in the platform.

Synths.jpg (440.83 KiB) Viewed 2216 times