Studio One 4.1.4 - V - Reason 10.3 - |Performance Compared|

[THE INTRANCER]

With all the anticipation of Propellerheads Reason 10.3 being released, I thought I'd give it a whirl and to see for myself just how well it does in this little test I did to compare it on my system and to show you the results of how it compares to Studio One 4 (the current latest version). I could have done something bog standard but instead decided to use it an opportunity to get very creative over this Easter weekend with new music... as you'll see and hear.

I'll add the test files later...for now you can see how things compare..

Video Updated.





Download Link - Studio One 4.1.4 & Reason 10.3 + Midi Test Files

[antic604]

Care to summarize the results? I've no time (or intention) watching that video, although judging by your joyous post S1 is on top, which is hardly surprising because it uses "tricks" to pre-rended everything that's not armed for recording (as does Cubase or Reaper), but why would you care about such detail

[xylyx]

I just skipped to where I could see the results. It basically is 115 instances of Hive in S1 versus 42 in Reason. It did appear as though the test was done with hyperthreading off in Reason, which gives better results for some, although I have always found hyperthreading has improved my system's performance with Reason.

[chk071]

115 to 42 instances? Hell... didn't think Reason's VST performance is such an issue.

Anyway, that's what i always anticipated. When Reason builds in VST support, they'll be light years behind the competition which is in this game for centuries.

[Jim Roseberry]

Care to summarize the results? I've no time (or intention) watching that video, although judging by your joyous post S1 is on top, which is hardly surprising because it uses "tricks" to pre-rended everything that's not armed for recording (as does Cubase or Reaper), but why would you care about such detail

Studio One uses "Hybrid Buffering"... which Logic and Samplitude have done for years.
ProTools is also now using Hybrid Buffering.

Hybrid buffering isn't Pre-Rendering...
A large processing buffer (more CPU efficient) is used for all tracks merely playing back.
Tracks that need to be monitored in realtime are processed using the ASIO buffer size.
This greatly improves performance.

When it comes to *effectively* working at smallest ASIO buffer sizes, Studio One is currently one of the best performers... as are Logic, Samplitude, and ProTools.
Hybrid Buffering is the reason why...

Reaper is also one of the most CPU efficient DAW applications.
It uses "Anticipative EFX Processing" which is a look-ahead buffer.

[Funkybot's Evil Twin]

...and pretty sure ASIO Guard in Cubase works similarly.

The problem with the Studio One approach is that their implementation works in such a way where many effects are completely incompatible with the hybrid-buffering system. It basically makes a copy of the signal chain behind the scenes, and any plugins that don't have hardwired VST Host Automation parameters (like many guitar amp sims, modular plugins, etc.) won't reflect parameter changes in the monitoring path. So you adjust the drive knob on your amp sim and nothing happens. You have to disable monitoring and re-enable it to hear the change. Very poor implementation IMO and one that got more and more frustrating for me as time went on.

[Trancit]

...and pretty sure ASIO Guard in Cubase works similarly.

The problem with the Studio One approach is that their implementation works in such a way where many effects are completely incompatible with the hybrid-buffering system. It basically makes a copy of the signal chain behind the scenes, and any plugins that don't have hardwired VST Host Automation parameters (like many guitar amp sims, modular plugins, etc.) won't reflect parameter changes in the monitoring path. So you adjust the drive knob on your amp sim and nothing happens. You have to disable monitoring and re-enable it to hear the change. Very poor implementation IMO and one that got more and more frustrating for me as time went on.

Afaik all problems you describe here is only valid if you want to have "Low latency monitoring" enabled...
If you turn it off, it´s not a problem at all... I might be wrong though...

[Trancit]

Studio One uses "Hybrid Buffering"... which Logic and Samplitude have done for years.
ProTools is also now using Hybrid Buffering.

Hybrid buffering isn't Pre-Rendering...
A large processing buffer (more CPU efficient) is used for all tracks merely playing back.
Tracks that need to be monitored in realtime are processed using the ASIO buffer size.
This greatly improves performance.

When it comes to *effectively* working at smallest ASIO buffer sizes, Studio One is currently one of the best performers... as are Logic, Samplitude, and ProTools.
Hybrid Buffering is the reason why...

Reaper is also one of the most CPU efficient DAW applications.
It uses "Anticipative EFX Processing" which is a look-ahead buffer.

These buffering methods result in perhaps being able to run 3 or 4 instances more... but never ever a difference of 70 instances...
Either it´s a problem with Reason and Hive or Win7 or a messed up system...
I would like to see a comparison (in numbers not video) on the same machine between S1 and Ableton or FLS... then I can tell you more...

[Funkybot's Evil Twin]

...and pretty sure ASIO Guard in Cubase works similarly.

The problem with the Studio One approach is that their implementation works in such a way where many effects are completely incompatible with the hybrid-buffering system. It basically makes a copy of the signal chain behind the scenes, and any plugins that don't have hardwired VST Host Automation parameters (like many guitar amp sims, modular plugins, etc.) won't reflect parameter changes in the monitoring path. So you adjust the drive knob on your amp sim and nothing happens. You have to disable monitoring and re-enable it to hear the change. Very poor implementation IMO and one that got more and more frustrating for me as time went on.

Afaik all problems you describe here is only valid if you want to have "Low latency monitoring" enabled...
If you turn it off, it´s not a problem at all... I might be wrong though...

Right, but that defeats the whole purpose of the dual-buffer design. With Low Latency Monitoring off, you're back to the old, single buffer way which uses more CPU most of the time.

Those issues don't exist in other DAWs (Cubase, Reaper) so it's clearly just poor design in that regard.

[Jim Roseberry]

Speaking strictly from a performance perspective:
When pushing heavy loads at the smallest ASIO buffer sizes (say working at 96k using a 32-sample ASIO buffer size), Cubase with ASIO Guard isn't among the best performers.

With Studio One, if you make any changes to the plugin, you do indeed have to disable/re-enable Low Latency Monitoring.
Not the most elegant solution, but you can effectively work at super small ASIO buffer sizes (meaning no audio glitches - even with large scale projects).

If you set Cubase to 96k using a 32-sample ASIO buffer size (with ASIO Guard at max and Multi-Processing enabled), you'll see large realtime performance spikes. Disable Mult-Processing, and those realtime performance spikes diminish.
If you're working with larger scale projects, disabling Multi-Processing isn't practical.

ProTools handles monitoring more elegantly... but won't allow ASIO buffer sizes lower than 64-samples.

Each DAW has strengths/weaknesses.
If you're a more hard-core composer, it's hard not to choose Cubase.
It's forte' just isn't running heavy loads at the smallest ASIO buffer sizes.

Digital Performer 10 uses pre-rendering... but it's among the worst performers when it comes to running heavy loads at the smallest ASIO buffer sizes.
When set to 96k using a 32-sample ASIO buffer size, insert a single instance of Helix Native (to play/monitor thru in realtime)... and it's nothing but gliches (even on the fastest machine money can currently buy).
DP10 has some great features. Pushing the limits of ultra low latency performance isn't its forte'.

[Trancit]

...Right, but that defeats the whole purpose of the dual-buffer design. With Low Latency Monitoring off, you're back to the old, single buffer way which uses more CPU most of the time.

Again afaik: No... the opposite is true...
1. the low latency mode doubles the (suitable) fx on this track, which causes more CPU load in the moment...
2. The drop-out protection is what allows your system running more instances of plugins... but it raises the buffer size internally which causes latency...
The low latency mode is just a workaround to cheat for the moment to be able to play your input with less latency...

Those issues don't exist in other DAWs (Cubase, Reaper) so it's clearly just poor design in that regard.

This might be true... but at least it made S1 much more useable on many systems... better something with a bit of problems than nothing...
While I have to say... I am in the lucky position to never play anything live... so for me it´s only advantage but no disadvantages at all!!

Speaking strictly from a performance perspective:
When pushing heavy loads at the smallest ASIO buffer sizes (say working at 96k using a 32-sample ASIO buffer size), Cubase with ASIO Guard isn't among the best performers...

This test in the vid was made with 512 sample buffer... no dropout protection cheat of any DAW makes a difference of 70 instances at this sample level... not even more than 5...

[Jim Roseberry]

These buffering methods result in perhaps being able to run 3 or 4 instances more... but never ever a difference of 70 instances...
Either it´s a problem with Reason and Hive or Win7 or a messed up system...
I would like to see a comparison (in numbers not video) on the same machine between S1 and Ableton or FLS... then I can tell you more...

Performance difference (depending on what you're running) can vary *radically* between the various applications.

I'll use the Presonus Quantum (audio interface) as an example (latency can be set extremely low).
Set Quantum to 96k with a 32-sample ASIO buffer size (total round-trip latency is 1ms).
That's pushing the limits of ultra low latency performance.

With Quantum set at 1ms total round-trip latency:
Load Helix Native (single instance) into your DAW of choice... and setup a patch that uses two Cab IRs.
Now enable record/monitor... to allow playing/monitoring in realtime thru Helix Native.
CPU use will be significant (even with the fastest machine money can buy).

Digital Performer 10 can't do this without glitches (still glitches with ASIO buffer set to 64-samples).
Cubase 10 can do this with an occasional glitch (performs better with Multi-Processing disabled).
Studio One can do this with zero glitches.
With ProTools, you have to set the ASIO buffer size to 64-samples (minimum allowed). Zero glitches
Samplitude Pro X4 can do this with zero glitches.
Reaper 5 can do this with zero glitches.
Ableton Live 10 can't do this without glitches.

There are things I don't like about Studio One 4, but its Hybrid Buffering scheme allows you to *effectively* work at the smallest ASIO buffer sizes.
With Digital Performer 10, you just can't effectively run these types of loads.

If you're not pushing the limits of ultra low latency performance, you may not care about any of this.
But performance difference under these circumstances can be radically different... to the point of making work practical... or not.

[Funkybot's Evil Twin]

Just to add to your point: that will also vary wildly based on the audio interface/driver. RME's will be much more efficient than Focusrite interfaces for example. Last time I looked at DAWbench, the Quantum interfaces performed very well, but still not as well as the UFX+ with Thunderbolt for example, and similarly to the UFX+ in USB mode if I remember correctly. The charts I found the DAWbench site seem out of date, but I remember seeing newer compares that included the newer RME and Presonus interfaces somewhere in the massive Gearslutz thread.

Found it here (bottom chart, first post): https://www.gearslutz.com/board/music-c ... -base.html

[reggie1979]

I'm really confused by all of this. The detail and minutia throw me. All I want to know is how/can I get my work done.

I used reaper for years and never had any issues. Studio one seems to be quite nice these days. Reason is so-so. It works, I can get the job done. Cubase seemed to be the hungriest. Always had to be careful of the CPU meter.

[Jim Roseberry]

2. The drop-out protection is what allows your system running more instances of plugins... but it raises the buffer size internally which causes latency...
The low latency mode is just a workaround to cheat for the moment to be able to play your input with less latency...

With Hybrid Buffering:

Tracks that are merely playing back, any amount of latency can (and is) compensated.
This effectively makes the latency of these tracks moot... as the latency is never actually experienced.

Enabling Low Latency Monitoring uses the ASIO buffer size just for those tracks where (software based) input monitoring is necessary.

This is a whole lot more CPU efficient than standard buffering.

In the case of pushing super heavy loads at the smallest ASIO buffer sizes, it's the difference between being able to effectively work... or not.