Does the performance of reason benefit with audio interfaces with built in DSP?

[okaino]

Im going to try to skip story time and get to the point of this as best as possible. To make a long story short yesterday I ran into the issue of the sends effects routed back as a mixer channel not being delay compensated. 

Searched this issue on the reasontalk forum found this
viewtopic.php?f=4&t=7519372&p=518753&hi ... on#p518753

which led to me to this link as well
https://help.reasonstudios.com/hc/en-us ... what-s-up-


The best solution I personally found  is to just route into the returns so i can get the delay compensation to happen which is my primary concern at this particular moment because im in the middle of a mix. Delay compensation is more important to me right now then the convenience of routing sends to a mixer channel.


In the forum article about the delay compensation issue with the sends

The first suggestion was to lower the buffer size which isnt possible for me this far in the mixing stage and also points to that the delay would be more exaggerated. 

2nd recommendation was to uncheck the "render audio using audio card buffer size setting." Which i had activated.


This brought me to the question....Does the performance of reason benefit with audio interfaces with built in DSP?

To also further define this question ill add i know products such as UAD and antelope audio have these interfaces, and you can use their plugins in reason which lightens the workload on the cpu because of the built in dsp. Im not interested in that at all for the sake of this question.

I want to know if Reason uses the built in dsp on audio interfaces based on the audio setting "render audio using audio card buffer size setting"


Any information towards this is greatly appreciated!

Thanks!

[buddard]

I want to know if Reason uses the built in dsp on audio interfaces based on the audio setting "render audio using audio card buffer size setting"

In short, no.

The benefit you get from using the audio card buffer size setting is that the VST plugins can process more samples each time it's its turn to run, and won't have to be called as often, thus reducing the processing overhead for many of them (depending on how they were implemented).

In earlier versions of Reason, the internal buffer size in Reason was always locked to 64 samples, and that would cause performance problems for many VSTs that were not implemented with this in mind.

[syncanonymous]

In earlier versions of Reason, the internal buffer size in Reason was always locked to 64 samples, and that would cause performance problems for many VSTs that were not implemented with this in mind.

Earlier than version 11?

[buddard]

In earlier versions of Reason, the internal buffer size in Reason was always locked to 64 samples, and that would cause performance problems for many VSTs that were not implemented with this in mind.

Earlier than version 11?

Earlier than 10.3.
You can still use the old 64-sample rendering by not checking the "use audio card buffer size" setting.

[Dante]

This brought me to the question....Does the performance of reason benefit with audio interfaces with built in DSP?

If you are using the DSP for tasks that you would otherwise have to use Reason for then yes. Its called 'offloading'.

For example, if you use UAD2 Mastering Plugins instead of Reason Mastering plugins then that frees up native CPU for other Reason tasks.

[mcatalao]

As Dante said, that only happens if you use UAD plugins in place of Reason's Re's or devices. And it will only work with that brand's plugins because the acceleration card will not offload.

The DSP of a normal audio card, is only to convert the digital audio of several sources into analogue audio.

You have 3 types of devices:
- Simple audio cards that only convert AD and DA on for multiple sources
- Simple acceleration cards that offload the processing tasks of the plugins they support (example UAD PCIe accelerators, AVID Accel...)
- Hybrid Audio card with DSP (UAD Apollo, RME Fireface 802, etc). The RME makes it a bit hard to offload the dsp, but its doable.

I've had UAD pci accels in the past, but imho there are some pitfalls on using these kinds of devices. First you end up being tied to a vendor and depend a lot on UAD or SSL or xx brand for your plugins. So if you need this or that plugin you end up using it and cost you DSP on the CPU anyway because though they have a lot of plugs, there's always something more out there. Then there is the price. If you think about it, a DSP2 quad costs about 800 eur (the DSP Accel only), and the least expensive hybrid counterpart is an audio card with 4 channels that costs about the same and it doesn't have the same performance. If whats important about an accel card like this is the amount of instances you can load of the most important plugins, that's a very limited option, imho. The result is you offload with their own plugins, and limit the instances you can load, and to load more instances you need additional cards (you can have up to 4x DSP2 quads in a pc).

TBH, unless it's really important to have 0 latency on these plugins, you'd be better off investing in a bigger CPU. An update between an I5 and an i7 or a i9 with double the cores and good speed and amount of ram, doesnt cost you 800 eur. Heck if you happen to create your own pc's you can build your i9 10850k under 1000 eur!

[kuhliloach]

Thanks for going over this! I was actually considering UA as a vendor--no more. If the only plug-ins that can utilize the hardware are their own the product isn't worth the price. Meaning, it won't perform any better than an interface that is ten times cheaper.

[syncanonymous]

Thanks for going over this! I was actually considering UA as a vendor--no more. If the only plug-ins that can utilize the hardware are their own the product isn't worth the price. Meaning, it won't perform any better than an interface that is ten times cheaper.

Whoa nellie, maybe for offloading DSP tasks...
there are other important issues to consider:

UA Apollo has +25dB analog input headroom
You aren't getting that for 200 currency units! afaiaa
for the cheaper device you'll be lucky to get +18dB input levels
which is an absolute deal breaker for me

And considering mic pres:
I have a RME UFX+
I just tested the mic pres in my Sound Devices 722 compared with the RME pres
The Sound Devices pres are considerably better
In my test I recorded acoustic steel string with a Rode NT2

those are just 2 basic differences off the top of my head
there will be more reasons if investigated deeply enough

like i/o to start it off...the UFX+ has a staggering amount of i/o possibility

[tumar]

No. Just invest in multiple core CPU. I had TC Konnekt 24D and Saffire Pro 24 DSP (and TC Powercore itself), never liked DSP effects handling.

[bitley]

For performance invest in Thunderbolt or PCI tech.

[mcatalao]

Whoa nellie, maybe for offloading DSP tasks...
there are other important issues to consider:

UA Apollo has +25dB analog input headroom
You aren't getting that for 200 currency units! afaiaa
for the cheaper device you'll be lucky to get +18dB input levels
which is an absolute deal breaker for me

(...)

Well, first the op was talking exclusively about DSP, so there's that...

TBH, I fail to see the relevance of the +25 dB headroom in this case. It's very important for the owner to know that for correct gain staging between different devices, for example to know if he/she has to pad down a hotter pré amp or D.I. but more than that, I fail to see where you're going to regarding this thread! It's imho just to confuse things. And if that's important you can find a bunch of other cards with 25 dB analogue "headroom".

Heck even my Beringer ADA 8000 has 26 dBu line inputs.

[groggy1]

UAD pros:
The Dsp card lets you get some fancy plugins that can only be run on UAD. So if you want a Neve branded 1073, you need the UAD “dongle”

Offloads some processing from your cpu. With modern processors, it’s helpful to some but not others

Cons of UAD
Expensive

Other manufacturers make a Neve 1073 clone that’s nearly as good (or better, depending on taste)

Using UAD brings in some latency during the mix stage. But UAD will tell you a million times that latency during mix stage is perfectly normal. And that their Apollo 0 latency during tracking is how you get around it. But for folks that want to blur line between tracking and mixing, it’s a bit annoying

[fixed typo]

[okaino]

Thanks for this information!

[tallguy]

Has there ever been any attempt to create a common platform for DSP cards? Like graphics cards had OpenGL?

[PeterP]

Has there ever been any attempt to create a common platform for DSP cards? Like graphics cards had OpenGL?

There is an attempt from ROLI and JUCE that's been in development for a while. It's interesting but I'm doubtful if they'll succeed in convincing the industry at large.

It's a bit like OpenGL shaders but for fx and instruments.

https://github.com/soul-lang/SOUL