Will I be able to enhance my workflow by using an external DSP audio interface?

So as everyone might have noticed by now, Black Friday is coming close. And I was thinking about upgrading my audio interface from UR-RT 2 to something like UAD apollo series or Apogee Symphony Desktop that has built-in DSP in it.

My question is that, by using one of those DSP audio interfaces will I be able to not worrying too much about my Cubase CPU meter (a.k.a. ASIO meter) anymore? Does DSP prevent CPU meter from going up? Lately my CPU meter is hitting the ceiling as multiple VSTs and plugin inserts going on without freezing or rendering the tracks. That is because I want to change things at anytime I want to (I sometimes render individual tracks into Stereo bus track if I really need to though).

Iā€™m asking this kind of stupid question here because I read that CPU meter is not actually showing CPU usage on my PC. When the meter is hitting around 80% on Cubase and I seemy CPU is actually only running around 30% in Task Bar. So I donā€™t really know what else to do to maximize and utilize my machine power for Cubase.

If the DSP helps me create/mix songs very well in the future in that way, Iā€™d like to have Apogee Symphony Desktop as this one seems stable enough to go along with Windows 10 USB connection so far. Can anyone confirm this fact that DSP helps creative workflow? Thanks and sorry if my question is not clear enough for you.

My PC spec
Win10
16GB RAM
Intel Core i7-8700K
NVIDIA GeForce GTX 1060 3GB
Audio IF UR-RT2
Cubase 10.5 (already updated from the patch back in 2021 March)

You are probably putting to much on a single channel or you donā€™t have a lot of cores in your cpu, make sure you are using group tracks and effects tracks appropriately. I have a bunch of DSP effects in my antelope interface and I never use them, the work flow isnā€™t the best. Maybe its better with UAD, I donā€™t know.
For reference I do several mixes a month and have only had to freeze tracks on a mix once this year. Iā€™m running 14 core (28 thread) with all cpuā€™s manually clocked at 5.1Ghz.

If youā€™re not experiencing tangible problems (clippings,drops,freeze,etc.) do not concern yourself about the CPU meter. Donā€™t let it scare you into buying more gear!

If ever you encounter real issues, a simple upgrade on you PC will solve it for a fraction of the price of HW DSP gear. And a fraction of the complications!

UAD and Apogee build great DSP gear! But unless you run ā€œextremeā€ projects, like a large symphonic orchestra in VSTs, using a decent PC wonā€™t require an extra DSP in HW.

Try first simple workflow ā€œclean upā€ like proposed by DodgingRain, and donā€™t waste resources on silly overkill resolutions like 32bit/384kHz.

You might also be able to identify 1-2 poorly written plugins that hordes too much CPU resources and replace it.

Make sure also you donā€™t have unwanted applications (not related to your actual work) running in the background.

But anyway, unless you experience audible problems, you can just turn off your CPU meter! :wink:

No, I donā€™t have that powerful CPU on my PC as Iā€™m an amateur. Itā€™s a Intel Core i7-8700K and recently optimized a setting for CPU power usage on Cubase and it got a little bit better. With that being said Cubase crashes when I load up some CPU intense vsts or plugins sometimes and I donā€™t like it. I thought DSP could also help me preventing crashes in that case. Also, yea maybe Iā€™m wasting some extra power on unnecessary audio processes. Thanks for your input!

You already have a faster CPU than I have - very unlikely to be the issue.
For large projects however, make sure you have at least 8GB of RAM (if not 16GB pref.).
Using an SSD is also a great upgrade if you donā€™t have already.

In any case, using external DSP is not a solution.
Itā€™s valuable for live latency-free monitoring/recording, not so much for mixing/producing.
As a hint, this excellent article from SOS does not mention DSPs at allā€¦

A badly written plugin or VST is more likely to make you crash than having too many.

I donā€™t think I use any bad quality plugins almost at all since I mainly use iZotope neutron 3 and Ozone9. But itā€™s probably matter of vsts at this point because I rely heavily on IKmultimediaā€™s CPU intense vsts very often. Itā€™s safe to say that their products made me think of buying a DSP HW lol. And yea to make a great sounding mix, DSP is not required at all tbh I get that. But I hope you understand that I posted this topic just to make my workflow unlimited from CPU limitation and to be able to work on anything at any time I want to. Thanks for your comment!

I donā€™t have an UAD myself. I somehow read from your posts that you might want to run your iZotopes or IK Multimedia plugins on an external DSP. However as I understand, UAD allows their specific plugins only on the DSP. A DSP wonā€™t run Intel CPU compiled code. With a DSP you could off-load plugins by using the equivalents from UAD and free up some realtime resources for iZotope.

Yeah in case there is some confusion - as Marco wrote the DSP in Universal Audioā€™s devices only runs their plugins. So if you buy an interface or a DSP-only product they will include some plugins. Your IK or Izotope plugins will not run on that DSP. So if you want to keep using those plugins you wonā€™t solve your problem that way.

What you can do is swap your IK or Izotope plugins for UAD-2 plugins. That way you can reduce the load on the CPU and shift it over to the DSP. But, if you didnā€™t get plugins you need included when you bought the hardware with DSP youā€™ll need to buy that separately.

I personally find that UAā€™s plugins generally speaking sound very good, and their hardware emulations sound excellent. I think the pricing is fair especially when there is a sale. But you do pay more than just for the hardware and you also are ā€˜stuckā€™ with using those plugins on UA DSP. That might not be a big problem, but itā€™s something to consider.

In terms of just upgrading your work(flow) experience getting a better CPU or new computer is probably the most economical solution. In terms of getting better sounding plugins then arguably getting UA plugins could be better.

Yes, exactly those plugins donā€™t run outside the DSP. Also, neutron or whatever random third-party plugins donā€™t run on DSP as well I know.

The purpose of this topic is to confirm if I can free up ASIO meter when I used those plugins that UA, Apogee or Antelope make. Most importantly this older thread (ASIO Meter (CPU)) brought me here

According to that topic (I also noticed lately) ASIO meter is not really showing your CPU or RAM processing information. When CPU performance is not really related to ASIO meter then how do you know that upgrading your CPU or RAM can secure more work space and processing power? And here comes my biggest question ā€œWhatā€™s the best way to secure/enhance more ASIO availability?ā€ and this question led me to search for DSP audio interfaces.

Also, since there is a confusing info on the internet that I can never confirm by myself and no one has ever clearly answered to.

Antelope once said in their official youtube video that using (their product like zen go) external DSP still consumes RAM on your machine. (Does that mean Antelope plugins with DSP still occupy ASIO availabilityā€¦? Then can I really expect Cubase performance improvement with DSP? )

What I really want to know is that if the same situation written above apply to all those DSP based plugins made by other companies. Will any of them actually improve my performance in terms of ASIO meter?

If you have an itch to buy HW DSP gear, please indulge yourself, but itā€™s not a panacea to reduce your ASIOmeter anxiety. Itā€™s more of a tourist trap to my opinion. I personally plead guilty of often buying gear I did not really neededā€¦ just donā€™t hurt yourself trying to rationalize it with an ASIOmeter readingā€¦ :wink::upside_down_face:

Few easy tricks to try before spending your money:

  1. Use a larger buffer size. Increase the size to its largest possible, until the latency becomes too perceivable to your taste. When ā€œunsureā€, give it a few hours of work to get accommodated, you will be amazed how much your brain can compensate for long latencies.

  2. Donā€™t use higher than 24b/96kHz resolution. Izotope tools will oversample it anyway in the few occasions the internal algorithms require it. In fact, using 16b/48kHz is (usually) unlikely to make any audible difference for most of what you will do. Give it a try. (or at least 24b/48k)

  3. Turn off ASIOmeter. Itā€™s highly unlikely you will regret doing so, ever.

Please share your observations, if you try those suggestions.

1 Like

Thanks for suggestions.

I already set my studio setup like the picture attached. (2048samples) image image

However I cannot go below 32 bit float. If I really wanted to, I need to go to pool and convert audio files info lower bit rate one by one but thatā€™s not something Iā€™d like to do.

Also, I donā€™t think I will ever disable ASIO meter because it tells me if I can load up other instruments before my Cubase crashes. Itā€™s very important to me.

Probably Iā€™m asking to much to my machine when Iā€™m too lazy to mixdown to 2mix and open it up on another mastering project to free up resources and process. I want to do everything in one project even itā€™s not ideal for most cases. I guess Iā€™ll just go with freezing midi or rendering in place if I really need to.

If you didnā€™t already do it, please run LatencyMon on your machine with every non-neccessary background processes disabled. If there are spikes, you might investigate which processes cause that. Like might be a dodgy network or video card driver. Check if there are any updates, especially for your mainboard. Remove any driver bloat like GeForce experience if ypu have that. Your music computer doesnā€™t want that at all.

I also nearly fell into the buy-more-gear trap (it is really easy to get into this this tbh), I also had spikes in LatencyMon on my system (on a brand-new Ryzen back then), but after I updated my BIOS, disabled everything not neccessary like onboard WiFi, internal sound chip and all those small things, it really added up in far better performance.

Your CPU isnā€™t too bad, really. Badly programmed plugins or CPU-hogs like certain Kontakt instruments can trigger the ASIO meter anyway, so the red icon for CPU overload lights up. Happens for me almost every time I load certain instruments into Kontakt, and I akready have two SSDs on my system (one for Windows, the other one dedicated to samples/VSTs). I also own NIā€™s Komplete, which contains a piano instrument using a particles engine. This thing puts quite some strain on the ASIO meter, like 20-30% if particles are enabled. Yet I have no issues, unless I use that dozens on times.

I do not care about the ASIO meter as I do not hear any audible clicks. First, things then are increasing the buffer size for my interface, slimming down resource consumption. I work with BBC Symphony Orchestra, where you can trash articulations not used for each track. At least frees up RAM and maybe a slight bit of CPU power. If you do orchestration with many tracks, you might check as well, if you can purge unused features on each track.

As said by many people, try to play around how you work. There is no general rule and it may be daunting to try out so many things. But going through the process will really help you, you might even discover new ways to improve your work and maybe even spark new musical ideas.

Have fun and produce music o/

1 Like

Generally speaking a faster CPU will of course process more data in the same given amount of time. This means that your CPU meter should drop for whatever processing is hitting that limit. More cores can help if your project is more parallel in nature where more threads are loaded . I posted a new thread with the computer-and-devices tag that includes two charts that compare DAW performance in Reaper with different CPUs. Iā€™ve been looking at those charts for a few years now and they seem to indicate that newer architectures generally offer better performance for us as DAW users.

So in other words your 8700K would most likely perform worse than a current generation CPU with the same amount of cores, and even with the same clock speed (because other improvements yield higher performance per clock).

The only problem with all of this is what you could upgrade to. Your platform is an older one so at this point itā€™s questionable if itā€™s worth getting a new CPU for your motherboard. It might be worth more to get a new CPU/memory/mobo instead.

Well, if you take an extreme example the answer is probably ā€˜yesā€™. iZotope plugins as far as Iā€™ve experienced can be pretty heavy on the CPU. So if you imagine that you have a bunch of those, and you instead choose equally ā€˜heavyā€™ DSP plugins running on a UAD-2 Octo setup, then I would absolutely imagine that you would save CPU cycles. Yes, some RAM would be taken up and some CPU cycles, but most of the work should be done on the DSP. Cubase and your computer should be concerned with hosting the GUI and shuffling audio back and forth between the DSP and the CPU, but that should be far less work than actually running heavy plugins on the CPU.

On the other hand if you donā€™t really offload ā€œenoughā€ to the DSP then of course you could - I imagine - see very little improvement.

But either way I think the idea here is that you should look at this with two questions in mind: 1) How much processing / how many plugins can I run on the DSP? and 2) How much processing / how many plugins can I then ā€œremoveā€ from the CPU?

I donā€™t think thereā€™s an absolutely clear generic answer. Youā€™ll have to look at how you work.

Just fyi: That setting doesnā€™t apply to stored files, it applies to the processing that happens as you mix. So files are read from your drives - either 16 or 24 bit fixed files or floating point files - and then theyā€™re processed using 32 or 64 bit floating point. So even if you converted all your stored files to 8-bit fixed point files they would still be processed at a minimum of 32-bit float. In other words keep it at 32-bit float and donā€™t worry about it.

My recommendation would be to also record/create new files at 24-bit fixed point. If samples etc. are loaded at 16 bit thatā€™s fine. No need to convert them.

1 Like