VST CPU performance in Renoise vs Reaper

(this thread was split from this What's your approach to resampling? - #25 by Ninjilla, if you want better context, check my discussion with @TNT there)

Yeah I honestly have close to zero feature requests for Renoise. It serves me well for what it is.

My only complaint is the VST performance for “newer stuff”. Nothing else, really (because VST can fill the gaps for any features the DAW doesn’t have - e.g. granular synthesis, spectral processing, etc).

I’ve been using Renoise less and less because of the problems I’m having.
I’ll now try a workflow with Redux within Reaper and see how it goes (at least until new updates). I can’t work with piano rolls, my brain works better with trackers.

2 Likes

@untilde Would it be possible to compile a somewhat similar project in Renoise and Reaper to show and test the CPU differences, so we can test this here and look for problems? Reaper is pretty good with CPU scheduling, but “terrible performance in Renoise” vs. “barely touching 40% CPU in Reaper” sounds like there’s a bug or something else.

Ideally, only free plugins should be used for such a project, so that everyone can test it without buying a bunch of plugin licenses.

2 Likes

did a quick test, looping random note events across random vital presets in 5 instances of vital at the same time and the cpu meter in Renoise doesn’t go beyond 8.9% (AMD Ryzen 7 7700 8-Core Processor 3.80 GHz, ram = 32 gb). Perhaps it is a combo with other plugins that is causing problems on your end?

Also have you ever tinkered around with the settings here for plugins?

Having auto suspend on for certain plugins have caused problems for me in the past.

This is 10 Vital instances in Reaper with 14 % CPU

and this is the same in Renoise, where it is ~24-27%

grafik

Do I compare the right numbers?

Reaper_Surge.RPP - pCloud (it was Vital also here, not Surge like in the file name)

Over here your demo file fluctuates between 3.6% and occasionally hits 4.4% max for a short time. It isn’t stable though, it fluctuates.

This is strange. My CPU should be stronger than yours (AMD 9700X here).
How much samples did you use and what’s the buffer size?

48000 kHz 512 samples

Perhaps the graphics card? Amd Radeon 7900 GRE combined with fast ram?

What soundcard do you use / what kind of drivers (asio?) and buffer settings? Try increasing the latency and see how this affects cpu use.

Asio 48000 512 samples (10.7 ms)

Linux, Behringer UMC 204, Pipe-Wire, No graphic card. Hmmm…

Just a long shot but you could try update the BIOS

It’s on the list, but the company will do this that built the PC. They didn’t upgrade the BIOS, which should be normal imo

It’s hard to compare CPU displays accross DAWs. Renoise shows the overall percentage of the audio real-time thread only, where 100% means that audio definitely starts to crackle. Also it does not take CPU throttling into account.

So either compare the CPU values reported by task manager after minimizing all GUIs or, even more reliably, don’t look at the meter at all and compare when audio starts to crackle, which is basically the only thing that matters.

Also when comparing CPU loads, make sure you are using the same audio driver with the same audio sample rate and buffer settings. Buffer sizes and sample rates make a big difference here.


Here’s a test using 16 Vital instances with the “A happy ending of the world” preset (sorry, it was the first one :))

https://files.renoise.com/tmp/Rns_Reaper_16xVital.zip

On my working machine, a 12th Gen Intel i5-1240P with 16 CPU threads Renoise and Reaper can both play that. Starting the Song in Renoise crackles, because Renoise doen’t know yet where the plugins are played. On Reaper it sometimes crackles when looping.

When not looping and not starting playback, just playing through, reaper can handle a few more tracks, Renoise barely can. But there’s no drastic difference.

So the CPU baseline is similar, but it also would be weird if it isn’t, as this is just one VST playing on one core. Apart from scheduling the VSTis to some CPU the host’s don’t do much here.

I guess it gets really interesting when using send & groups tracks and also Track DSP FX. There surely will be differences.

2 Likes

I don’t have reaper to compare it with, but indeed on first playback run there are stutters in the beginning in Renoise, then after looping or subsequent playback the cpu usage stabelizes around 19 - 20 % (around the same cpu usage for renoise in the windows 11 taskbar). This is on 48000hz, 512 samples Asio buffer settings.

I think AMD have released several BIOS updates to their new CPUs as of late and probably more to come I suspect. If you need some assistance updating just ask, usually it’s just download the new bios and place in on a USB thumb drive, then go into your BIOS and find the update button somewhere :wink:

For Windows it should be important to get the latest chipset drivers directly from AMD, not your motherboard vendor, so I think you might have to be a little patient with the Linux drivers?

I’m also getting a new 9000 series CPU soon :badteethslayer:

Interesting test for my old i5-9600k with 6 cores, no hyperthreading.
Reaper can run all 16 instances without any issues but in Renoise the audio engine stops a few times before being able to run with massive stuttering both DAWs at 70% CPU usage. To remove all the stuttering Renoise can only run 8 instances and even then still give occasional stutter. Both Reaper and Renoise should run at at 44k, 512 samples on a Volt ASIO.

Maybe some setting is wrong here or could it be hyperthreading is important to Renoise and/or Vital? If hyperthreading is important then it’s gonna be interesting to test the upcoming Intel CPUs.

1 Like

Hey, I’m really happy with this thread. Thanks for taking your time.

A quick tip: Press Ctrl+Alt+P on Reaper to see a detailed view of CPU usage

I’ll check the provided examples in two different machines.

All background programs are closed and non-essential processes killed.

Samplerate is 44k, WASAPI shared mode. Windows 10 with latest updates.

ps: keep in mind that Vital is set at 2x oversampling in your example, which should affect testing slightly.

1. Intel i5 8400 (6/6) + 1x8gb RAM (pretty weak office PC in which I sometimes use Renoise)

Reaper:
image

Idle CPU usage is 20%.

Average CPU usage playing the whole loop from start to finish is 45%. Some slight crackling during the first couple seconds (when it peaks at 100%), but it gets stable after that.

Renoise:

Idle CPU is ~78%.

Intense crackling. With less than 1 second of playback Renoise overloads (set at 90%).

image

Then I disabled overload prevention and tested again:

Lots of crackling, unbearable.

NOW, this is where things get interesting (and curious).

I hit STOP twice to kill everything.

image

Then I hit play again from the start:

Weird, isn’t it? CPU peaks at about 85%. There’s some slight crackling, but nothing compared to what it sounds like when hitting PLAY immediately after loading the project. Lots of questions.
After hitting stop again (once), it idles at 20%!

image

In dumb-proof wording, what happens internally when we double click STOP?
How does Renoise behave when loading plugins at startup versus after double clicking STOP?
I’m very curious.

  1. AMD Ryzen 3600 (6/12) + 2x8gb RAM (I’ll test this later this evening at home)
CPU: 6-core 12th Gen Intel Core i5-12400F (-MCP-)
speed/min/max: 1857/800/4400 MHz Kernel: 6.11.0-1-rt7-MANJARO x86_64 Up: 7m
Mem: 1.5/15.38 GiB (9.7%) Storage: 1.62 TiB (57.4% used) Procs: 285
Shell: Bash inxi: 3.3.36

all thru jack and multicore support
Reaper 34~40 %
Renoise 36~38 %

As mentioned above, let’s please ignore the CPU display in Renoise and Reaper. It’s hard to compare this:

@untilde Also, how many realtime audio CPUs have you enabled in Renoise’s Audio preferences? What is the audio buffer size for WASAPI in Renoise and Reaper?

Double stop does a panic. Flushes all audio buffers and stops all plugins.

I’ll fix the CPU overload when starting the song after loading it. So let’s ignore that for now too because that demo song isn’t really a realistic example. This does happen, because in Renoise plugin instrument can move freely from any track to another one. So in order to find out where a plugin plays, it needs to play. This can be avoided by setting the Audio Routing for the plugin in Renoise from “Current Track” to some dedicated track.

1 Like

Also note Reaper’s “Anticipative FX Rendering” option.

This basically “freezes” tracks and preloads them automatically. It’s a great feature. Renoise doesn’t have anything like that. I think no other DAW has a similar feature.

That alone might be the reason why Reaper works so much better for you - CPU performance wise.

3 Likes

I can load ~36 instances in Renoise before it starts crackling. The CPU usage in HTOP is around 50% on each of the 16 cores.

512 samples / 48 000 kHz / Pipewire

With Reaper I can load more than 65 instances. CPU usage hits almost 100% on all cores, no xruns. I stopped at 65.

AMD Ryzen 9700x, 32 GB RAM

Try measuring CPU load with something like Sysinternals Process Explorer you’ll get more correct numbers

1 Like