Jump to content
JeffSouth

The X2 over sampling button

Recommended Posts

Hi guys.  I use a lot of Nebula 4 libraries for comps and such.  Can you guys confirm that the X2 over sampling button is really working?.  I keep it on, but recently I turned it off before loading an instance of Nebula and I simply am not seeing any noticeable improvement in latency.   The Nebula seems to struggle the same amount, yet turning the button off should be like going from 88 to 44 and this should be a massive cou saver, yet I don't see it.

 

But this is the same for other normal vst, like I have Kush Silika on the mix bus, it is CPU intensive, also IK tapes, turning the X2 button off just doesn't seem to make any difference at all on my CPU, yet if I actually begin a project at 88 instead of 44, the CPU power is significantly more stressed.

 

How can we confirm this X2 button is actually doing something?  When I turn that off it seems to me I should at least be able to notice a small difference in CPU stress.

 

Thanks.

Share this post


Link to post
Share on other sites

It isn't meant to be a latency reducer. It is meant to produce a more accurate sound from some effects or VSTI's because of aliasing effects. The amount of CPU power used would depend on what the effect is doing with the signal and how much of the CPU is used by that effect.

  • Like 1

Share this post


Link to post
Share on other sites
3 hours ago, reginaldStjohn said:

It isn't meant to be a latency reducer. It is meant to produce a more accurate sound from some effects or VSTI's because of aliasing effects. The amount of CPU power used would depend on what the effect is doing with the signal and how much of the CPU is used by that effect.

Yes, I realize it is to oversample because of aliasing, but running a ug, or especially several throughout a project is running all those plugs at 88 instead of 44.  Have 6 different plugs working at 2x and turn it off so they all drop to 44, this should be a very noticeable drop in pressure on your CPU.   

 

I am seeing, from what I can tell, zero effect from the X2 button.  And the more plugs you have going, say 12 at once, you drop those from 2x to just normal you should see a sharp drop in the CPU usage, at least a noticeable change.  I'm seeing what looks like zero difference, as if the X2 button isn't actually doing anything.

Share this post


Link to post
Share on other sites

It's not going to be easy to verify that upsampling is working. The effect is normally going to be quite subtle (assuming it does anything at all) unless you have some scenario where obvious aliasing can be heard (or seen with a spectrum analyzer). And that's actually a rare condition, difficult to make happen even on purpose. I've never had a synth or effect that caused audible aliasing, and if I ever did it would get retired immediately.

In terms of CPU usage, I doubt you'd see a discernable change, unless maybe you upsampled every plugin. Even then, CPU usage will normally bounce around more than that just due to WIndows background processes. By contrast, bumping up your overall sample rate is far more impactful, as it affects everything, including filling output buffers for monitoring and input buffers for recording as well as plugin processing.

Under most circumstances, upsampling individual plugins is not going to have a noticeable effect on overall latency. 

Of course, all this is moot if you can't hear (or even measure) the difference, in which case you didn't need upsampling to begin with.

 

 

  • Like 1

Share this post


Link to post
Share on other sites

The easiest way to verify plug-in oversampling is use a plug-in that reports sample rate like S-Gear 2.

Here is a video using CbB 2021.09 build 99 showing the effect of the 2x button when S-Gear 2 has oversampling with default values enabled in a 44.1K project.

3ObLEpR.gif

An audible test is possible inserting TTS-1 into a project and set the plug-in oversampling to 96K in AUD.ini. The plug-in will distort just like if the project is set to 96K.

  • Like 3

Share this post


Link to post
Share on other sites

I should not have mentioned aliasing at all.  I was only interested in the CPU usage of the X2 is.  So, then, if the X2 is isn't using but such little CPU, then I think Cakewalk should give us x4 and x6 options for plugins.  Thanks for showing the Shear example.  This shows it's working, and also tells me that this X2 is is very little pressure on even my old laptop CPU.  

 

So goodness sakes, Cakewalk, in a new update, give us at least a x4 option as well, it would be so awesome to use some of my old Waves plugs, for instance, with x4 or x8, stuff like Soundtoys and my Nebula stuff.

 

If this X2 is so darn easy to run, I can say for sure that if you gave us even x4 and this was mentioned on a forum like Gearspace (Gearslutz), there would be people jumping to Cakewalk like crazy because many folks still love a lot of plugs that do not have built in os.  

I have known people were interested in switching just because it has X2.  You add just x4 alone in to Cakewalk in the next update and you would see a nice jump in people turning to Cakewalk.  

CAKEWALK, GIVE US A X4 BUTTON AT LEAST IN THE NEXT UPDATE and that news would be all over the audio forums.  I would be one of the ones out there harping on it because I don't know of any other daw that can stay at 44 yet run it's plugs in oversampling with a simple button push.  You have to buy a extra plug to load them in.   

Wow, there are massive alliasing threads on Gearslutz and KVR, if we Cakewalk users were able to go in there and tell everyone the new free Cakewalk update allows you to run all plugs at x4 with a simple button push and the CPU hit is barely noticeable, good gracious, you would get a stream of new people jumping to Cakewalk.

A little arrow option with X2, x4, and x8, oh my, that would be HUGE 

Share this post


Link to post
Share on other sites
1 hour ago, JeffSouth said:

... if the X2 is isn't using but such little CPU, then I think Cakewalk should give us x4 and x6 options for plugins.

I'd caution against generalizing based on your observations. All you've determined is that your computer can handle the miniscule overhead of upsampling a specific compressor, in your specific project with your specific plugins.

This is far more about the plugin's efficiency than that of Cakewalk, since nearly all the extra processing that upsampling incurs happens within the plugin. It'll be running exactly the same instructions under Cakewalk as it would under Reaper or Studio One. Yes, being able to upsample individual plugins is indeed a great feature worthy of bragging about in the marketplace, but I doubt any knowledgeable DAW user is going to jump ship based on that feature alone.

1 hour ago, JeffSouth said:

I should not have mentioned aliasing at all. 

This discussion is meaningless without talking about aliasing. Aliasing is not only relevant, in the context of compression it's literally the only reason oversampling is ever prescribed. If you have no audible aliasing, you don't need oversampling to begin with. If you don't need 2x you likewise don't need 4x or 6x.

  • Like 2

Share this post


Link to post
Share on other sites
On 9/6/2021 at 2:45 PM, JeffSouth said:

I have known people were interested in switching just because it has X2.  You add just x4 alone in to Cakewalk in the next update and you would see a nice jump in people turning to Cakewalk. 

Send them the link to the documentation page scook posted.

"Benefit from 96kHz processing without the hard disk and CPU strain that recording at 96kHz produces"

Pretty tasty.

Having said that, I never oversample past 2X, because I can't hear any difference. There can be a difference with certain types of plug-ins, such as distortion and synths that are heavy on the square waves, but whether it's audible or not is also down to individual plug-ins. The thing people should be doing if they really care about this is doing some critical listening with the track solo'd and also doing null tests, where you compare bounces of the track with and without the oversampling enabled using analysis tools.

I think it's better to understand and know what's really happening with the tools than it is to just crank the oversampling up. Supposedly oversampling can also introduce issues of its own, but I don't know what they are. I know a good plug-in that cuts it output in half if Cakewalk oversamples it while running the 64-bit engine.

It's important to remember that you can get ugly artifacts by slamming the input of a plug-in with the output of the track/synth or previous processor. I was getting this recently. The best tool for finding this inter-sample clipping is the freeware analyzer Bitter.

If one wishes to really enter the oversampling wars, Meldaproduction plug-ins can set their internal oversampling as high as 512X, and have an option for higher quality oversampling. This is one of the "pro" features that gets added to the FreeFX Bundle plug-ins when you buy the upgrade license (wait for a 50% off everything sale):

image.thumb.png.471a8d1a04bf64fa0914a5c1cc058e70.png

As you can see, I keep it at 2X for rendering only, and turn off the high quality oversampling. The Meldaproduction documentation is uncharacteristically very informative about oversampling:

Quote

Oversampling can potentially improve sound quality by processing at a higher sample rate. Processors such as compressors, saturators, distortions etc., which employ nonlinear processing generate higher harmonics of the existing frequencies. If these frequencies exceed the Nyquist rate, which equals half of the sampling rate, they get mirrored back under the Nyquist rate. This is known as aliasing and is almost always considered an artifact. This is because the mirrored frequencies are no longer harmonic and sound as digital noise as this effect does not physically occur in nature. Oversampling (or oversampling) reduces the problem by temporarily increasing the sampling rate. This moves the Nyquist frequency which in turn, diminishes the level of the aliased harmonics. Note that the point of oversampling is not to remove harmonics, we usually add them intentionally to make the signal richer, but to reduce or attenuate the harmonics with frequencies so high, that they just cannot be represented within the sampling rate.

To understand aliasing, try this experiment: Set the sampling rate in your host to 44100 Hz. Open MOscillator and select a "rectangle" or "full saw" waveform. These simple waveforms have lots of harmonics and without oversampling even they become highly aliased. Now select 16x oversampling and listen to the difference. If you again select 1x oversampling, you can hear that the audio signal gets extensively "dirty". If you use an analyzer (MAnalyzer or MEqualizer for example), you will clearly see how, without oversampling, the plugin generates lots of inharmonic frequencies, some of them which are even below the fundamental frequency. Here is another, very extreme example to demonstrate the result of aliasing. Choose a "sine" shape and activate 16x oversampling. Now use a distortion or some saturation to process the signal. It is very probable that you will be able to hear (or at least see in the analyzer) the aliased frequencies.

The plugin implements a high-quality oversampling algorithm, which essentially works like this: First the audio material is upsampled to a higher sampling rate using a very complicated filter. It is then processed by the plugin. Further filtering is performed in order to remove any frequencies above the Nyquist rate to prevent aliasing from occurring, and then the audio gets downsampled to the original sampling rate.

Oversampling also has several disadvantages of which you should be aware before you start using it. Firstly, upsampled processing induces latency (at least in high-quality mode, although you can select low-quality mode in the plugin settings), which is not very usable in real time applications. Secondly, oversampling also takes much more CPU power, due to both the processing being performed at a higher sampling rate (for 16x oversampling at 44100 Hz, this equates to 706 kHz!), and the complex filtering. Finally, and most importantly, oversampling creates some artifacts of its own and for some algorithms processing at higher sampling rates can actually lower the audio quality, or at least change the sound character. Your ears should always be the final judge.

As always, use this feature ONLY if you can actually hear the difference. It is a common misconception that oversampling is a miraculous cure all that makes your audio sound better. That is absolutely not the case. Ideally, you should work in a higher sampling rate (96kHz is almost always enough), while limiting the use of oversampling to some heavily distorting processors.

Also keep in mind that many other plug-ins also have their own internal oversampling settings, which when combined with Cakewalk's, can take the frequencies out to freakin' ultraviolet.

Share this post


Link to post
Share on other sites

It's worth re-iterating that some plugins don't support up-sampling, particularly bridged 32 bit ones.

Also some plugins might sound different - for example the rate / depth in TH3/THU's chorus pedal goes at half the speed when running at either X2 or when my audio i/f is set to 96KHz.  I must get around to reporting this to Overloud.

 

  • Like 1

Share this post


Link to post
Share on other sites
12 hours ago, msmcleod said:

It's worth re-iterating that some plugins don't support up-sampling, particularly bridged 32 bit ones.

Also some plugins might sound different - for example the rate / depth in TH3/THU's chorus pedal goes at half the speed when running at either X2 or when my audio i/f is set to 96KHz.  I must get around to reporting this to Overloud.

 

Yeah, this is strange. In fact there quite a few popular VST 3 plugins that act this way. 

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×
×
  • Create New...