So finally we have plugins that run on 3 different popular platforms !

I had the chance to benchmark the plugins using all formats, Native , UAD-2 version and TDM version.
Of course the code might be different but it is VERY interesting to compare the plugin counts.

Stereo instances of the Brainworx Digital V2 plugin

30 TDM Plugins Pro Tools 8.1.1 HD2 PCIe Accel
80 UAD-2 Quad plugins
443 AU plugins with heavy duty settings. 1050 with flat EQ. Logic 9.1.5

Pro Tools ran out of RAM @ 750 RTAS with default - 330 RTAS with Heavy duty settings.

Here is the result with the TDM / UAD / Native versions of SPL Vitalizer.

17 TDM Stereo Vitalizers ( Each instance occupied 56 percent of one DSP so there is room for more plugins but no more Vitalizers ) HD2 PCIe.
56 UAD instances on a UAD-2 Quad.
390 Stereo AU´s in Logic.( with heavy crunchy settings )

Mac Pro Hexacore 3.46.

And yet, if all 1050 of those plugins were autotune, you still wouldn't be able to fix Matt Cardle.

What I've been saying for quite a while... And I didn't need to run any benchmarks to know. Those DSP chips are no competition to a modern CPU.

Alistair

Native is much more efficient...

Now somebody just needs to make native plugins that sound as good as Universal Audio's and there wouldn't be any questions at all

What audio were you running, how did the playback sound, were any other plugins or VSTs used in the session, yadda yadda ?

If you were comparing apples/apples then this would be a valid test. However, that isn't really the case.

First of all, I agree that TDM can be underwhelming in terms of DSP power. However, the one thing most everyone will agree on with TDM systems is that they have basically no perceivable latency. Now, of course there is some latency but it is much lower than native systems can typically run.

Second, a large number of the plugins for the UAD platform oversample. I can't say for certain whether the Brainworx one does or not, but most of the newer ones do and it definitely contributes to the quality of those plugins. I'm not here to argue whether oversampled plugins sound better or not, but they definitely do take quite a bit more dsp to run.

So, if you want to get close to doing an apples/apples test, you need to first run the Brainworx in Logic with the buffer set to 32 samples, which is the closest native buffer comparable to TDM latency, and compare that to TDM. I'm guessing you won't be able to load 1050 instances at a buffer of 32 samples.

And to compare to the UAD version, you would need to first change your audio settings to 192KHz in Logic to try and duplicate the oversampling that the UAD version of the plugin is doing. Again, I very much doubt you will get near the amount of instances.

I'm not at all arguing against native as I think in most cases native dsp will outperform both TDM and UAD. I'm just making the argument that your testing methodology is a bit flawed. Until you get a little closer to comparing apples/apples, I don't think your claim that native is 30x more powerful than TDM and 13X more powerful than UAD is valid.

Cheers

You know this for a fact? (Easy to test).

Alistair

I agree about the TDM version and needing to set the buffer as low as possible to get a fair comparison but not so with the UAD. The Plugin in question is as equivalent as 2 plugins can be on two different platforms so whether it does or doesn’t upsample is irrelevant because if it does it will on both platforms and if it doesn’t then that would allow the UAD to run more instances because it doesn't need to upsample.

But the buffer setting is an interesting though. Correct me if im wrong but really to be equivalent (leaving TDM out) you would have to set the buffer much higher to match what the inherent latency the UAD causes? I’m not sure what it is but it’s definitely much higher than 32 or 64 samples.

Interesting thread thanks for posting it!

That UAD-coded plugins oversample? Yes, this is a fact. Now whether they oversample at 4x, 8x, 16x, etc. I'm not sure and afaik that information is not readily available. The point being that if you were comparing a UAD plugin that oversampled at 4x to a native equivalent you would need to run the native host at 192KHz to try and simulate the same amount of DSP usage that you would get running the UAD version at 48KHz. Will you ever get apples/apples exactly? No, but this will get a lot closer than the original tester's methodology.

All of them? And importantly, does the Brainworx Digital V2 running on the UAD card oversample? That is the important question here.

Alistair

You buy UAD for all of their plugin's, which the majority of them are not available in TDM or AU without the UAD Cards. Don't loose sight of that.

Trying to imply that using PTHD is about plugin's is simply crazy. Next to no one today buys a TDM system because they can run more plugin's. I have an HD2 and I run next to no TDM plugin's, more than 90% of my plugin's are RTAS. Again, I don't use HD for Plugin's. Routing of Outboard, basically no Latency at any stage of the recording/mixing process, I know it will always work and never have to even think about my DAW ever, ever, ever, ever, ever. I was always fussing with Logic, Cubase, DP, LE, etc....Never, ever, never, ever, never, ever with PTHD. Get it?

This is absolutely not true. Let's say for instance that you are running both the UAD version and the native version in Logic running at 48KHz. If you load the UAD version, as that plugin upsamples, it will use more dsp than the native version as the native version does not upsample. If the UAD version is doing 4x upsampling, that would be the equivalent to running the native version at 192KHz.

You are confusing the UAD with TDM. If you are running UAD on a native system, it uses the same amount of sample buffer that the host is set to. So, if Logic is set to a buffer of 128, then the UAD plugins have a latency of 128 samples. Using UAD on a TDM is slightly more problematic as apparently UAD has issues with Pro Tools' delay compensation. That's a separate issue.

But my main point on the latency is this. If you are going to compare Pro Tools to Native, you have to take into account the latency. Pro Tools has lower latency than native systems, which means it is running at a buffer < 32 samples. So to do an equivalent comparison between TDM and native, one would have to run the host at a buffer of 32 (the lowest native equivalent) to try and simulate the same dsp usage.

If you are using 90% RTAS plugins, I find it very unlikely that you don't have latency. If you don't find that a problem, you might as well use HD|Native. It will give you much more power than a HD2 setup. :-)

Alistair

You're going to have to explain that one to me. Been doing this for several years. Now if you are talking about the need to track through plugins, that is the only time I use TDM plugs, once I'm done, they are changed to RTAS. The only plugins that remain in a mix are the one's I put on Aux. Ala, no latency.

Native will never give anyone more power, I'll always have the extras processing of the TDM cards in addition to what I get from the computer. I want to know the I can record a track at any stage of the process, and that seems promising with Native, but as of today, it is not guaranteed, at least with plugins on the track when I'm recording.

No, to my knowledge, not all UAD plugins upsample. Maybe someone with more specific information could chime in here, but afaik, only the developers would know this information.

Yes, exactly right. But until someone can confirm whether it does or not, or what ratio of oversampling (4x, 8x, etc.) then an apples/apples test again is really not possible.

Alistair, oh Alistair, lets go back to simple math class, I'm sure its the same in the Netherlands as it is here in the states.

Picture pretty school teacher at the chalkboard, now I know you're a young boy, so let me remind you to pay attention to the chalkboard and not the school teacher. Here we go:

2+2=4 (Computer + TDM Hybrid)

2+0=2 (Computer Alone)

Now in simple terms:

Native + TDM = More Than Native Alone

Now I think probably what got us here, is that lately you've been paying way too much attention to the school teacher and forgetting the fundamentals learned on the chalkboard.

Pump some white noise through any plugin and look at the output with an frequency analyser. Do you see a sharp fall close to Nyquist due to a low-pass filter for anti-aliasing purposes or not? If not, it doesn't oversample.

It is a digital EQ with no saturation. 99.999999999% chance it doesn't oversample. So I would say that unless the contrary is proven it is safe to assume it doesn't oversample.

Alistair

Guys, guys, this is foolish discussion. Until you can buy every single plugin sold by UAD and use them Natively, TDM, and their Cards, this argument is futile. It is not why people by the UAD cards, it's ONLY because you have to to use their plugins......Geez.....For supposed smart guys, sometimes you seem kinda dense.......

I get the math and how upsampling works but if I'm understanding you right your saying that the UAD version of the Brainworx EQ upsamples and the Native one doesn't? There are native plugins that upsample so this isn't a feature that only UAD has so if the native one doesn't upsample then i would assume the UAD wouldn't either.

Come on, come on, who's gonna take me on!!!!!!! Dying to get in the game coach, come on, let me go.............Jump......Jump......Jump

I don't think that's a very good assumption to make. Saturation, tube emulation, etc. has nothing to do with whether a developer decides to oversample with a plugin, whether it be a UAD plugin or a native plugin.

The only point I'm trying to make here is that the OP's testing methodology was flawed, therefore his claim that Native is 30x more powerful than TDM and 13x more powerful than UAD is also flawed.

Here's a list from UAD with all of their upsampled plugins, The Brainworx is not one of them.

You can dl the manual off their website or here's the link.
http://www.uaudio.com/media/assetlib.../uadmanual.pdf


With this info I think its fair to assume that comparing UAD to native version of this plugin is accurate.

Most of the native plugins I've used that upsample give you the option to turn upsampling on/off, such as Slate Digital,etc. The reason they do that is because it takes a hell of a lot more DSP to run the plugin with oversampling. So, it's very possible a developer like Brainworx would choose to upsample with the UAD version and not with the native version due to CPU usage concerns.

You are absolutely right though that it is entirely possible. I just wouldn't make the assumption.

UAD DSP vs Logic native performance? Oh, UAD anytime.

I'd rather cruise in a Audi than scream around in a Honda.

Thanks, A Fak. Good find.

My only question on that list is whether "Upsampled UAD Plugins with additional latencies" is just a list of the upsampled plugins with additional latency, or all the upsampled plugins.

Anyone know?

I can guarantee you it is not:
International Comparison of Math, Reading, and Science Skills Among 15-Year-Olds &mdash; Infoplease.com
World education rankings: which country does best at reading, maths and science? | News | guardian.co.uk
PAGE ONE -- U.S. Teens Rank Low in World Tests / High school students dismal in math, science - SFGate
Statistics on American K-12 Public Education - The Broad Foundation - Education
U.S. Teens Lag as China Soars on International Test - Bloomberg

Both a HD|Native rig and an HD2 will give you 192 voices (at 44.1/48Khz). The big difference is that the HD|Native rig will always have those 192 voices. When you run out of TDM power and start adding RTAS plugins, you lose voices (and increase latency)!

An RTAS on an aux uses two voices. One for the input, one for the output. (4 for stereo auxes). An RTAS on a track after a TDM plugin also uses two voices. (4 for stereo tracks). If you have consecutive RTAS plugins, the signal stays native in between the plugins so no extra voices are used. The same thing with an instrument track: If you put an RTAS plugin after an Instrument, the signal remains native until after the RTAS plugin and doesn't use any extra voices. On the other hand if you want to use an RTAS plugin then a TDM plugin and then an RTAS plugin again, you lose two voices for each RTAS (4 for stereo) because the signal has to be routed to and from the DSP cards twice.

So to resume: HD2+Native = Less than Native alone!

You need at least a HD3 rig to compete with a HD|Native rig. And for mixing and plugin power a HD|Native rig will compete with a HD6 rig.

Stop the condescending tone. You clearly don't understand the topic.

Alistair

I know the uad precision line started over sampling @ 192k many moons ago.

And that was for better quality even back then so I'm sure there newer stuff had quality processing in mind.

Good thread though,but does any know if the eq null with all 3 formats?

Sent from my PC36100 using Gearslutz.com App

So why don't Universal Audio have the decency to market it like that?

Alistair

For anything that has saturation or non-linear processing, oversampling can make sense. That isn't the case with this particular plugin.

Alistair

And yet, they still decided to oversample with precision plugins. It's been confirmed that the UAD Precision plugins do in fact oversample. Take a look at the list A Fak provided.

If the Precsion plugin line oversamples and does not use saturation, your argument that the Branworx wouldn't either is not very valid.