UAD vs.CPU comparison

calimike · Apr 26, 2015

Comparaison puissance CPU entre une UAD2 Octo et un i7 - forum Universal Audio UAD-2 Quad - Audiofanzine

Great test. An Octo has about one third the power of an i7 CPU, or almost half the power after CPU spikes. I'm surprised, I didn't realize it was that powerful.

The test was done with a notebook CPU, but since they basically stopped making faster desktop CPUs 5 years ago, it's not a big difference.

The test shows that if UAD was native, many UAD plugins would consume a lot more CPU than most native plugins. Which makes sense given how bad most native plugins sound.

DanButsu · Apr 26, 2015

You'd have to do the test at 192kHz as UAD upsamples. That will put significantly more stain on the notebook i7. I did not see any mention of sample rate, I am guessing it's at 44.1kHz

DanButsu · Apr 26, 2015

Native vs UAD VSM-3; only nulls at 192kHz based on this post:

http://uadforum.com/general-discuss...releases-vertigo-vsm-3-native.html#post164169

So the real test would be how many VSM-3s can you run natively on an i& chip at 199kHz. Would be fun to know

billybk1 · Apr 26, 2015

I really like is the sheer scalability of the UAD processing system. With UA raising the UAD/Apollo device limit to (6) per hardware system, if you were so inclined and as your needs grow, you could connect (2) Apollo QUAD's via TB and then add another (4) OCTO PCIe cards (in the box or in a TB chassis) and create a (40) SHARC DSP farm. That is equivalent to about (120) UAD-1 cards!!! I remember back in the day, when I had a (4) UAD-1 card system and I thought I was in high cotton!

Cheers,

Billy Buck

zvenx · Apr 26, 2015

DanButsu said:
Native vs UAD VSM-3; only nulls at 192kHz based on this post:

http://uadforum.com/general-discuss...releases-vertigo-vsm-3-native.html#post164169

So the real test would be how many VSM-3s can you run natively on an i& chip at 199kHz. Would be fun to know

HI, I was about to ask you where you got the 192 figure from, and if it was indeed the VSM-3 bug (confirmed by Dirk on GS on the UAD version of VSM-3), so I am not sure where you get the 192 upsample from.
https://www.gearslutz.com/board/10994588-post109.html

I go further from UAD instance chart:

"All measurements are taken at a sample rate of 44.1 kHz. Some UAD plug-ins use more DSP at higher sample rates."

which suggests to me, at least for some this isn't true.

My own tests had gotten this:
http://www.kvraudio.com/forum/viewtopic.php?p=5934328#p5934328

rsp

DanButsu · Apr 26, 2015

zvenx said:
"All measurements are taken at a sample rate of 44.1 kHz. Some UAD plug-ins use more DSP at higher sample rates."

which suggests to me, at least for some this isn't true.

rsp

The sample rate will affect the native versions, the UAD plugins that upsample do so on the SHARC chip, not the CPU.

I was under the impression that the VSM-3 upsampled. Any confirmations anyone?

Anyway, at 192kHz I can only run 5 native VSM-3s before I get a CPU overload!

And 4 UAD VSM-3s is the max on my Quad

Serenity · Apr 26, 2015

Should perhaps be noted that 16 instances of VSM3 will only use 80% of an Octo. So it actually has the power to run 20 of them, but it's not possible due to not being able to split instances across multiple DSP chips.

But yeah, there's no doubt that native plugs are more cost effective.

DanButsu · Apr 26, 2015

Okay, based on page 551 of the UAD Plug-Ins Manual v8 and Table 35 (Upsampled UAD plug-ins with additional latencies), there is no mention that the VSM-3 upsamples. So that's interesting

Kcatthedog · Apr 26, 2015

I was able to load more than 4 in console and they didn't say they were disabled, but if it consumes that much resource a good argument for native and ua

I have 2 instances in a 96@24 mix with lots of other ua plugs; no probs ?

DanButsu · Apr 26, 2015

I closed everything and only ran Pro Tools and redid the test:

Revised numbers at 192kHz I loaded and ran 72 Native VSM-3s on a 2011 i7 Mac mini Quad 2.0GHz. I got up to 76 at 44.1kHz

4 UAD instances hit 55% my Quad, but I could not load anymore as Serenity mentioned. We are limited to one VSM-3 per chip!

EDIT:
Funny... just added 8 in Console, then it maxed out...

Well, I'm more confused than anything now.

Ciao

billybk1 · Apr 26, 2015

DanButsu said:
Okay, based on page 551 of the UAD Plug-Ins Manual v8 and Table 35 (Upsampled UAD plug-ins with additional latencies), there is no mention that the VSM-3 upsamples. So that's interesting

The VSM-3 surely upsamples to 192k on the UAD! It is very easy to verify. As with native plugins, when you double the sample rate the processing needs double as well. A native plugin using 5% @ 44.1 will use 10% @ 88.2k. The UAD DSP works the same way. Non-upsampled UAD plug-ins will double DSP usage going from 44.1k to 88.2k. Upsampled DSP plug-ins will barely increase (typically only increase about 10%). The VSM-3 uses 40% DSP @ 44.1k and marginally increases to 44% @ 88.2k (very similar DSP load behavior to the UAD FATSO which is also upsampled). If it was not upsampled it should be using 80% DSP! Also, another clear indicator that a UAD plug-in is upsampled is it will have an added plug-in delay due to the upsampling. All upsampled UAD plug-ins have this added sample delay. Typically, it is 55 samples. Although a few of the UAD plug-ins incur less, some a bit more. The VSM-3 has an added 55 sample plug-in delay.

Cheers,

Billy Buck

Cass Anawaty · Apr 26, 2015

This operates on the assumption that they are sonically equal. Not all of us are on board with the premise, at least in the tests I ran a while back.

Serenity · Apr 26, 2015

Cass Anawaty said:
This operates on the assumption that they are sonically equal. Not all of us are on board with the premise, at least in the tests I ran a while back.

It remains to be seen when they release the update for UAD. Dirk said that a bug fix in the native version is the reason they didn't null.

slamthecrank · Apr 26, 2015

They don't null b/c of the bug in the recent UAD version - supposed to be fixed in an upcoming release.

Serenity · Apr 26, 2015

I checked how many instances I could run of VSC-2 reliably on my old MacBook Pro 2.3GHz i5. I got to 220. So an Octo would give me about 60% of the performance of my MBP for that specific plugin. A computer that cost me $1000 back then, and that's a full computer.

So yeah, UAD is certainly not cost effective. But then again, it might be worth it still.
People spend a lot of money on a Kemper only to use it in the studio. Something that could just as well run as a plugin on your computer instead of underpowered motorola chips. But the software is so good that people think it's worth it.

DanButsu · Apr 26, 2015

How many VSM-3s can you run? My i7 mini maxed out at 76!

I'll have to check out how many VSC-2s it can run.

Still love my uad plugs and the extra juice they give to my system and there's certainly 2 dozen I would not like to work without, just interesting to compare!

Eric Dahlberg · Apr 26, 2015

Serenity said:
So yeah, UAD is certainly not cost effective. But then again, it might be worth it still.

Well, for most of the plugins, it's the only option. In the case of the VSM-3, though, I wish I could sell the UAD version and get the native version. I love the sound but rarely use it because of the DSP usage.

Kcatthedog · Apr 27, 2015

need coffee very confused

are people saying they get way more BW instances ?

Hoenerbr · Apr 27, 2015

Wow 76 instances? Yeah but that must sound like a MESS!

DanButsu said:
How many VSM-3s can you run? My i7 mini maxed out at 76!

I'll have to check out how many VSC-2s it can run.

Still love my uad plugs and the extra juice they give to my system and there's certainly 2 dozen I would not like to work without, just interesting to compare!

Serenity · Apr 27, 2015

Eric Dahlberg said:
Well, for most of the plugins, it's the only option.

Yeah, it's certainly the only option if you want the UAD plugins. But the point is that you pay a lot for very little processing power.
For the price of two Octos you would be able to get a base Mac Pro or the most powerful Retina iMac, which has roughly three times the performance plus a lot more.

So yeah, it all comes down to how much the UAD plugins are worth to you. The native plugins are really good these days, certainly not up to UAD standard in every case, but still good enough for most people.

UAD vs.CPU comparison

Venerated Member

Administrator

Administrator

Shareholder

Established Member

Administrator

Hall of Fame Member

Administrator

Hall of Fame Member

Administrator

Shareholder

Shareholder

Hall of Fame Member

Hall of Fame Member

Hall of Fame Member

Administrator

Purveyor of musical dreams fullfilled.

Hall of Fame Member

Hall of Fame Member

Hall of Fame Member