Sonar and UAD-2 review

Hi All
Just wanted to post quickly about my findings so far with a new UAD-2 running under Sonar. These have been getting a fair amount of attention lately and so I thought I'd post my findings so far with mine.

First, my system:

Intel Core 2 Duo E8500 Wolfdale 3.16GHz
EVGA nForce 780i motherboard
4 GB RAM
NVIDIA GeForce 8600 GT
MOTU 896HD (BLA Modded)
2 UAD-1 cards
1 UAD-2 duo

First of all, the card went in fine and driver installation went without a hitch. UAD-1 cards still work with no changes in performace that I've been able to see. The significantly reduced latency with the new UAD-2 card is a really nice thing. I can track just fine with UAD plugs enabled. While I use a separate audio path for monitoring and thus don't generally monitor direct from Sonar when recording audio, I do a fair amount of work with virtual instruments and being able to have a reverb bus with a plate 140 available when tracking is awfully nice.

Next, host CPU usage. If you've been following the chatter on the UAD forum you may know that there has been a fair amount of discussion about increased host cpu usage with the UAD-2. Here are my findings under Sonar.

First, I created a very basic audio track with a brief snippet of piano on it and then added an 1176. Then I cloned it 23 times (for a total of 24 tracks). First of all, 24 1176 instances uses 69% of my UAD-2 duo. The bad news is that when I press play my CPU usage spikes up and if it plays at all it is massively distorted (usually I just get a dropout almost immediately). Disabling the UAD card results in normal (minimal) CPU usage and normal playing, re-enabling the card results distortion. This happens with ASIO drivers with the sample buffer set to 128.



Shifting to 1028 ASIO buffers results in normal playback, but curiously, anything below that gives me massive CPU spikes and audio distortion.

Oddly, if I disable the multi-processor engine and play the above project at a buffer setting of 128, the project plays fine with somewhat elevated but still usable CPU usage.




Somehow, the multi-processor engine is seemingly related to the increased host CPU usage.

One more comparison. The same project with each 1176 plug replaced with Voxengo's Marquis Comp (24 tracks, 24 plugs), I see the following usage:



And finally with no plugs (just for a baseline):



Hope that helps folks.

Dean

Hi Spindle,
I didn't mean to imply that the Marquis was similar in sound to the 1176....I think you may be able to cause the Marquis to sound somewhere close-ish to the 1176, but I wouldn't say that they are terribly similar. I used the Marquis in my comparison simply because it's my most frequently used non-UAD compressor...really just because it was a compressor, it wasn't UAD and it was at the top of my list (although I do like it quite a bit).

Dean

ORIGINAL: Razorwit

One more comparison. The same project with each 1176 plug replaced with Voxengo's Marquis Comp (24 tracks, 24 plugs), I see the following usage:



And finally with no plugs (just for a baseline):



Hope that helps folks.

Dean

And this is pretty good data point that UA could take it's plugs native. A compressor is a compressor...I think it would be hard to say that the math involved is so much more complicated on the UAD 1176 vs the Voxengo compressor that it would be so much more CPU usage to take the 1176 native.

Dean...thanX much for posting.

Frank, do you still notice this if you take the Sonar performance check out of the box in the UAD control panel? That thing doesn't work at all for me no matter what version of their software I use. I get pops, cracks, dropouts in UAD plugs, automation issues, and all this stuff prints to the export when I export my audio. As soon as I take that check out of the box, everything works perfectly.

Danny, actually I have it disabled..
I have issues either way.. but they are different depending on how the Sonar-compatibility-mode is checked..
if I run the cards at ~60% dsp power everything's working ok,..
anyway, seems as if I need to switch those settings constantly to whatever I am doing in Sonar.

Well ...... if anyone had any doubt about how "ancient" the UAD2 technology already is, the above says it all.

UAD 2 DUO -> 24 x 1176's = ~%70 load

That gives "theoretical" max's as follows - assuming constant steady %100 glitch free use at %100 which you can see from the above you aint going to get:-

- UAD 2 SOLO - 17 x 1176's = ~%100 load

- UAD 2 DUO - 34 x 1176's = ~%100 load

- UAD 2 QUAD - 68 x 1176's = ~%100 load

Remember - as crappy as these figures are, this is a just a COMP - imagine how bad the figures would be if you were loading the NEVE stuff or some of the better Reverbs !

And latency was set to 128samples - 2.9ms - thats OK, but you wouldnt want to go any higher if live input monitoring / tracking is what you want because if you are st to 2.9ms / 128 smaples, your overall "real world" latency taking into account AD/DA conversion, system processing and peripherals etc... will be at least twice the audio config setting - ie: > 6.0ms.

And to further prove why UAD dont post "native" figures, Voxengo's Marquis Compgo - by all reports a brilliant VST thats just as as good as anything UAD in its own ways, uses somthing like 1/3 rd the CPU load in Sonar with no "dongle" UAD 2 card.

Seriously, if you like UAD stuff and want to pay for the out of date UAD 2 technology, more power to you - but saying UAD plugs sound "better" as oppossed to sound "different" to other high end plugs is like saying the lead guitar track I recorded in ProTools last night sounds better than the same lead guitar track recorded in Sonar 7 - its rubbish.

I just want to give major props to Razorwit for posting his genuine results - good on you- I have no doubt as people see more and more posts like this, this whole UAD 1 / UAD 2 myth will finally be busted and then if people want to buy and use them, they can without thinking or beleiving or being "spun" or "hyped" into the illusion that they have the "best and most powerfull" DAW plugs out there.

Tim

Dude - get over it. If you don't like them, don't buy them. I buy them cause I like what they do. I do music, I don't sit around playing with the computer all day long. If they do the job, great, if they don't... get on with it. This is about making music and they are just another tool to do so.

ORIGINAL: timboe

Well ...... if anyone had any doubt about how "ancient" the UAD2 technology already is, the above says it all.

UAD 2 DUO -> 24 x 1176's = ~%70 load

That gives "theoretical" max's as follows - assuming constant steady %100 glitch free use at %100 which you can see from the above you aint going to get:-

- UAD 2 SOLO - 17 x 1176's = ~%100 load

- UAD 2 DUO - 34 x 1176's = ~%100 load

- UAD 2 QUAD - 68 x 1176's = ~%100 load

Remember - as crappy as these figures are, this is a just a COMP - imagine how bad the figures would be if you were loading the NEVE stuff or some of the better Reverbs !

And latency was set to 128samples - 2.9ms - thats OK, but you wouldnt want to go any higher if live input monitoring / tracking is what you want because if you are st to 2.9ms / 128 smaples, your overall "real world" latency taking into account AD/DA conversion, system processing and peripherals etc... will be at least twice the audio config setting - ie: > 6.0ms.

And to further prove why UAD dont post "native" figures, Voxengo's Marquis Compgo - by all reports a brilliant VST thats just as as good as anything UAD in its own ways, uses somthing like 1/3 rd the CPU load in Sonar with no "dongle" UAD 2 card.

Seriously, if you like UAD stuff and want to pay for the out of date UAD 2 technology, more power to you - but saying UAD plugs sound "better" as oppossed to sound "different" to other high end plugs is like saying the lead guitar track I recorded in ProTools last night sounds better than the same lead guitar track recorded in Sonar 7 - its rubbish.

I just want to give major props to Razorwit for posting his genuine results - good on you- I have no doubt as people see more and more posts like this, this whole UAD 1 / UAD 2 myth will finally be busted and then if people want to buy and use them, they can without thinking or beleiving or being "spun" or "hyped" into the illusion that they have the "best and most powerfull" DAW plugs out there.

Tim

ORIGINAL: tazman

Dude - get over it. If you don't like them, don't buy them. I buy them cause I like what they do. I do music, I don't sit around playing with the computer all day long. If they do the job, great, if they don't... get on with it. This is about making music and they are just another tool to do so.

Roberto, this is an unfair and emotional assessment. Tim is providing excellent information so people can make an *informed* choice regarding UAD cards. By objective standards(and now empirical evidence) the cards are definitely underpowered compared to both the UAD1(only 2.5x increase in 8 years?) and current technologies. To say anything else is simply a religious statement. Lest I be branded a "hater"(thank you for not being like some others in this regard)...I am about to order my own UAD2 to see what all the fuss is about.

P.S. Can you go over to my other thread and tell me what your favorite plugs are so i can spend my vouchers?

I don't agree that it helps people make an informed decision. Everyone's needs are different. Plug-ins are bought for their sounds (or what they do to sounds), not for the power of the cards they run on. I would still have a UAD-1 even if it could only run one instance of the LA2A or 1176LN. The real things cost between $1800 and $3000. The card is much cheaper and sounds really good.

Are the UAD-2's underpowered? I don't think so, but to be honest, I don't care!!! The cost of a UAD-2 quad (which I will be getting soon) is ~4 times the UAD-1, but it fosters 10x the power. That, to me, is a bargain.

Anyway, the plugs I use all the time are the Neve 1073, 1176LN, LA2A, LA3A, the Pultec's, Fairchild 670 and the 88RS.

If I could only choose one... Fairchild 670 for what it does on the Man Buss. The LA3A is getting to be my go to compressor, but I do like what the LA2A does on bass guitar.

The fact is UAD takes a load off your CPU, no matter how big or small it may be..
Why people would complain a product that reduces the load on your CPU as with UAD, against having a product that drains your CPU as with any other plugin..
Is beyond me!!

ORIGINAL: plainfaced

The fact is UAD takes a load off your CPU, no matter how big or small it may be..
Why people would complain a product that reduces the load on your CPU as with UAD, against having a product that drains your CPU as with any other plugin..
Is beyond me!!

Look at the CPU meter in the first picture. The PC is nearly maxed out. There is no decrease in load. On the contrary. Anyway, as it doesn't run properly this way, it isn't even worth considering.

The alternative scenario is that it forces you to turn off multi-processor mode so it INCREASES the total load on your system! Look at the CPU graph in the second picture. There is only about 50% of one core left for native plugins. In other words, about 75% of the total system's processing power has been chewed up by the UAD card. There is no way anyone can translate this into "takes a load off your CPU".

Although the Marquis isn't a direct comparison, look at the CPU graphs in that test. There is much less load than any of the tested modes using the UAD cards. The PC can still do plenty of native processing.

Hopefully UA will address these issues.

Also there is the price consideration. At 1500$ the Quad is less powerful than the equivalent cash invested directly in a DAW. These are all important factors when purchasing expensive equipment like the UAD cards.

UnderTow

ORIGINAL: plainfaced
Yeah sorry I was just on a rant, basing my opinion on the old UAD-1 argument..

After re reading the post.. quite concerning.

Not totally related. Im on UAD-1, and downloaded the most recent software.. It seems to have a few problems - like motorboating plugs..
I havent tested it properly yet..

No worries. Hopefully these are just issues with the early versions of the drivers.

UnderTow

the test in this link had somewhat better results in both reaper and sonar

http://www.chrismilne.com/uadforums/viewtopic.php?f=11&t=10024

Just an update. After reading over some of Billy BUcks excellent work over at the UAD forums I decided to try my testing again using WDM drivers. Here are my results:

24 1176 plugs at 2.7ms latency:



24 Marquis Comp at 2.7ms



24 1176 at 20ms



24 Marquis at 20ms




All audio played fine with no distortion.

Looks like the problems I was having under ASIO with multiprocessing at low latencies went away after shifting to WDM drivers. I shifted back to ASIO to see if the problems came back and I can reliably reproduce them. Another interesting note is that at low latency the UAD plugs used more host CPU than the Voxengo, but at higher latencies that was not true.


Dean

UADs have always been a bit finicky it seems. Out of curiosity, I tried 18 tracks of 1176LN, but on a UAD-2 Solo. I've got a quad core on Vista 64 with 6GB RAM.
The highest spikes were around 22% in Sonar (multiprocessing enabled). Strangly, when I disable the UAD-2 there is no appreciable difference in CPU usage. This is at 256 samples latency on an Echo AudioFire8. UAD-2 DSP is at 95%; PGM at 10% and MEM at 0%.

I also have 2 UAD-1e's isntalled, but they are not being used. For fun, I assigned the Plate 140 to the UAD-1s, and threw a few of those in the mix, in addition to all the 1176LNs, to see if the UAD-1 usage affected the UAD-2. There was no effect. So at this point, for me everything is better than good!

I hope you're able to work out these issues.