KVR Audio

Christian Budde · Post by **Christian Budde** » Wed Jun 15, 2005 10:13 am

Hello everyone,

using my VST Plugin Analyser i tested the performance of several Compressors/Limiters. I don't say a word about how they sound, that decision is up to you...

(i have an AMD, 2GHz, but I think only the relative change is interesting)

L1V Limiter: 0.8%-0.9% (*)
Kjaerhus Audio classic Compressor: 0.9%
mdsp Compressor: 1.7%-2.5% (*)
E-Phonic X-Pressor: 0.4%-0.6% (*)
GY GComp: 1.5%
W1 Limiter: 1.1%-2.0% (*)
W1 Limiter (BetaBugsAudio): 1.7%-2.6% (*)
buzzroom GeneComp3: 0.9%-1.4% (*)
buzzroom GeneComp: 0.3%-0.5% (*)
buzzroom GranComp3: 0.2%-2.8% (*)!
buzzroom GranComp: 0.1%-1.7% (*)!
buzzroom buzMaxi: 0.4%-1.1% (*)
C3 Multiband Compressor: 4.1%-4.3% (*)
Slim slow Slider StereoCompressor: 0.8%-0.9% (*)
Simple Compressor: 3.9%
Simplesqueeze: 6.2%-7.3% (*)
Blockfish (filet files): 1.7%-1.9% (*)
BGTech Compressor: 1.2%-1.4% (*)
Sinus Peak Compressor: 0.5%-0.9% (*)
Kjaerhus Golden Compressor: 4.0%
Voxengo Polysquasher: 1.0%-1.1% (*)
Voxengo Elefant: 6.2%-6.4% (*)
4front XLimiter: 2.3%
4front YLimiter: 0.3%-0.7% (*)
Vanilla Compressor: 4.6%

(*) = input dependent (first value is processing with an impulse, the second processing with white noise)

Do you miss your favorite compressor? PM me, so that i can make a measurement.

And maybe someone would like to make a chart out of it...

Christian

P.S.: I just checked the demo versions, the real versions may perform different.

Ronny Pries · Post by **Ronny Pries** » Wed Jun 15, 2005 11:33 am

Good one, i'd like to check out how much cpu the UAD compressors/limiters actually use. Is the cpu usage easy to determine using your plugin analyser? If so, i'd like to contribute.

Perfect Solitude · Post by **Perfect Solitude** » Wed Jun 15, 2005 11:37 am

Check IHSAN ADL too you dolt.

kritikon · Post by **kritikon** » Wed Jun 15, 2005 2:24 pm

Holy guacamole

Voxengo Elefant: 6.2%-6.4% (*)

That's alot of CPU for a compressor - more than the Synthedit ones, I see.
Mind you GCO-1 isn't too far behind.

But...I guess it's worth it, as I notice that the real character comps are the ones that use the most CPU. I'd rather have a sweet compressor sound with a big hit rather than a bland one that barely taxes the hamsters.

Armadillo · Post by **Armadillo** » Wed Jun 15, 2005 2:31 pm

Elephant ain't no compressor you know

I'm missing the TbT stuff from that list though.

Ronny Pries · Post by **Ronny Pries** » Wed Jun 15, 2005 3:11 pm

Regarding Elephant, it'd be interesting to know wether oversampling was enabled or not.

kritikon · Post by **kritikon** » Wed Jun 15, 2005 4:19 pm

Elephant ain't no compressor you know

I thought Elephant was a limiter.
So, when did limiters cease being compressors?

Christian Budde · Post by **Christian Budde** » Wed Jun 15, 2005 4:30 pm

Performance Analysis is very easy with my plugin analyser. Just select the 'Performance Tab', edit the size to 44100 samples (that would be 1 second of audio @ 44100 Hz Samplerate) and click 'Run Performance Test'. You will get retrieve the time (in milliseconds), which was needed to calculate that effect (in realtime!).
The time may vary a little, but it's more or less stable.
If you assume, that processing of one second of audio shouldn't take longer than 1 second, then you can divide the Milliseconds by 10 and you retrieve the CPU-Usage. The absolute value is practically a little bit too low, because your host will produce overhead.

Btw. i used 3x oversampling for elephant. Without it's about 1.1%-1.2%...

Christian Budde · Post by **Christian Budde** » Wed Jun 15, 2005 4:40 pm

TbT Saturated Driver: 1.1%-1.6%
TbT Pocket Limiter: 3.2%-3.6% (stereo linked), 2.2%-2.6% (mono)

I can't get a demo of "IHSAN ADL" the website seems to be broken.

Armadillo · Post by **Armadillo** » Wed Jun 15, 2005 4:40 pm

kritikon wrote:
Elephant ain't no compressor you know
I thought Elephant was a limiter.
So, when did limiters cease being compressors?

so a motorbike is a bike as well then.
and a Zebra is a horse.

Shane Sanders · Post by **Shane Sanders** » Wed Jun 15, 2005 4:45 pm

Armadillo wrote:
kritikon wrote:
Elephant ain't no compressor you know
I thought Elephant was a limiter.
So, when did limiters cease being compressors?

so a motorbike is a bike as well then.
and a Zebra is a horse.

My zebras are limitless and ride bikes.

Aleksey Vaneev · Post by **Aleksey Vaneev** » Wed Jun 15, 2005 5:43 pm

Indeed, oversampling switch in Elephant affects CPU usage considerably (4x oversampling is like having 5 or 6 Elephant plug-ins running in parallel).

Aleksey Vaneev · Post by **Aleksey Vaneev** » Wed Jun 15, 2005 5:43 pm

Somehow I thought somebody is going to offer sonic performance measurements.

bduffy · Post by **bduffy** » Wed Jun 15, 2005 5:53 pm

Aleksey Vaneev wrote:Somehow I thought somebody is going to offer sonic performance measurements.

Like, "which one compresses better"?

Aleksey Vaneev · Post by **Aleksey Vaneev** » Wed Jun 15, 2005 7:20 pm

bduffy, yes. Of course, I doubt this is possible.

As for the CPU usage I think it could look more revealing to take 1% CPU as 100% performance and show others accordingly (i.e. 2% CPU means 50% processing efficiency).

Having information about oversampling and possibly about availability of variable (or soft) knee could help. From what I know variable knee requires more CPU resources to calculate.

Compressor/Limiter Performance Measurements