The No.1 Website for Pro Audio
 Search This Thread  Search This Forum  Search Reviews  Search Gear Database  Search Gear for sale  Search Gearslutz Go Advanced
CPU usage difference for 44.1 and 96 kHz Virtual Instrument Plugins
Old 2nd September 2011
  #1
Lives for gear
 
rushton's Avatar
 

CPU usage difference for 44.1 and 96 kHz

How much of a difference should there be? I have a 2011 2.2 mbp and with 5 exs24's, subboombass and predator playing with a few fabfilter pro-q's and timeless' treating them it's CPU bar is sitting around the halfway mark at 96 kHz. This doesn't seem quite right to me after having whole projects with heaps of plugs running with oversampling on and I'm barely using a quarter of the CPU.
Old 2nd September 2011
  #2
Gear Head
 
Toseben's Avatar
 

I don't know how much you mean with "heaps of plugs" but that seems pretty accurate to me. In 96kHz CPU processes over double the amount of data compared to 44.1kHz.
Old 2nd September 2011
  #3
Lives for gear
 
rushton's Avatar
 

Quote:
Originally Posted by Toseben View Post
I don't know how much you mean with "heaps of plugs" but that seems pretty accurate to me. In 96kHz CPU processes over double the amount of data compared to 44.1kHz.
By heaps I mean roughly 30 software instrument tracks averaging 2-3 effects plugs per track. Not all playing at once of course but up to 20 at once. That was barely 1/4 of the CPU bar.
I figured it would be about double but this just seams a bit extreme.
I am asking the question as I have had problems with random projects over using the CPU before.
Old 2nd September 2011
  #4
Lives for gear
 

Actually 96k hits FOUR times more.
Old 2nd September 2011
  #5
Lives for gear
 
rushton's Avatar
 

Quote:
Originally Posted by TimOBrienFlorida View Post
Actually 96k hits FOUR times more.
Does it really? Why is this? It would definitely explain what I am seeing.
Old 2nd September 2011
  #6
Lives for gear
 

A 96k file is between 3 and 4 times bigger than a 44.1k file.

Doubling the rate QUADRUPLES the data and stresses your cpu.

Lookup the math, then lock your daw on 24/44.1k and forgetaboutit....
Old 2nd September 2011
  #7
Lives for gear
 

There's some incorrect math in this thread.

I think there's a mixup between thinking about 44.1kHz-16bit vs 44.1kHz-24bit on the small side of the equation and 192kHz vs 96kHz on the big side.
Old 2nd September 2011
  #8
Gear Head
 
Toseben's Avatar
 

Quote:
Originally Posted by TimOBrienFlorida View Post
A 96k file is between 3 and 4 times bigger than a 44.1k file.

Doubling the rate QUADRUPLES the data and stresses your cpu.

Lookup the math, then lock your daw on 24/44.1k and forgetaboutit....
I'm not sure with my math so can you explain to me: Given that we have 24bit 44.1kHz and 24bit 96kHz files -> 96:44.1 = 2,18
Doesn't this mean that CPU has to use 2,2x the processing power compared to 44.1kHz?
Old 2nd September 2011
  #9
Quote:
Originally Posted by TimOBrienFlorida View Post
A 96k file is between 3 and 4 times bigger than a 44.1k file.

Doubling the rate QUADRUPLES the data and stresses your cpu.

Lookup the math, then lock your daw on 24/44.1k and forgetaboutit....
Quote:
Originally Posted by Jason West View Post
There's some incorrect math in this thread.

I think there's a mixup between thinking about 44.1kHz-16bit vs 44.1kHz-24bit on the small side of the equation and 192kHz vs 96kHz on the big side.
Bingo.

96/44.1 = ~2.18 the number of samples. Since the number of samples increases by ~2.18 times, you can ballpark the CPU bandwidth by using that proportion.

A 24 bit signal consumes half again (50% more) data bandwidth and storage than 16 bit -- since you're devoting 50% more raw data space.

Now that doesn't necessarily equate to more processor cycles required for the same data handling, but it does mean more data stored to and pulled from the hard drive and that can be a factor in overall performance.

(RAM handling can be harder to pin down, since a 32 bit computer can store higher resolution values by duplexing available memory; IOW, my 32 bit computer can still operate with numeric resolutions that require 64 bits of storage by, in a sense, doubling up on the number of memory addresses used to store a given number.)
Old 2nd September 2011
  #10
Quote:
Originally Posted by Toseben View Post
I'm not sure with my math so can you explain to me: Given that we have 24bit 44.1kHz and 24bit 96kHz files -> 96:44.1 = 2,18
Doesn't this mean that CPU has to use 2,2x the processing power compared to 44.1kHz?
Your CPU has a maximum data handling rate more or less based on the clock speed and the number of cores that can be properly utilized in a given operation (depending on your OS and version, your OS may not be able to utilize extra cores as well as other OS's/versions; older versions of OS X were particularly problematic in this regard).

The clock speed (the number of processor cycles per second) is basically the bottleneck. As long as you don't throw more operations at it than it can handle, everything is hunky dory. But if you exceed the number of operations it can perform in a given time, the system will need to resort to using the buffered info -- if the buffers aren't large enough to cover for the shortflow of data, a glitch or drop-out [a long glitch] will result.
Old 2nd September 2011
  #11
Gear Addict
 

To get on a more even playing feild, you have to increase the buffer size to compensate for running 96k. Keep in mind that running at higher rates gives you less than half the processing latency. It also cuts in half the latency of the converters themselves thus the slightly "less than" half.

The other thing to watch is using plug-ins with oversampling. If it is using 4x oversampling to hit 192k and you run the plug at 96k, you are now 4x96k=384. Some plug-ins will scale the oversampling down to compensate but most don't so you have set the oversampling lower to even it out.

Even after doubling your buffers, adjusting oversampling and changing to the higher rate, it works out that it is still considerably more processor intensive. On top of that, you get more stress on samples being streamed from disk as stated above. I usually get around 50% performance hit at 96k after making the above adjustments.
Old 2nd September 2011
  #12
Lives for gear
 
rushton's Avatar
 

Awesome, I'm really starting to understand how processors actually handle audio and generally work now. Thanks guys!
Post Reply

Welcome to the Gearslutz Pro Audio Community!

Registration benefits include:
  • The ability to reply to and create new discussions
  • Access to members-only giveaways & competitions
  • Interact with VIP industry experts in our guest Q&As
  • Access to members-only sub forum discussions
  • Access to members-only Chat Room
  • Get INSTANT ACCESS to the world's best private pro audio Classifieds for only USD $20/year
  • Promote your eBay auctions and Reverb.com listings for free
  • Remove this message!
You need an account to post a reply. Create a username and password below and an account will be created and your post entered.


 
 
Slide to join now Processing…
Thread Tools
Search this Thread
Search this Thread:

Advanced Search
Forum Jump
Forum Jump