Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
Mixbus takes 200+ CPU
01-10-2018, 04:43 PM
Post: #1
Mixbus takes 200+ CPU
Running top, Mixbus32C (latest) takes what seems to be a lot of CPU. DSP is at 77% with the transport stopped. jackd is running using:

/usr/bin/jackd --sync -P80 -ndefault -dalsa -dhw:M1010LT -r44100 -p256 -n4

CPU is: i5-3570 CPU @ 3.40GHz
RAM: 16 GB

For what I'm doing at the moment, punch in/outs, it seems OK though.

   

current: Soca Philosophy
Elixir Collage-1
Find all posts by this user
Quote this message in a reply
01-11-2018, 03:03 AM
Post: #2
RE: Mixbus takes 200+ CPU
(01-10-2018 04:43 PM)jonetsu Wrote:  Running top, Mixbus32C (latest) takes what seems to be a lot of CPU. DSP is at 77% with the transport stopped. jackd is running using:

/usr/bin/jackd --sync -P80 -ndefault -dalsa -dhw:M1010LT -r44100 -p256 -n4

Can you repeat the experiment with jack running on the onboard sound? I had 2x 1010LT in an older setup and it seems they (or their Linux driver) are a bit sloppy with IRQ/CPU resources.

Cheers,
MMM
Visit this user's website Find all posts by this user
Quote this message in a reply
01-12-2018, 09:55 AM
Post: #3
RE: Mixbus takes 200+ CPU
I've asked Paul Davis, and Robin contributed also.

Paul:

"There's almost no reason whatsoever to measure CPU usage for realtime audio usage.

There's only number that matters: the audio hardware gives you N msec to process T samples - how close are you to missing that deadline? i.e. if the hardware gives you 8msec and your system uses 7msec, then any small disturbance to how long it takes will lead to a dropout ("xrun"). If your system uses 0.5msec, then there's so much headroom that you can probably watch a couple of youtube videos and rebuild ardour without worrying about
dropouts. This measure ("DSP Load") is what Ardour and JACK tell you, instead of some system-based "CPU use" number. If it says "50%" it means your systems uses half the available time; if it says 1%, it means it uses 1/100th of the available time. On all 3 major OS platforms, xruns start fairly reliably when this number gets up to about 80%.

Measuring this in terms of "percentage of theoretical CPU cycles actually used for computing" really misses the point entirely."

Robin:

" CPU load can go up to N * 100% with N = number of CPU-cores.

200% CPU usage (as displayed by top) means on average two of your CPU cores are busy. On a quad-core system that's nothing to worry about. It shows that Mixbus' parallel-processing works as designed.

Total load depends on plugins and track-count (in Mixbus every track does DSP)."

current: Soca Philosophy
Elixir Collage-1
Find all posts by this user
Quote this message in a reply
01-12-2018, 05:25 PM
Post: #4
RE: Mixbus takes 200+ CPU
This is correct

" CPU load can go up to N * 100% with N = number of CPU-cores.

to go a bit further. I have seen in other applications (non audio related) ,that probably use other parallel processing (openmp-- I dont know the type of parallel processing used in mixbus) that in quad core machines with hyper-threading on, cpu hits 800% on full load.




(01-12-2018 09:55 AM)jonetsu Wrote:  I've asked Paul Davis, and Robin contributed also.

Paul:

"There's almost no reason whatsoever to measure CPU usage for realtime audio usage.

There's only number that matters: the audio hardware gives you N msec to process T samples - how close are you to missing that deadline? i.e. if the hardware gives you 8msec and your system uses 7msec, then any small disturbance to how long it takes will lead to a dropout ("xrun"). If your system uses 0.5msec, then there's so much headroom that you can probably watch a couple of youtube videos and rebuild ardour without worrying about
dropouts. This measure ("DSP Load") is what Ardour and JACK tell you, instead of some system-based "CPU use" number. If it says "50%" it means your systems uses half the available time; if it says 1%, it means it uses 1/100th of the available time. On all 3 major OS platforms, xruns start fairly reliably when this number gets up to about 80%.

Measuring this in terms of "percentage of theoretical CPU cycles actually used for computing" really misses the point entirely."

Robin:

" CPU load can go up to N * 100% with N = number of CPU-cores.

200% CPU usage (as displayed by top) means on average two of your CPU cores are busy. On a quad-core system that's nothing to worry about. It shows that Mixbus' parallel-processing works as designed.

Total load depends on plugins and track-count (in Mixbus every track does DSP)."
Find all posts by this user
Quote this message in a reply
01-13-2018, 05:28 AM (This post was last modified: 01-13-2018 06:25 AM by Frank.)
Post: #5
RE: Mixbus takes 200+ CPU
To monitor my systems i use Ksysguard

From the manual:

System Monitor is the Task and Performance Monitor.
It features a client/server architecture that allows monitoring of local as well as remote hosts. The graphical front end uses so-called sensors to retrieve the information it displays. A sensor can return simple values or more complex information like tables. For each type of information, one or more displays are provided. Displays are organized in worksheets that can be saved and loaded independently from each other. So, System Monitor is not only a simple task manager but also a very powerful tool to control large server farms.

One can graphs of CPU usage per CPU, context switches.

Also disk performance: and network load

temperatures and values to choose CPU !
   

CPU performance graph
   


In Windows there is a similar tool: least regarding the CPU performance.
I noticed that Windows had a CPU sleep mode. Which is desastreus for RT performanc: to wake up a cpu takes lots of cycles.

One can remove that mode in the registry. look for CPU sleep. I think
I have no windows anymore : and i checked for my notes , but the gang who took over my job destroyed everything befor i could make a copy.

regards


Attached File(s) Thumbnail(s)
   

Frank W. Kooistra

- Mixbus 4.3 , MixbusMC32 4.3, AD/DA: Focusrite Liquid56, Saffire 26 I/0, Saffire Pro 24 DSP, UbuntuStudio 17.10, OSX 10.11.6, No Harrison CP-1 Control Pad, TTC-1 Touchpad, Lemur
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)