Computer Audio Asylum

Music servers and other computer based digital audio technologies.

Return to Computer Audio Asylum


Message Sort: Post Order or Asylum Reverse Threaded

Devoting an entire core to audio

76.206.10.165

Posted on December 15, 2016 at 11:46:42
scruffy_
Audiophile

Posts: 1576
Location: Los Angeles
Joined: August 5, 2005
Why hasn't anybody developed a software that dedicates an entire core of the CPU to a program (or a specified USB port) such as an audio player? Is it possible?

~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Humility is the true mark of genius. Just get used to it."
-Anonymous

 

Hide full thread outline!
    ...
RE: Devoting an entire core to audio, posted on December 15, 2016 at 12:21:06
AbeCollins
Audiophile

Posts: 28648
Location: USA
Joined: June 22, 2001
Contributor
  Since:
February 2, 2002

Processor affinity is not new and many operating systems can accomplish this at the user level. In other words, you can do this yourself. Programmers can also assign specific cores or CPU's in their applications. However, you have to understand the application in order to determine if there's any real benefit vs letting the OS scheduler have complete control over where each thread resides.

Look up "processor affinity". Each operating system has their own specific commands and/or APIs to manage affinity (or binding). taskset is common in Linux systems, cpuset in BSD, pset_bind in Solaris, etc.



 

RE: Devoting an entire core to audio, posted on December 15, 2016 at 15:04:24
scruffy_
Audiophile

Posts: 1576
Location: Los Angeles
Joined: August 5, 2005
Thanks! I can now see how to set the affinities in Windows 10, but would be very cumbersome because you would need set the affinity of foobar to one thread and then set all the other apps and services to do the inverse. And then you would need to redo it whenever to close and reopen an app or service.

~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Humility is the true mark of genius. Just get used to it."
-Anonymous

 

Process Lasso Pro, posted on December 15, 2016 at 17:12:44
Kal Rubinson
Reviewer

Posts: 10647
Location: New York
Joined: June 5, 2002
Useful tools for controlling your facilities.

 

RE: Devoting an entire core to audio, posted on December 16, 2016 at 01:55:01
Mercman
Reviewer

Posts: 6278
Joined: October 20, 2002
Contributor
  Since:
May 20, 2004
Fidelizer Pro claims to move non-audio processes to their own core.

 

RE: Devoting an entire core to audio, posted on December 16, 2016 at 02:35:24
soundchekk
Audiophile

Posts: 1906
Joined: July 11, 2007
Process affinity tweaks are known, discussed and applied for almost a decade over here.

The main challenge is not to assign a process or interrupt to a certain CPU.
That you can simply do with the taskmanager.
The actual challenge is to move all other (dynamic) processes away from that CPU.

An example for affinity:

http://www.windowscentral.com/assign-specific-processor-cores-apps-windows-10

Of course you could also start a program from cmdline with an affinity flag assigned.

###

Linux is IMO a bit more advanced in that area. You can "isolate" an entire CPU first. (Don't know if you can do that under Windows)
This way no process will be assigned to that CPU anymore.
And than you can manually assign a process to that isolated CPU exclusively. (you could also do that in a/the process startup script to make the assignment automatic of course)



Enjoy.

 

RE: Devoting an entire core to audio, posted on December 16, 2016 at 05:20:07
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
The ability is built into windows taskmanager. You can even see a graphical representation.

Depending on how 'audio dedicated' the PC is, this may or may not 'sound' better and is a trial and error thing.

 

The whole thing , posted on December 16, 2016 at 05:22:43
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
was on the Chinese Tam Audio site a few years ago, with scripts on how to do it. Some inmates were then loudly proclaiming the Apple route to 'best' audio.

 

RE: Devoting an entire core to audio, posted on December 16, 2016 at 05:25:13
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
It is not cumbersome at all and you can write a script to do it if you want.

However, there is transparency built into using Taskmgr, as opposed to using opaque software that claims to do whatever to improve 'timing'.

 

RE: Process Lasso Pro, posted on December 16, 2016 at 05:26:23
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
This is not needed, if one wants to have clarity about what is happening to processes and priorities.

 

RE: Devoting an entire core to audio, posted on December 17, 2016 at 08:26:23
scruffy_
Audiophile

Posts: 1576
Location: Los Angeles
Joined: August 5, 2005
Thanks! I can easily change the affinity of foobar to a specific thread, but I cannot see where I can easily change my other apps and services to not use that particular thread. Moreover, all the changes get reset when rebooting the computer or reopening any app or service.

Maybe I'm missing something?

~~~~~~~~~~~~~~~~~~~~~~~~~~~
"Humility is the true mark of genius. Just get used to it."
-Anonymous

 

RE: Devoting an entire core to audio, posted on December 17, 2016 at 14:17:53
SBGK
Audiophile

Posts: 444
Joined: March 22, 2012
think you've got threads and cores mixed up

just have a look at process lasso, it's got everything you're asking for

fideliser also has similar functionality
http://mqnplayer.blogspot.co.uk/

 

RE: Devoting an entire core to audio, posted on December 17, 2016 at 21:58:01
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
Windows may prevent you from making changes to certain core processes. Whether total isolation of audio processes is achievable or desirable is a moot point and my own experience has been that a 'clean' audio computer and the assignment of player priority are rather important as well.

Adding software to 'realign' processes and priorities doesn't necessarily improve SQ. For one, you are adding software and processes to 'reduce' the influences of such. For another, you are 'fixing' your audio drivers to use a certain fixed set such as MCSS, WASAPI, and perhaps buffers to another person's 'optimal' perceptions in their own systems. I reckon that you should organise your own. For me, using KS and defeating all uneeded windows audio processes in a super slim and clean computer system powered by an excellent power supply works better than most.

Finally, there is the issue of cpu power in a single core for all the audio processes you want to run, and the power of the other cores to run other required processes for your system. With a low power cpu, this may actually cause interruptions to your audio.

The whole thing is very complex and trial and error.

 

RE: The whole thing , posted on December 18, 2016 at 00:11:28
soundchekk
Audiophile

Posts: 1906
Joined: July 11, 2007
Chinese.... Come on.


We've been discussing it over here during times of CPLAY and CMP and earlier.

The Fidelizer guy just build on everything that's been discussed and used it for his tool.


 

RE: The whole thing , posted on December 18, 2016 at 10:22:05
Windows X
Manufacturer

Posts: 209
Location: Thailand
Joined: February 28, 2011
Could you elaborate who

-Introduced MMCSS optimizations without affecting system configuration like applying registry tweaks
-Write program to actually do some API works to increase audio thread priority instad of batch script raising process priority alone

I haven't seen anyone doing these things long before inventing Fidelizer until now. And you're right that I've tried many tweaks from here. Who doesn't? I also have Phil's AO license and tested his latest version with Nimitra tonight. You don't simply get good sound from copying someone's tweaks. I prefer MSDN and scanning registry and actual audio related files as better sources for my research.

I believe AudioAsylum is a good place to start tweaking your computer for better sound quality. But it doesn't mean everyone will end up using only tweaks found in this community. Please, be innovate and talk with respect to earn respect from people. Happy listening. :)

Regards,
Keetakawee

 

RE: The whole thing - the issue for me is, posted on December 19, 2016 at 01:00:40
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
One starts with an OS, some perhaps loaded with all kinds of crap.

One buys 'optimisation' software which modifies the OS bahaviour for audio.

One then buys another player which also makes more alterations to the OS.

One then uses another software suite that dynamically alters OS behaviour to get the 'best sound'.

What does one end up with?

AS Prince Charles said once, a grey gooey!

 

RE: The whole thing - the issue for me is, posted on December 19, 2016 at 03:01:32
Mercman
Reviewer

Posts: 6278
Joined: October 20, 2002
Contributor
  Since:
May 20, 2004
Or one can do what I did in my review of AudiophileOptimizer. A clean install of Win 10 Pro 64 on 2 partitions. Apply AO to one partition. And yes, I need a music program. I used Roon. Also listened with Roon streaming to the HQPlayer. AO made a big difference compared to the partition with just Roon and the HQPlayer.

Your approach has never been clearly explained. Without getting you pissed off, how about giving us a clear explanation of what you do so I can try it. Pretend that you are writing a review so that we all might understand it.

This post is not meant to get you upset, so please take it as a discussion, not an insult.

 

RE: The whole thing - the issue for me is, posted on December 19, 2016 at 09:16:39
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
You are changing the subject from what I posted and you appear now to support what has always been good practice ie a clean and minimised OS. Previosuly, you had cocnsistently preferred multiple installs of 'optimising' software to give 'better' sound. So, good for you.

It is surprising that you chose to inatall W10 when Audiophil has always preferred Win Server 2012R2. If you had installed this in your trials, you may actuallly find even better better SQ. Reason? Server is slimmer than W10 and can be made very slim indeed.

My approach has always been simple and I have posted on this many times. Use the best hardware including power supplies. Do not use a laptop where components are packed close to and where extra software drivers are needed to run the system (over that of a desktop), plus where a fan (or fans) are mandatory. I do not install any software that is not needed for running an audio system, including graphic drivers fit for high resolution use but not needed for audio. I do not use anti virus software and I avoid the web. I use the best power supplies and mains regenerators, not oblique mains conditioners that are not clear in what they actually do. .

I am in favour of finding out the basics of what what 'optimising' software does, and try to understand how to slim a system and give priority to audio. I do not believe in sales words such as 'this device or software improves timing', which in itself has no meaning.

My 'purist' systems run on less than 280 threads and 5000 handles. They run on very little ram and perhaps 2-3% cpu power playing 384k 32 bit files. They do not go to the web to find pretty pictures for audio files. They do not rely on optimising software that add additional processing to get rid of 'uneeded ones'.

 

RE: The whole thing - the issue for me is, posted on December 19, 2016 at 09:28:09
Mercman
Reviewer

Posts: 6278
Joined: October 20, 2002
Contributor
  Since:
May 20, 2004
As stated in the review, Windows 10 was used since many of my readers have this OS. Server 2016 will be the source of a future review.

I guess the specifics of what you are slimming will remain with you.

 

What I said are the specifics, posted on December 19, 2016 at 10:52:56
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
as the starting point for better audio quality. If you are not even prepared to accept or discuss these, why bother asking?

 

Your world may be different from mine, but, posted on December 19, 2016 at 11:17:04
E-Stat
Audiophile

Posts: 24683
Joined: May 12, 2000
Contributor
  Since:
April 5, 2002
running two independent music streams each with their own FLAC decompression process using LMS uses < 1% of total CPU time on my 2010 Dell Studio with a quad core i7-860.

Processor affinity would be superfluous.

 

RE: Devoting an entire core to audio, posted on December 20, 2016 at 19:35:28
Windows X
Manufacturer

Posts: 209
Location: Thailand
Joined: February 28, 2011
Just curious. Does anyone notice the difference in sound between high power processor like i7 and low power ones like Celeron or Atom?

 

this is one of those, posted on January 6, 2017 at 03:59:35
fmak
Audiophile

Posts: 12784
Joined: June 1, 2002
self righteous statements without justification or reference to other high or low power systems.

 

Thanks again for more grins, posted on January 6, 2017 at 06:03:45
E-Stat
Audiophile

Posts: 24683
Joined: May 12, 2000
Contributor
  Since:
April 5, 2002
I just about sprayed coffee from my nostrils when I read your post!

It's clear you don't understand the objective of the kernel parameters you suggest tweaking. Hint: it's to give them scheduling priority so that they don't have to wait for any of the CPU cores to execute. :)

Here's a visual of my LMS server with one music thread in operation across the space of thirty seconds:





 

Page processed in 0.037 seconds.