[Solved] This one weird issue with Equalizer APO 1.2.1 and Big.Little CPUs

[anphex]

Edit: Pops, clicks and distortion solved. Look at the 3rd edit of this post

Hello everyone,

I recently upgraded my home audio system and wanted to share my experience, hoping it might help others facing similar issues.

The Upgrade: I've set up a new surround sound system using my existing equipment and a new amplifier. An important change was connecting my PC directly to the TV via HDMI. This was prompted by LG's recent update that fixed the HDMI eARC audio PCM Passthrough issue, allowing 7.1 surround sound instead of just 2.1. This setup also helps avoid HDR desync and image loss problems.

Performance Improvement Tips: The setup works wonderfully, and I recommend it. Additionally, I found that disabling GSync improves visuals, as it causes gamma curve delays and isn't really beneficial if your refresh rate exceeds your panel's maximum.

The Main Issue - Audio Pops and Clicks: Despite the successful setup, I encountered an issue with audio pops and clicks, especially noticeable in calm, atmospheric music. This reminded me of a similar problem I had previously with my Intel 12700k CPU. At that time, Windows and eAPO routed the FIR-processing to the CPU's smaller cores, which sometimes weren't fast enough, causing audio disruptions. The temporary fix was to disable the e-cores.

Fast forward to 2023, I upgraded to an Intel 13700k, and initially, everything worked fine. However, with my new setup, I made two key changes:

1. I switched from using Audyssey to REW FIR + EAPO for audio processing on my PC.
2. I significantly increased the number of FIR-using channels, currently 4.

Reflection and Questions: This is the first time I'm using eAPO with multiple channels instead of Audyssey since a while. And again, I have to disable the e-cores if I want to have pop-free music. I wonder why these audio issues occur in the first place since e-core SHOULD be fast enough, don't they? While I know eAPO is somewhat outdated and considered legacy software, I'm not aware of any viable alternatives. Does anyone have insights or suggestions on how to address this issue so I don't have to decide between ecores or error free music?


Thanks for reading, and I appreciate any advice or thoughts you might have, even if it's a super niche problem.


Edit: The more I listen to music the more I feel as if there is also ... less distortion? May it actually be that the ecores couldn't process the FIR at all and the pop/clicks were just the worst manifestations? Weird.

Edit 2: Also for everyone using APO. Follow the tips in this thread: https://www.audiosciencereview.com/...nding-the-windows-audio-quality-debate.19438/ Disabling the Original APO made an audible improvement.

Edit 3: My initial solution was like throwing all car tires away just because one was slightly damaged. Instead of disabling all e-cores in the BIOS, a far more simple and effective solution is to use a free tool like "Process Lasso" and permanently assign the audio processes to the p-cores. and give them slightly higher priority. Also uninstall the "original APOs" in the Equalizer-APO. If you want no-compromise system performance for audio, also disable hyper-threading and raise the priority to realtime. But be careful with the latter one.

 

[AnalogSteph]

You could try identifying the Audiosrv PID via Task Manager's Services tab, and then manually set its thread affinity. At least I would think that APOs would be running in that process...

 

[anphex]

You could try identifying the Audiosrv PID via Task Manager's Services tab, and then manually set its thread affinity. At least I would think that APOs would be running in that process...

Sir, you solution is so incredibily elegant. I knew you could assign priorities in the task manager but CPU affinity? That was a new for me. Immediately I found a tool for long term adjustment of this affinity.

I am now using Process Lasso in the already very extensive free version and permanently gave all system audio processses a "higher than normal" priority and locked them out of using the e-cores.

So yes, my problem is solved, with just one elegant suggestion. Wow. Thank you so much!

 

[anphex]

I noticed that it may not be the e-cores fault but the windows core scheduler that mixes the p- and e-cores up for eAPO.
I am currently running the audiodg.exe only on the e-cores with "high" priority and it's also pop free.

 

[AnalogSteph]

Definitely interesting. I have yet to play with any such processors myself, the latest I have access to would be an i7-11700 or an i3-12100, neither of which sport any E-cores or would be likely to be upgraded any time soon.

 

[anphex]

New insights - I played around with different sample rates and it emerged that the higher the FIR file sample sample rate, the more likely it was for pops, distortion and plain underruns to happen. 44,1 Khz is fine, 48 Khz worked too, glitches started to appear more at 96 Khz and at 196 Khz it wasn't usable at all.

This was only on the e-cores, or more precisely ONE e-core since audiodg.exe does not seem to make use of parallelism at all.
When I switched to ONE p-core, everything was perfect, even at 196 Khz.

This makes me think if the p-cores have some FIR-specific instruction that the e-cores do not have - or it just boils down to pure higher clock of the p cores.