Audio stutters with USB Dacs on Macbook M1 Pro

[pengin15]

Just throwing out some ideas on why the Qudelix 5K (and only that) seemed to work:

• Protocol - From what I understand Qudelix 5k is one of the few native USB3.0 DAC, while most DACs uses USB2.0. In theory, this would mean other DACs using USB3.0 (such as on top of my head, iFi Zen V2) should also work just fine. Scratch that. Posts below indicates that Qudelix 5k is not USB3.0
• Powered - Unlike most DAC/amps folks around here would have, the Qudelix 5K is powered straight from your PC's USB port. It does not require a separate power line. I think someone in this thread also mentioned that their Audioquest Dragonfly also worked fine with no issue, which might lend some credence into this theory.

I'm this close to just selling my M1 Pro MBP and just get myself another one with 32GB RAM, which is absolutely ridiculous.

 

[Zensō]

I'm this close to just selling my M1 Pro MBP and just get myself another one with 32GB RAM, which is absolutely ridiculous.

Don’t. I have an M1 Mac Studio with 32GB RAM and I’ve had to deal with the stuttering issue. I have it mostly solved using a plugin to increase the audio buffer, but 32GB is not a guarantee that it will be 100% free from stuttering.

Anecdotally, it seems the M2 doesn’t have this problem, but I haven’t confirmed it myself.

 

[DWPress]

So far the M2 Mini with 24GB RAM is fine. If I couldn't have justified the upgrade of the machine for other business use I'd be even more upset with Apple.

 

[kuil5230]

I just upgraded to macOS Sonoma beta and so far I haven't encountered any audio stutters. OMG
My setup includes a MacBook Pro M1, Fiio K5 Pro, and HD 660S.

 

[pengin15]

So far the M2 Mini with 24GB RAM is fine. If I couldn't have justified the upgrade of the machine for other business use I'd be even more upset with Apple.

I wish I could've gone with a Mac mini, but I travel too much to be able to use one. The other alternative to the M1 Pro MBPs are of course the M2 Pro/Max series, but all anedcotal evidences seem to conclude that they run hotter with less battery life compared to their M1 counterparts.

I just upgraded to macOS Sonoma beta and so far I haven't encountered any audio stutters. OMG
My setup includes a MacBook Pro M1, Fiio K5 Pro, and HD 660S.

OMG. Could I trouble you to check your memory pressure in Activity Monitor, and whether you can trigger the stutters when you increase RAM usage? I usually use www.trackthis.link to automatically open 100-200 browser tabs or so to artificially increase memory pressure to test.

IF they really fixed it, it might just be worth it to lose access to SoundSource until late September

In other news, Fiio's cables for flashing the Fiio K7's MCU just arrived this morning and after fiddling around with it all morning, I finally managed to flash the MCU. As @jayadubya rightfully pointed out - a very annoying and painful process, and the bad English instructions doesn't help. Am testing the new firmware right now, let's see how it goes.

@jayadubya I can probably give you some tips on how to correctly position the jig if you still haven't managed to flash yours.

 

[dweeeeb2]

I haven’t read all this but on my intel laptop power management played a big roll. If I set power options to best performance and plug the laptop in I get glitch free playback.

 

[kuil5230]

OMG. Could I trouble you to check your memory pressure in Activity Monitor, and whether you can trigger the stutters when you increase RAM usage? I usually use www.trackthis.link to automatically open 100-200 browser tabs or so to artificially increase memory pressure to test.

My RAM usage consistently remains above 80%, and I have never experienced any lag even when it exceeds 90%. For me, macOS Sonoma has already resolved this issue.

 

[theREALdotnet]

My RAM usage consistently remains above 80%, and I have never experienced any lag even when it exceeds 90%. For me, macOS Sonoma has already resolved this issue.

Fingers crossed!

 

[Zensō]

I haven’t read all this but on my intel laptop power management played a big roll. If I set power options to best performance and plug the laptop in I get glitch free playback.

Thanks, but this is isolated to the M1 Macs. No one was having this specific issue with older Intel Macs.

 

[pengin15]

I bit the bullet and installed macOS Sonoma's developer beta 1 (losing SoundSource in the process, which sucks) to see if it fixes the issue. It unfortunately did not for me and, while the frequency might be somewhat lesser, the same issue showed up when RAM usage passes the 13.5GB threshold. Console app showed the same "kernel trap" issue so whatever Apple did, they didn't seem to have fixed the underlying issue. In any case, Sonoma seems to be running at a slightly lower RAM bandwidth than Ventura in my observation, so it's pretty likely that the reduced stuttering might be attributable to the fact that it takes longer to reach the necessary memory pressure threshold than Apple having fixed coreaudiod.

On a related note, I'd encourage anyone trying macOS Sonoma to bug (ha!) Apple using Feedback Assistant. It's beta season and in my experience, they are more attentive to bug reports during this time.

I haven’t read all this but on my intel laptop power management played a big roll. If I set power options to best performance and plug the laptop in I get glitch free playback.

Very likely a different issue than what we're dealing with here. Audio playback always shits the bed when CPU is being taxed, and this is not exclusive to OS X. I've had my share of this issue in my previous Intel macbooks. This is also why forcing the CPU to work more through power management fixed your issue. In the case of Apple Silicon, anecdotes seem to point to how RAM management interfacing with coreaudio being the problem, although that doesn't seem to be the farthest extent of the issue since inexplicably non-USB2.0 protocols doesn't trigger the bug.

N.B. For those reading this later and wondering, FiiO's K7 MCU patch did not resolve the issue.

 

[Zensō]

I bit the bullet and installed macOS Sonoma's developer beta 1 (losing SoundSource in the process, which sucks) to see if it fixes the issue. It unfortunately did not for me and, while the frequency might be somewhat lesser, the same issue showed up when RAM usage passes the 13.5GB threshold. Console app showed the same "kernel trap" issue so whatever Apple did, they didn't seem to have fixed the underlying issue. In any case, Sonoma seems to be running at a slightly lower RAM bandwidth than Ventura in my observation, so it's pretty likely that the reduced stuttering might be attributable to the fact that it takes longer to reach the necessary memory pressure threshold than Apple having fixed coreaudiod.

On a related note, I'd encourage anyone trying macOS Sonoma to bug (ha!) Apple using Feedback Assistant. It's beta season and in my experience, they are more attentive to bug reports during this time.



Very likely a different issue than what we're dealing with here. Audio playback always shits the bed when CPU is being taxed, and this is not exclusive to OS X. I've had my share of this issue in my previous Intel macbooks. This is also why forcing the CPU to work more through power management fixed your issue. In the case of Apple Silicon, anecdotes seem to point to how RAM management interfacing with coreaudio being the problem, although that doesn't seem to be the farthest extent of the issue since inexplicably non-USB2.0 protocols doesn't trigger the bug.

N.B. For those reading this later and wondering, FiiO's K7 MCU patch did not resolve the issue.

Ah bummer.

I’ve seen reports of Thunderbolt 4 docks placed between the computer and DAC solving this. I have a CalDigit TS4 dock arriving today. I’ll report back.

 

[silwynar]

Ah bummer.

I’ve seen reports of Thunderbolt 4 docks placed between the computer and DAC solving this. I have a CalDigit TS4 dock arriving today. I’ll report back.

Got a CalDigit TS3+ between my Macbook Pro 14" Base and a Topping DX3+
Sadly same issues, so a dock doesnt seem to help.

Using the Element solution with a 1024 buffer also didn't really help. Maybe the stutters got a tad better, but still annoying.
Also kinda cumbersome as I have to resetup the inputs everytime I plug my macbook off my dock. :/
2048'er buffer sadly gave me noticeable input lag.

Urgh, so annoying.
Really close to plugging my speakers via 3.5mm into the CalDigit, but making my lovely Topping go to waste? Mhmhm.

 

[Zensō]

Got a CalDigit TS3+ between my Macbook Pro 14" Base and a Topping DX3+
Sadly same issues, so a dock doesnt seem to help.

Using the Element solution with a 1024 buffer also didn't really help. Maybe the stutters got a tad better, but still annoying.
Also kinda cumbersome as I have to resetup the inputs everytime I plug my macbook off my dock. :/
2048'er buffer sadly gave me noticeable input lag.

Thanks. I just set up the CalDigit TS4 and can confirm that it DOES NOT solve the stutter problem in my setup either (M1 Mac Studio, 32GB RAM > RME Babyface Pro FS). I needed a new dock anyway, so not a total loss.

Urgh, so annoying.
Really close to plugging my speakers via 3.5mm into the CalDigit, but making my lovely Topping go to waste? Mhmhm.

Annoying indeed. I use my interface for music production, so the analog out is not a solution for my situation. I'm considering selling my M1 Mac Studio and picking up either an M2 Mini or Studio.

I can't recall, has anyone tested a USB 3 or Thunderbolt DAC or interface? Wondering about the iFi iDSD and the UAD Apollo among a few others...

 

[silwynar]

I can't recall, has anyone tested a USB 3 or Thunderbolt DAC or interface? Wondering about the iFi iDSD and the UAD Apollo among a few others...

Regarding that the Qudelix 5k is a DAC that uses USB3 via USB-C , there is the idea that every audio interface not using USB2 is not affected by this issue.
(wrong informations, see replies below)
As far as I read into it, this theory is not confirmed though. We just know that the Qudelix 5k is working without issues.

Is it confirmed that the M2 chips don't have this issue? This would mean that it maybe isn't an software, but actual hardware issue and we're all doomed? haha

 

[unpluggged]

Regarding that the Qudelix 5k is a DAC that uses USB3 via USB-C

It does not. It's a USB Audio Class 1.0 device that uses USB 2.0 interface: https://docs.qualcomm.com/bundle/pu...SERIES_BLUETOOTH_AUDIO_SOCS_PRODUCT_BRIEF.pdf

 

[silwynar]

It does not. It's a USB Audio Class 1.0 device that uses USB 2.0 interface: https://docs.qualcomm.com/bundle/pu...SERIES_BLUETOOTH_AUDIO_SOCS_PRODUCT_BRIEF.pdf

My bad! Picked up misinformations there then. Thanks for correcting!

 

[jayadubya]

My RAM usage consistently remains above 80%, and I have never experienced any lag even when it exceeds 90%. For me, macOS Sonoma has already resolved this issue.

Sonoma! My home county. Could this be the fix?

 

[Zensō]

Sonoma! My home county. Could this be the fix?

Apparently not:

Audio stutters with USB Dacs on Macbook M1 Pro

Just throwing out some ideas on why the Qudelix 5K (and only that) seemed to work: Protocol - From what I understand Qudelix 5k is one of the few native USB3.0 DAC, while most DACs uses USB2.0. In theory, this would mean other DACs using USB3.0 (such as on top of my head, iFi Zen V2) should...

audiosciencereview.com

 

[pengin15]

It does not. It's a USB Audio Class 1.0 device that uses USB 2.0 interface: https://docs.qualcomm.com/bundle/pu...SERIES_BLUETOOTH_AUDIO_SOCS_PRODUCT_BRIEF.pdf

Thanks for this. Crossed out my post above so no one gets mislead.

I wish it would have been as simple as getting an M2 (Pro/Max) device, but I otherwise really, really love how quiet and how good the thermal on my M1 Pro MBP is, so switching to an M2 Pro/Max device and swallowing the worse thermals and battery life is really not preferable. Apple did not announce the M3 during WWDC (rumored to be a new smaller-node "tick" chip), so let's assume M3 Pro will be at least a year away. This is quite annoying.

Should we do a brief sum up on what works, what alleviates, and what doesn't work thus far?

What worked for me

• Built-in HDMI -> HDMI extractor -> Toslink -> DAC - Can confirm that this completely fixed the issue for me. Zero stutters.

What alleviated the issue for me

• Soundsource "more reliable" latency setting / SoundSource - Element - Blackhole 2ch chain - Can also confirm that this alleviates the audio skips, if only very slightly. Think six/seven stutters from the initial ten. I don't think there are any perceivable differences between just using SoundSource alone and using the Soundsource - Element chain.
• macOS Sonoma (as of DB1) - It does seem to reduce stutter, but I think only because it might just have improved enough on memory management that the "stutter threshold" doesn't get breached as often. You can try increasing your memory pressure and the same skips will start again. Console still shows the same "kernel trap" error whenever this happens, so the fundamental cause have not been resolved.

What just doesn't work for me

• FiiO's new MCU firmware for its K7 DAC/Amp - Doesn't seem to have any effect whatsoever in my experience.

Other possible solutions so far?

• M2 doesn't seem to suffer from this issue?
• The brute force way - just get more RAM. I was considering this but yeah, as @acbarn noted, won't really solve the issue. Also, I just remembered that the way macOS handles RAM is opposite Windows in which they are not empty but rather macOS sort of "fills everything to the brim" so that it can do its thingamajig and loads stuff faster. IDK I'm not smart enough for this but I reckon this might not be as effective as it seems to be.

back when I started having this issue on my then new MBP M1 Pro using Big Sur, on a Reddit thread (or maybe MacRumors, can't remember now) someone mentioned a tiny application called MacClickFixus. it was created to keep the audio bus open continuously for people having issues with audio click and pop noises when the interface is initialized.

back then anyway, keeping it running completely fixed the external DAC dropout problem for me. sadly though, it doesn't seem to work anymore after one of the Ventura updates. it's also an Intel binary (that's what I have), so maybe the added layer of Rosetta doesn't help.

the binary is still out there if people want to give it a go: macClickFixus.app.zip - Google Drive

no, I'm not the author of it.

I keep coming back to this post by eelpout. Does anyone know of any app that might do the equivalent of macClickFixus does for Ventura onwards?

And I thought the worst thing Ventura did was to abandon the old System Preferences - look like it actively worked to do away with our solutions too...

 

[antcollinet]

Out of interest - what data rate (bit depth / sample rate) are people having this problem using?

I use 24/48 and see no issue. Though I also have 32GB ram.