Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
I need someone to help me test a bug.

So far, I have tried two different 2012 MBA's on 2 different Thunderbolt displays, with the same result.

Essentially, when the GPU and CPU are being taxed, the sound coming out of the Thunderbolt slowly gets more and more distorted. It starts out fine, but slowly the sound degrades into static, until it's unlistenable.

If l click from "Display Audio" to "Internal Speakers" the sound is fine, and if I wait 10 seconds and switch back to "Display Audio", the sounds is fine but then builds right back up to the distortion.

I've been working with Applecare on this, and so far they first replaced a Thunderbolt display, but a new display had the same problem, so then they replaced the Macbook Air... but there was still the same problem.

I have installed NOTHING on this computer. I also disconnected all keyboards/mice/network cables, etc etc.

So I've replicated the bug a bunch of different ways, but one consistent way is to restart the machine, and then start a song, and open up photobooth and set it to record a video. Within a minute or 2, the distortion starts up. If you have a game like diablo3 or minecraft, these will also trigger the bug (if you have it).

I had minecraft installed to test things out, and it would also trigger the problem, but I then did a clean restore to rule that out, and the problem still persists, on two 2012 MBA's on 2 different TB monitors. How is that even possible?

Also - I have a 2011 MBA that I use on the same TB display, same keyboard/mouse, and it's never had a problem like this.

So if anyone can run a test to see if ANYONE else can replicate this problem, that would be great. Apple has been very helpful in replacing the Monitor, and then the computer, but I don't see how I could have the same bug twice, where I've ruled out everything I can. I am having the same problem now on a totally new pair of monitor/computer, with no software installed???

It just baffles me. I am wondering if there is a problem with the ivybridge interaction between GPU and audio.

If someone could run through the scenario for me, that would be really awesome. I can't just say "Well I'll ask for an exchange" since I've already done that with the computer and the monitor.
 
Last edited:

jsolares

macrumors 6502a
Aug 8, 2011
844
2
Land of eternal Spring
I need someone to help me test a bug.

So far, I have tried two different 2012 MBA's on 2 different Thunderbolt displays, with the same result.

Essentially, when the GPU and CPU are being taxed, the sound coming out of the Thunderbolt slowly gets more and more distorted. It starts out fine, but slowly the sound degrades into static, until it's unlistenable.

If l click from "Display Audio" to "Internal Speakers" the sound is fine, and if I wait 10 seconds and switch back to "Display Audio", the sounds is fine but then builds right back up to the distortion.

I've been working with Applecare on this, and so far they first replaced a Thunderbolt display, but a new display had the same problem, so then they replaced the Macbook Air... but there was still the same problem.

I have installed NOTHING on this computer. I also disconnected all keyboards/mice/network cables, etc etc.

So I've replicated the bug a bunch of different ways, but one consistent way is to restart the machine, and then start a song, and open up photobooth and set it to record a video. Within a minute or 2, the distortion starts up. If you have a game like diablo3 or minecraft, these will also trigger the bug (if you have it).

I had minecraft installed to test things out, and it would also trigger the problem, but I then did a clean restore to rule that out, and the problem still persists, on two 2012 MBA's on 2 different TB monitors. How is that even possible?

Also - I have a 2011 MBA that I use on the same TB display, same keyboard/mouse, and it's never had a problem like this.

So if anyone can run a test to see if ANYONE else can replicate this problem, that would be great. Apple has been very helpful in replacing the Monitor, and then the computer, but I don't see how I could have the same bug twice, where I've ruled out everything I can. I am having the same problem now on a totally new pair of monitor/computer, with no software installed???

It just baffles me. I am wondering if there is a problem with the sandybridge interaction between GPU and audio.

If someone could run through the scenario for me, that would be really awesome. I can't just say "Well I'll ask for an exchange" since I've already done that with the computer and the monitor.

It's a known issue, the sandybridge macs also have it, you can see it discussed on anandtech and many other sites
 

calvol

macrumors 6502a
Feb 3, 2011
995
4
Formerly, I worked in the chip industry, and when the PCIE bus standard came out, there were a lot of internal shielding requirements imposed on PC OEMs because PCIE basically runs at 2.5GHz, and grossly interfered with the upper channels of WiFi in the 2.4GHz band. Fast forward to today, (not in the industry anymore), but I would bet there are continued noise issues with Thunderbolt, specifically the cable, acting as an antenna, causing issues with the audio amp on the logic board. No doubt this is the dirty little secret of TB!
 

daz369

macrumors newbie
Jul 12, 2011
4
0
same issue here. I upgraded from 2011 MBA (which has never had the issue).
Within an hour of watching EyeTV on the 2012 MBA the Thunderbolt speakers are distorted.
Changing back to the internal/onboard speakers and then back to Thunderbolt display audio, temporarily fixes it.

There's a thread (among a few) on Apple Support Communities:
"HT1338 Audio Problem with 2012 MacBook Air and Thunderbolt Display" where quite a few people are having the issue".

I also called Apple Support who are wanting me to perform SMC reset and various other things, I hope to get some time later today to go through it with them on the phone.
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
I have actually gone through all the applecare phone steps, even sent them the machine, and am waiting to hear back from apple engineering.

So this is just a random case? Or are there just not many people using a TB monitor with a 2012 MBA? If so, do I just need to use a different computer, like a retina or regular MBP?
 

astormsau

macrumors newbie
Jun 14, 2012
11
0
Have exactly the same problem, happens in iTunes after listening to music for an hour or two. Changed the display with Apple within 14days of purchase and had the same issue with the new display, figured its either an inherent problem in the sound card in the monitor or something thunderbolt related.

See this post for some more info from others with similar issues: https://forums.macrumors.com/threads/1241915/

and my thread: https://forums.macrumors.com/showthread.php?p=15169406#post15169406

Good luck, will be interested in a solution :)
Drew
 

Daginator

macrumors newbie
Apr 22, 2012
27
0
Do you have a soundclip of the "distorted sound"? My last air that I swapped for a new air had this issue I think. Happend one time and I paniced and restarted it. Have not connected my new air to the display yet.
 

TLewis

macrumors 65816
Sep 19, 2007
1,294
120
Formerly, I worked in the chip industry, and when the PCIE bus standard came out, there were a lot of internal shielding requirements imposed on PC OEMs because PCIE basically runs at 2.5GHz, and grossly interfered with the upper channels of WiFi in the 2.4GHz band. Fast forward to today, (not in the industry anymore), but I would bet there are continued noise issues with Thunderbolt, specifically the cable, acting as an antenna, causing issues with the audio amp on the logic board. No doubt this is the dirty little secret of TB!
Dunno. While that's certainly possible, that doesn't seem to explain the slow sound degradation. I'd suspect something more along the lines of the speaker output somehow being fed back into the microphone circuitry (not necessarily via the microphone itself, but possibly via noisy inductive/capacitive coupling somewhere in the circuitry). Alternatively, maybe something's borked with the echo cancellation??

As a quick test, I'd try muting the microphone. If that solves the problem, bingo. If it doesn't, that doesn't necessarily say anything either way.
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
Do you have a soundclip of the "distorted sound"? My last air that I swapped for a new air had this issue I think. Happend one time and I paniced and restarted it. Have not connected my new air to the display yet.

I do. In this clip, recorded with my iphone, I have the Mountain Lion preview video playing off Apple's site. At 1:30 in, I switch from the display back to the laptop. This one is after it's built up a bit, so just imagine it less or more distorted depending on how long it has been playing.

Dunno. While that's certainly possible, that doesn't seem to explain the slow sound degradation. I'd suspect something more along the lines of the speaker output somehow being fed back into the microphone circuitry (not necessarily via the microphone itself, but possibly via noisy inductive/capacitive coupling somewhere in the circuitry). Alternatively, maybe something's borked with the echo cancellation??

As a quick test, I'd try muting the microphone. If that solves the problem, bingo. If it doesn't, that doesn't necessarily say anything either way.


It's not this. The problem seems to be linked to heavy GPU usage or CPU usage.

Like I said, I opened up the machine from a fresh boot, turned the sound all the way down, opened up PhotoBooth and set it to record a video, let it go for a few minutes, and then turned on itunes and the sound was already messed up. I could see the CPU usage in the activity monitor, it was decent but not maxed out.

I tried the same thing with a handbrake conversion of a video file, which doesn't use GPU - wasn't able to push the sound to distortion even with a good 15 minutes of max CPU.

I've tested this in every permutation I can think of with hardware and software, and it is almost certainly linked somehow to GPU stress. Either way, I've had the exact same problem on 2 totally different computers/monitors:

MBA-1, TB-1 = Problem
MBA-1, TB-2 = Same problem
MBA-2, TB-2 = Same problem

Also, I have a 2011 MBA that does not have this problem on TB-1 or TB-2.

If it was something else, I could overlook it, but not being able to watch video do anything that needs GPU is too much to overlook.
 

Attachments

  • example.m4a.zip
    856.2 KB · Views: 89

BlueOcean

macrumors member
Jun 15, 2012
69
0
This is rather worrying. When I get my MBA, I'll try hook it up to my TV (HDMI) to see if the issue is the same. Is is only the TBD that's giving the issue?

If the problem is consistent, that seems bad enough for me to return the Air and get a Pro instead. In fact, it seems like something that Apple should recall products for (especially since this is a compatibility issue with 2 products they both produce), but I guess that's unlikely.
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
This is rather worrying. When I get my MBA, I'll try hook it up to my TV (HDMI) to see if the issue is the same. Is is only the TBD that's giving the issue?

If the problem is consistent, that seems bad enough for me to return the Air and get a Pro instead. In fact, it seems like something that Apple should recall products for (especially since this is a compatibility issue with 2 products they both produce), but I guess that's unlikely.

That's all I tested it on.

I did hook it up to an HDMI adapter to my TV, and the TV display just went black every 8 seconds or so, over and over... but that's a different bug for a different thread. Again, my 2011 MBA doesn't have this problem.
 

asting

macrumors 6502
Jun 10, 2012
378
3
That's all I tested it on.

I did hook it up to an HDMI adapter to my TV, and the TV display just went black every 8 seconds or so, over and over... but that's a different bug for a different thread. Again, my 2011 MBA doesn't have this problem.

Mine does it fine through a cheapo no name Mini dp to hdmi adapter. It outputs sound just fine too. I didn't do too much testing, but it seemed good for the half an hour or so i used it.
 

calvol

macrumors 6502a
Feb 3, 2011
995
4
Dunno. While that's certainly possible, that doesn't seem to explain the slow sound degradation. I'd suspect something more along the lines of the speaker output somehow being fed back into the microphone circuitry (not necessarily via the microphone itself, but possibly via noisy inductive/capacitive coupling somewhere in the circuitry). Alternatively, maybe something's borked with the echo cancellation?.

I'm thinking something is filling up the USB buffers slowly, causing buffer overrun, whatever the source-- PCIE noise, echo, or mike feedback. Sounds like it's unique to the 2012.
 

TLewis

macrumors 65816
Sep 19, 2007
1,294
120
I'm thinking something is filling up the USB buffers slowly, causing buffer overrun, whatever the source-- PCIE noise, echo, or mike feedback. Sounds like it's unique to the 2012.
It's Thunderbolt, not USB, but you might be on to something.

After listening to the sample, and the OP reporting that it's related to heavy GPU, I'm now wondering if heavy GPU usage is somehow preventing the outgoing Thunderbolt output buffers from being filled/handled properly. Too many interrupts by the GPU, or maybe the GPU is preventing interrupts from being properly handled? Is there some system monitoring app that displays interrupts per second?
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
It's Thunderbolt, not USB, but you might be on to something.

After listening to the sample, and the OP reporting that it's related to heavy GPU, I'm now wondering if heavy GPU usage is somehow preventing the outgoing Thunderbolt output buffers from being filled/handled properly. Too many interrupts by the GPU, or maybe the GPU is preventing interrupts from being properly handled? Is there some system monitoring app that displays interrupts per second?

Dunno. But I am just waiting to hear back from Apple. If theres not a surefire fix, I will probably just ask for a refund authorization. I've had it about a month now, but haven't been able to use it once, and a week of that I had no computer at all because Apple wanted me to send it back.

It is definitely correlated with the GPU. I did lots of CPU testing with things that maxed out all 4 cores, and had no problems. But if I were to load something that used GPU, like say Diablo 3 or Minecraft, or even an iTunes movie, within minutes the sound would degrade and eventually become totally wrecked.

I thought it was heat, too, but after about 40 minutes of a handbrake encode maxing out all 4 cores, with no problems, I ruled that out.

What REALLY sucks is that I am a designer and spend all my working time in Photoshop, and that causes the problem too with the openGL features, so I can't even listen to music while I work. Now, I *could* use the laptop speakers.... but I've got about $3200 bucks invested into the MBA and the TBD, including applecare on both, so I really can't bring myself to overlook it or just compromise.
 

wditters

macrumors member
Apr 10, 2012
80
0

This must be a driver issue in OSX. I have a Lenovo T430s with Thunderbolt and Integrated Graphics HD 4000 and that machine has no issue. I have been able to watch 4 complete 1080p movies completely without issues connected to a Samsung Full-HD TV via TB->HDMI on the Thinkpad. On the MBA however I have have ran into the same issue each time I tried. This issue really is exclusive to Apple.
 

dark knight

macrumors regular
Oct 28, 2008
154
5
bizarrely, i just got the exact same symptom from my new rMBP connected via USB to an MBox 2 (for Protools software).

Its such a strange problem that it takes a couple of minutes to build up static, to the point that the music becomes unrecognisable. As of 30mins now i seem to have fixed the problem by re-downloading the driver package (same version i believe) and reinstalling. Its definitely data related rather than the regular audio signal picking up interference. The issue jumps around with severity depending on what application you are listening to. Some apps seemingly less affected than others until it all falls over into static.

It may have something to do with the presence of thunderbolt. my previous unibody did not have TB and never had this issue. same copy of Lion, same driver install etc.
 

TLewis

macrumors 65816
Sep 19, 2007
1,294
120
Its such a strange problem that it takes a couple of minutes to build up static, to the point that the music becomes unrecognisable. As of 30mins now i seem to have fixed the problem by re-downloading the driver package (same version i believe) and reinstalling. Its definitely data related rather than the regular audio signal picking up interference. The issue jumps around with severity depending on what application you are listening to. Some apps seemingly less affected than others until it all falls over into static.
Sean's issue appears to be GPU-related. Try doing something that hammers the GPU, and then see how long it takes for the static to build up.
 

dark knight

macrumors regular
Oct 28, 2008
154
5
OMG! you are right

having a the Retina MBP i can choose which GPU is active (with an app).

if i force the computer to use the powerful GPU the problem did not seem to fire up. then i forced the integrated intel HD4000 graphics and played video, used google earth etc.

Problem reoccured (damn!!!)
oddly, switching back to the NVIDIA gpu did reduce the problem but not restore the clean audio. i'm going to restart now and then only use the NVidia gpu to see if the problem can be instigated whilst in this mode.

Heck this is annoying :/
 

dark knight

macrumors regular
Oct 28, 2008
154
5
Well, if you only use the big GPU then the problem takes longer to occur, but it does. :/

So.....Mountain Lion address this next week?

----------

in an odd twist, my issue does add USB to, apparently, the same problem

----------

from a logical perspective (and one of no software knowledge at all) it seems that the audio signal is being degraded between its source (on the main board somewhere) and one of the two buses (USB or TB). however it is not degraded when the same source is routed to the internal amp/speakers.
the source of the degradation is related to the GPU heating up, possibly the fan kicking off? or does the MBA not have a fan? Though that said, once started, the problem does not go away when cooled down. So it must be software related, perhaps as was commented earlier, some kind of buffer overflowing. It takes a restart to fix the problem.
 

Daginator

macrumors newbie
Apr 22, 2012
27
0
I have the exact same issue! When the sound started to get distorted I plugged in my headphones in the headphone jack. I could still hear sound thru the display but now the sound was fine. What are you guys going to do? Replace the screen even if it doesn't fix the problem (just to get a fresh display perhaps) or wait for a fix?
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
After like 3 weeks of calls back and forth, I was told that if I want to do a workaround, then I need to disable WiFi while I am plugged in to the Thunderbolt Display, and use the ethernet instead.

they are leaving my case open and are going to notify me when there is a fix that will come in the form of a software update. I am not sure if that means Mountain Lion or a dot release or just a software fix, but /shrug that's all I can do unless I want to return it, which I don't.

They did say that a lot of people have experienced this problem and it's a known issue they are woking on a solution for. It sucks that there is no fix for it, since I know that disabling WiFi won't work for everyone. Hopefully there is a real fix and this isn't just something that drags on forever, it would really kill the resale value to have a computer that doesn't function right with a thunderbolt display.
 

calvol

macrumors 6502a
Feb 3, 2011
995
4
Sean, thanks for posting that followup. My suspicion was correct, i.e., there is an interaction between the PICE bus running at 2.5Ghz and WiFi in the 2.4Ghz band. I was a systems engineer in the PC industry when PCIE first came out, and we had to use excessive shielding in laptops to reduce the interference. I always thought this would be the Achilles heel of Thunderbolt. It would be interesting if you could operate WiFi at 5Ghz only and see if you have the same problem.

Note: PCIe bus: 100Mhz x 25 = 2.5Ghz (PCIe 1.1 spec).
 

Sean Dempsey

macrumors 68000
Original poster
Aug 7, 2006
1,622
8
Sean, thanks for posting that followup. My suspicion was correct, i.e., there is an interaction between the PICE bus running at 2.5Ghz and WiFi in the 2.4Ghz band. I was a systems engineer in the PC industry when PCIE first came out, and we had to use excessive shielding in laptops to reduce the interference. I always thought this would be the Achilles heel of Thunderbolt. It would be interesting if you could operate WiFi at 5Ghz only and see if you have the same problem.

Note: PCIe bus: 100Mhz x 25 = 2.5Ghz (PCIe 1.1 spec).

I will switch to 5ghz and test it out for a while.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.