Sometimes the kernel pops, sometimes it doesn’t. We’re waiting for a Leopard fix, and have reason to believe we may not have to wait much longer. Photo CC EastBayAnt.

Updated: As expected, the Mac OS X 10.5.3 update has been released, and it promises to address USB audio playback issues. That may or may not fully resolve issues users have been encountering; I expect we’ll know more soon.

Let’s get one thing straight: if you’re having audio problems under Mac OS X 10.5.2 and you’re looking for someone to blame, it’s reasonably safe to blame Apple, in case that wasn’t already obvious. That’s just this particular case, and it’s not the first (or last) time an OS update caused issues for audio, but that’s my best appraisal of the situation.

I made no secret that I was disappointed with the level of support for emerging OSes from M-Audio and Digidesign, and I stand by that complaint. The response from M-Audio and Digidesign was prompt: based on what I heard from them (and they wrote me personally), I don’t think they made any “excuses.” Representatives from both product lines apologized for lagging drivers, and promised to do better. We’ll of course watch to see if they deliver on that promise in the long haul, but they were at least able to offer some specific clarifications and updates so owners of their products can make some progress right now. (Read my original complaint, and follow-ups from M-Audio and Digidesign.)

Accountability and OSes

That said, all evidence points to audio performance problems on Mac OS X Leopard 10.5.2 specifically being the responsibility of Apple. Pointing out problems introduced by an operating system is not making an excuse; it’s stating an obvious fact, whether in regards to Microsoft or Apple. In this case, the symptoms are not restricted to a single product vendor. Apple’s own Logic Studio is an affected piece of software. (Heck, even Skype and iTunes may have issues.) I’ve gotten reader reports of problems with a variety of hardware, not just M-Audio and Digidesign. Problems don’t seem to affect everyone, but then, most bugs affect only some users, not all.

Consensus from every vendor I’ve talked to — software and hardware — is that an OS-level change in 10.5.2 caused problems. The likelihood is, Apple will have to resolve those issues. So it’s not worth getting angry at your device vendor, because it’s almost certainly not their fault. Likewise, I’m not sure it’s worth getting angry at Apple — 10.5.2 just didn’t work as expected, and the best we can do is to find a temporary workaround and wait for the next update. You can get angry if you want, of course. It just won’t make any difference. (I have about a 20-year history at this point of yelling at computers; I find it at least lets off steam.)

But let’s talk about accountability, since various readers are bringing it up. At the risk of pointing out the obvious, I think accountability has to involve both music and audio vendors and OS vendors.

As I see it, the basic responsibility of hardware vendors is to test their current hardware on new OSes, to release complete documentation in terms of what works and what doesn’t, and to update their drivers to support the current driver model and specifications of a new OSes in a reasonable amount of time. By this measure, there’s room for improvement at many vendors, and it’s my belief that improvement in these areas will mean more and happier customers.

But OS vendors have responsibilities, too. And since this isn’t 1985 or 1990, that means basically Microsoft and Apple. (The Linux community is a different animal, so I’ll leave them alone.) A creator of a device driver can’t possibly be responsible or accountable for fixing OS-level performance issues with scope beyond their own hardware. And that’s been, unfortunately, the situation with Windows Vista, particularly in its first 8 months, and now Mac OS X Leopard in its first 8 months. Problems haven’t affected everyone, but they have been widespread enough that we have a right to be disappointed. As a user, I think I have the right to be disappointed. As a writer, it’s my obligation to point it out. And I hope they do better in the future.

10.5.3: Help is on the way?

I’m equally obligated to watch for these issues to be fixed. In the case of Apple, the rumor mill suggests that a fix is imminent:

Mac OS X 10.5.3 about to hit [APC Mag]

This is not official information, as Apple doesn’t make their OS update process public. Technically, I shouldn’t say that 10.5.3 is rumored to have audio and video applications as a focus. I also shouldn’t say that the major change in the current seed is reported to be “kernel performance.” Of course, that’s exactly what I and many other parties had predicted: anyone running 10.5.2 who is experiencing this issue already knows there’s a problem. The good news is, a fix is likely on the way.

Look, updating and testing operating systems isn’t easy, and my job is not to sit here and be an armchair quarterback. What I can do, though, is do as much homework as possible, and try to use the large community we have here on CDM to provide as much information as we can to help you make informed decisions. I hope that 10.5.3 is an OS we can recommend, as 10.5.2 was not. So, we’ll keep watching.

If you have 10.5.2, have this symptom, and for some reason can’t downgrade, do let us know when an update becomes available and what your experience is.

  • Blame blame blame and shame shame shame on Apple!

    Anyway, I'm looking forward to trying 10.5.3

    Any news about Tiger's nasty Airport-related issue? They're soooooo slow on fixing issues!

  • Actually, slow is okay by me, given that in each case this is a fix for something else that created *new* problems — I say take your time!

    Tiger's Airport issue — nope, no word yet, though you can bet I'm watching. Best bet is to downgrade Airport support or turn the thing off if you're having an issue. At least it's a lot less nasty than 10.5.2.

  • Peter – I am not sure why you exclude Linux from the discussion, because precisely the same issues exist there. Do users have a right to expect an OS kernel to provide for the needs of realtime/media applications? Do they have a right not to see the performance of those apps suffer because of changes made to benefit other classes of applications? The one interesting thing about Linux in this arena is that at present, there is no claim that you can "do it all" with a single OS kernel – people who want absolute minimal audio latency use ones built precisely for that purpose. In some ways this makes the Linux situation better – the people who build those kernels for users know just what to do and are focused on the use cases faced by those users. But in other ways it makes it worse since there is no single OS kernel that anyone can convincingly talk about. We still face issues in our world with things like Linus' recent announcement that he was not happy with the direction that work on steadily merging more and more of the low-latency changes for Linux were going in, forcing some very hard choices on kernel developers trying to cater to the realtime/media crowd.

  • I'm not excluding Linux from the performance questions — I just think you can't make a direct comparison in accountability, because there's a community responsible for development, and (ideally) open-source drivers, etc. And that's a good thing, of course, as illustrated by everything that's going wrong with Windows and Mac OS at the moment.

    That said, I think you're absolutely right — it's interesting to watch exactly the SAME debate play out on Linux as far as kernel and performance. (And, in fact, a lot of the issues on both Vista and Leopard are kernel-related.)

    Personally, I think the "realtime/media" crowd is going to pretty soon translate to "everyone." That doesn't make the problem any *easier*, though (well, actually, it makes it harder).

    But, Paul, I'd very much like to know how you feel about the evolutions on Linux. I can also say that frustrations on Mac OS and Windows are causing developers — and, heck, ME — to look at Linux differently. For music users, at least, real-time kernels are a huge draw for Linux. I find myself more drawn to Linux's capabilities here in 2008 than ever before.

  • Well, I can see 2 or 3 places to comment.

    The big picture is that Linux is missing only a couple of things that stop it from being at least as good as other OS's for the sort of activities discussed here on CDM:

    1) missing native plugins

    2) some audio interface devices have no drivers

    The second point is becoming less true every day. Some of this is partly thanks to the ever increasing adherence to the USB audio class specifications, which means that a "driver-less" device that works on Windows and OS X also "just works" on Linux. The other reason is the hard work being put in by Pieter Palmers and a very few others at the FFADO project to provide firewire drivers for devices based on DICE-II and BeBob ieee1394 chipsets (i.e. everything).

    Plugins are a different story, and I won't comment more on them here other than to note that its really the only thing that I hear over and over from existing audio software users about why they cannot switch to Linux. Plugin developers: you are in control!

    The kernel itself is a separate story. The situation today is that anyone who wants to use Linux for media work needs to start by understanding that there is not "one Linux" – unlike Windows and OS X, there are many versions that are tailored to meet the needs of different users. Commercial/enterprise systems such as Red Hat are primarily aimed at people with thoroughly corporate needs, and these are often somewhat at odds (though not as much as you might think) with the needs of a recording engineer or composing musician.

    However, the emergence of excellent "overlays" like UbuntuStudio (on top of Ubuntu), 64Studio (on top of Debian), PlanetCCRMA (on top of Fedora), JAD (on top of SuSE) and so on now removes the excuse that "you can't get this stuff to work out of the box", because you just can. Each one of these "overlays" (or in some cases, whole distributions, or both), offers a kernel built especially for low-latency/realtime work, along with many other specially considered features and details aimed at media work. UbuntuStudio in particular has done an amazing job of making it possible to install Linux very quickly and be making noise (or maybe even music) with minimum hassle. Its a bit more than firing up an old Amiga, but given that the box will happily let you browse the web, read email, and compile your own DAW while running audio software with 2-10msec latency and no glitches, I'd say that the small bit of extra setup is a reasonable tradeoff.

    The kernel itself is evolving, as ever. For years, Ingo Molnar (employed by Red Hat) has been the most visible part of the a group of developers adding better and better realtime support. This basically means fixing the kernel design and specific sections of its code so that nothing can get in the way of a realtime process. Recent versions of these kernels have performance similar to so-called hard-RT OS's, typically measuring the time from an interrupt being raised to when a thread in an application that is "related" to the interrupt (e.g. your DAW or synth or whatever) is running again in the range of 5-20 microseconds. This just keeps getting better and better, and its certainly at least as good, if not better, than current OS X. Windows is significantly behind both of them.

    Recently, however, Linus announced that he wasn't thrilled by the direction Ingo and others had taken in their next steps, and announced a code reversion of one of the most recent "add more RT to mainstream" changes. This has forced the hand of the RT kernel guys. They have been working on steadily merging more and more of their code into the mainstream kernel (good for users who mistakenly think that any old Linux distro is the same as any other). Linus' decision means that either they (a) solve some very hard and distributed problems throughout the kernel that affect latency and/or (b) give up on continuing their merging efforts and focus just on the RT version of the kernel. Signs are good so far that they will do (a).

  • Thanks for that, Paul. Before this becomes an "is Linux ready for the desktop?" discussion, here's the point that to me is relevant here: forgetting about Linux for a moment, what we've seen on the Mac OS and Windows is that *lack* of control over the OS and kernel can be enormously costly to the industry. Why is that we're able to levy all sorts of criticisms of Linux (okay, the Linuxes as you rightfully point out), without examining the cost of poor real-time support in mainstream OSes?

    The other thing that's apparent from your discussion, and from my own anecdotal experience, is that it's absolutely possible to have an OS that performs mainstream tasks that's also perfectly appropriate to real-time auido work. That is, not all mainstream OSes are ideally optimized for real-time media, but all real-time media-optimized OSes can be used for other things, too.

    From an end user perspective, I know the real deal-killer on Linux is the lack of additional software — not only plug-ins, but also hosts (Ardour is great, but people are pretty personal about their choice of hosts). But without getting into that discussion, I think there really is an advantage to fixing real-time kernel performance issues. If it turns out that that's not really a priority for Microsoft and Apple, I think it's worth reevaluating Linux. It can't be all bad — it is the OS that ultimately launched digital vinyl DJing. 🙂

  • Jon

    I'm so happy I'm still on 10.4

  • ericdano

    It's funny that Metric Halo, MOTU, and other vendors don't have any issues with 10.5.2. And if they did, they fixed them in a timely manner.

    M-Audio/Digi………it is really lame they are consistently behind. There really is no excuse.

  • ernesto (costa rica)

    ericnado: i think you are right in that some vendor offer better support than others, but what I've lived and suffered with 10.5.2, running an otherwise top level computer, makes me to believe that there is a problem with the OS that, actually, goes beyond audio.

  • ernesto (costa rica)

    sorry for the typo!

  • spinner

    10.5.2 does seem to cause problems for audio but it has allegedly sorted out a lot general consumption issues (Don't run PT & it's a bloody stable OS). It's noteworthy that newish apps like Live & Reaper are running trouble free. PT, Logic

    & and others are hauling around what, 15 year old code…. that must have an impact on stability.

    The Digi Core driver has always been dodgy I just hope now that they've taken so long they've sorted it once & for all.

  • @Eric: We're conflating two issues here.

    I had complained about availability for M-Audio/Digidesign drivers for Leopard, Vista, and 64-bit Windows.

    The other issue is that 10.5.2 has caused other issues with audio on Apple laptops. It is not recommended by Serato and Native Instruments, for instance, both of whom are *software* vendors, not hardware. I just got a specific report in regards to MOTU hardware. There are a lot of variables, and not everyone is having issues (which appears to be dependent on which chipset their laptop is on).

    It sounds as though Apple changed something relating to kernel-level performance. The "age" of people's code isn't the key variable here. (Also, even that's misleading; product lines regularly get code refreshes, particularly with things like the Intel switch.)

    Now, whether we want to criticize Pro Tools, that's a separate discussion. But the OS on the Mac I currently would recommend is 10.4.11 (minus the most recent AirPort update). If 10.5.3 resolves these outstanding issues, then I would feel able to recommend Leopard.

  • mark

    I have had nothing but trouble with my m-Audio fast track pro. BUT HERE'S THE KEY QUESTION: wouldn't upgrading to 10.5.3 put Mi-Audio and other vendors TWO updates behind? For those of us who can't downgrade, wouldn't it be best to wait until the vendors first catch up to 10.5.2 before installing 10.5.3? Or does everyone think Apple will solve the problem?

  • @Mark: no, my sense from M-Audio and Digidesign was, explicitly, you can expect them to target a fixed update, not 10.5.2. The major changes to the driver model were in 10.5; these are bug fix releases from Apple.

  • Richie Allen

    AND HERE IT IS!!!!! !0.5.3 is NOW OUT! Pretty hefty update, too. ~420MB for my C2D MBP.

  • Mark

    Sorry Peter, I didn't quite understand… I spoke with M-Audio this morning, after the update, and they of course offered no insight. What is your advice? Should folks try and download 10.5.3, and hope this fixes our problems? Or should we wait for the vendors, which could be forever… Anybody download 10.5.3 and find your audio is happy again?

  • Mark, I'd give it a shot. M-Audio is behind on their Leopard drivers, so 10.5.3 may still not work, but it's worth a try. I think the next thing that will happen is, if 10.5.3 *is* more stable, it'll encourage more action on the developer side.

  • contakt321

    I upgraded, sadly my PreSonus FireBox still isn't being recognized even though it's sync-ing. I am tempted to just sell this thing. I havent been able to get it to work for the month I have had it and don't have the time nor patience for tech support.

    That being said – PreSonus' tech support via email has been great. I just want a working interface. Apple is bumming me out.

  • I recently bought a macbook that came with OSX 10.5.2 pre-installed. I have a M-Audio Fast Track Pro soundcard and I really didn't have any problems running Live, Reason or Audacity. Anyway, I've just upgraded to 10.5.3 and everything is alright so far. The video performance seems a little bit snappier, which is good. Audio is as good as before. I'll keep you guys posted if anything weird happens.

  • mark

    Drats. 10.5.3 does not fix my problem with the M-Audio Fast Track Pro. I can't even use iTunes, as a loud digital sound threatens to blow out my speakers and amps. Okay, now I'm not going to blame Apple anymore, but M-Audio, I'm looking at you…

  • Stefan

    I'm having problems with my fast track pro. I already had it with 10.5.2. It's some kind of distortion coming out of the fasttrack. It isn't my monitors because the distortion is also coming from the headphone output. I'm not having the problems always, sometimes when I restart or switch off and put back on the fasttrack the problems are gone. I recently installed the 10.5.3 update, still having the same problem. Really don't know what to do….


  • Still seeing no issues. Using a Presonus Firestudio on 10.5.3 Mac pro 2.66 (First Gen) 4 gigs of ram. M-Audio has been pretty unreliable, anyone using a Firewire 410 with issues?

  • Pingback: Create Digital Music » Adieu, XP; How Vista SP1 is Doing, and Why This OS Generation Has Been So Tough()

  • Mr. X

    It’s funny reading this old article when brwosing thru the web searching for compatibility issues with the new OSX 10.9 Mavericks. We couldn’t get any of the class compliant stuff up running. Seems like there’s a bug in the OS. And all the customers are complaining “why your stuff doesnt work” at our hotline and when we try to explain “it worked with previous versions, so it’s not our fault when it’s not working no more.” the customers are gettin more angry because they know that moaning at a apple hotline has exactly ZERO effect…

  • Mr. X

    It’s funny reading this old article when brwosing thru the web searching for compatibility issues with the new OSX 10.9 Mavericks. We couldn’t get any of the class compliant stuff up running. Seems like there’s a bug in the OS. And all the customers are complaining “why your stuff doesnt work” at our hotline and when we try to explain “it worked with previous versions, so it’s not our fault when it’s not working no more.” the customers are gettin more angry because they know that moaning at a apple hotline has exactly ZERO effect…

  • Mr. X

    It’s funny reading this old article when brwosing thru the web searching for compatibility issues with the new OSX 10.9 Mavericks. We couldn’t get any of the class compliant stuff up running. Seems like there’s a bug in the OS. And all the customers are complaining “why your stuff doesnt work” at our hotline and when we try to explain “it worked with previous versions, so it’s not our fault when it’s not working no more.” the customers are gettin more angry because they know that moaning at a apple hotline has exactly ZERO effect…