Sudden volume changes in DP6.01

[nadeama]

Hi,

I've had this problem since DP 5, and have done extensive testing on it, both last year and about when DP 6 came out. Please see this thread for more detail on that:

http://www.motunation.com/forum/viewtop ... 12#p250669

Here is what I know for sure about these problems (volume outbursts and MIDI bombs, which I feel are related because they appeared at exactly the same time for me). I know that some people will say that I can't be sure of anything, since all our systems are different, etc., but I feel I at least have had as much experience with these issues as anybody; I've tested them on 2 different computers (G5 and Mac Pro), on 2 different OS (Tiger and Leopard), and on 2 different major versions of DP (DP 5.x and DP 6.0 - 6.01). We have to be able to at least base our discussions on something, so let's pretend that the following is pretty solid:

- Neither of these problems has anything to do with VIs. This seems to be the most general misconception about this issue. I've had it in projects that contained only external MIDI assignments, projects that contained only internal VIs, and projects that contained both. So at least some form of the problem is indifferent to which kind of MIDI assignments you have.

- Although I've never seen this problem in projects that contained only one MIDI track (like a solo piano), I've experienced at least the volume outbursts in projects containing only 3 or 4 MIDI tracks. So, you really don't need to have a project with dozens of MIDI tracks to get the issue (at least the volume outburst issue).

- The MIDI bomb issue can be so severe as to crash all my PC Giga slaves at once (I guess they just can't handle the big burst of MIDI data being sent to them all of a sudden). Thoses PCs are rock solid otherwise. Not good...

- I would contend that these issues are quite simply bugs introduced in DP 5, because I've never had them before, and I can play my test sequence (see link above) with no problem whatsoever in DP 4.61, but I get the issues both in DP 5 and 6. Except that.... some people say that they've started experiencing this as early as DP 4.x. This has been and remains the biggest puzzle for me. Why didn't we all start having these problems with the same version? I can't explain this.

- Inserting CC7 at the beginning of a track doesn't help (at least not in my case). I have numerous projects containing several sequences where all MIDI tracks have CC 7 inserted and automation turned on, and I still get the issues.

- MOTU has said that they are aware of the problem, but can't reproduce it. Over a year ago I offered to send them my computer (the problem was reproducible with no other hardware attached) so they could have a machine on which to reproduce the bug(s). They refused (full disclosure: I had asked them to pay for shipping). During the year, I received 2 messages to my technote saying simply "resolved". No other detail even though I asked for clarifications. DP 6 has now come out, and obviously nothing is "resolved".

So, that's what I know about this. Sorry to repeat things I've already said in previous threads. but all these threads tend to get lost and buried, so I thought I should state again what I've learned after numerous tests.

The last time I posted about this, I said that I would call up MOTU and offer again to send them my computer, but this time I'll offer to drive there in person and show them the problems on my test sequence. Unfortunately, I got busy with gigs and didn't feel confident as to when exactly I would be able to let go my computer for a couple of weeks. So I haven't called them yet.

However....

I'm tired of this s h i t. A lot of people have been having these issues for a long time now, and I find it unacceptable that it's still not fixed (heck, they haven't even reproduced it!). So, I'm going to call them up tomorrow. If they accept my offer and want my computer, I'll just rent another machine while they have mine or something. I dunno.

I'm afraid I'm going to have to put the poor tech support guy on the spot a little bit, because in my case, he either tells me tomorrow that they're serious in wanting to fix this and thus want to take a look at my computer, or I switch to Logic. We'll see what they say...

[Frodo]

- Neither of these problems has anything to do with VIs. This seems to be the most general misconception about this issue.

It's not so much a misconception, per se. It could be that the issue manifests itself easily or more frequently across the user base with VIs because they are used more these days than external MIDI modules. I will concede that the issue must be *triggered* by MIDI (afaik), whether VIs or external modules.

I've got one friend who runs a studio and uses DP. He says that of all the things he's had to deal with that volume surges are not one of them. He doesn't do any VIs or MIDI at all-- just audio and fx. That's only one case of an audio-only person that I could find. But these threads are about gathering stats. If there's some audio-only person out there experiencing surges, I hope they chime in with their symptoms.

- I would contend that these issues are quite simply bugs introduced in DP 5, because I've never had them before, and I can play my test sequence (see link above) with no problem whatsoever in DP 4.61, but I get the issues both in DP 5 and 6. Except that.... some people say that they've started experiencing this as early as DP 4.x. This has been and remains the biggest puzzle for me. Why didn't we all start having these problems with the same version? I can't explain this.

This is my stumper as well. Could it have been some OSX firmware update that might be getting "in the way"? I just don't know. I didn't see it until 5.0 on my G5. That's a lot of updates ago.

- Inserting CC7 at the beginning of a track doesn't help (at least not in my case). I have numerous projects containing several sequences where all MIDI tracks have CC 7 inserted and automation turned on, and I still get the issues.

I can concur. Automation on or off-- surge-o-rama.

- MOTU has said that they are aware of the problem, but can't reproduce it.

I'd have to say you are quite accurate in that assessment. MOTU has certainly had plenty of reports from users, many of whom have received the same "can't *reproduce* the problem" response. Not only were some of my tech reports marked as "resolved", some of them were deleted entirely after months without a response.

With that said, I do know of one person who works for MOTU who is now going the extra mile to pursue the surge issue specifically. They may not be able to reproduce the issue (yet), but I'm encouraged to know that someone there genuinely cares.

Over a year ago I offered to send them my computer (the problem was reproducible with no other hardware attached) so they could have a machine on which to reproduce the bug(s). They refused (full disclosure: I had asked them to pay for shipping). During the year, I received 2 messages to my technote saying simply "resolved". No other detail even though I asked for clarifications. DP 6 has now come out, and obviously nothing is "resolved".

Not surprised. However, if you have a project that exhibits the surge behavior, zip the project without audio (if any) and send it to MOTU along with your System Profile and a crash report (if any) that may have occurred.

So, that's what I know about this. Sorry to repeat things I've already said in previous threads. but all these threads tend to get lost and buried, so I thought I should state again what I've learned after numerous tests.

Nadeama, you've done well to add your notes to the bundle. They are intelligent notes indeed, and in many ways inspiring for the depth and extent to which you've gone to examine the problem. I can only welcome you to the "headbanger's ball" where surge victims are wondering on their own if they've lost their minds by blaming themselves for user error.

I don't know if this is related or not, but I was working on a project tonight for about 8 hours. For the most part I was doing okay-- no VIs, just 4 stereo audio tracks. I solo'd one of the tracks and all was well. But when I MUTED the other three tracks, I heard them through the solo'd track. Very odd. This was in DP 6.01 and 10.5.5. I began to wonder about that dreaded audio leak issue that I've associated with this general surge issue. Hmm.

I'm tired of this s h i t. A lot of people have been having these issues for a long time now, and I find it unacceptable that it's still not fixed (heck, they haven't even reproduced it!). So, I'm going to call them up tomorrow. If they accept my offer and want my computer, I'll just rent another machine while they have mine or something. I dunno.

I'm afraid I'm going to have to put the poor tech support guy on the spot a little bit, because in my case, he either tells me tomorrow that they're serious in wanting to fix this and thus want to take a look at my computer, or I switch to Logic. We'll see what they say...

Fatigue. I know the feeling all too well.

As far as Logic goes, the grass is not always greener-- or even the same shade of green-- on the other side. Logic has a whole different set of issues a world apart from DP. I'd spent half of today in Logic and wound up porting everything back into DP because some things just didn't make sense. There *are* some nice features, but I got tired of jumping through hoops and spinning plates on sticks.

On another note, Nadeama, I owe you big time on a test that you asked for a while back. I was traveling internationally at the time, and I think 6.01 was released right after you posted your test request. I've not forgotten your request, but I want you to know that I consider such tests very important when a user needs confirmation. Please accept my apologies for not following through. I've got a rough mix to finish tomorrow and a whole CD's worth of roughs to send out before the end of the day Tuesday, but I will delve into your test once I can get these rough mixes out of the way.

Peace,
Frodo

[dogBoy]

I just want to know if we can eliminate an external controller from the mix of possible culprits.
Insightful posts on this subject, I like the cache build up idea.
nadeama, you really put your stuff on the line by sending motu your computer, thanks

[michkhol]

Hi,

I've had this problem since DP 5, and have done extensive testing on it, both last year and about when DP 6 came out. Please see this thread for more detail on that:

http://www.motunation.com/forum/viewtop ... 12#p250669

I have DP5 and I'm willing to run this test. Is it possible to publish it in the DP5 format?

I'm afraid I'm going to have to put the poor tech support guy on the spot a little bit, because in my case, he either tells me tomorrow that they're serious in wanting to fix this and thus want to take a look at my computer, or I switch to Logic. We'll see what they say...

One of major reasons I use DP vs Logic is that Logic has only 128-point external resolution on all its faders (they say it's more with a control surface). Not good for mixing at all. It seems like I have no choice.

[nadeama]

Alright, so I just got off the phone with a MOTU tech support representative, a nice guy named Joe.

Basically, here's how it goes: he's not aware of this issue, he wants me to recreate my default template (in case I have a corrupt template, Ha!), and wants me to send him my test sequence.

Same crap I went through last year. It's not my template, because I create new templates from scratch every time I upgrade DP (I only use a default template and it's very basic). I also created a new default template from scratch on the OS X partition that I recently created just for testing this problem. So it's not my template.

Hopefully, I might get lucky and he might be able to reproduce the problem with my test sequence. I'm not holding my breath though, since I sent them that same test sequence last year and they were not able to reproduce the issue then.

Still, I'm going to do what he asks just for the heck of it.

I explained to him that I've been through all this before, that I knew what will happen is that he'll eventually (probably after several weeks) tell me he can't reproduce the issue. I told him that I had offered to send them my computer last year, and that now I was ready to make the drive and bring them the machine myself, and leave it with them for a while. His answer was: "oh, that won't be necessary...".

So after that, I said that I just wanted to make it clear that I would do what he asked, but that if we couldn't resolve the issue this time around, I was switching DAW. He said "very well". I don't think he liked me very much.

So here's the bottom line as I see it: they just don't care. There could be one thousand of us all calling on the same day about this, and they'd dance around the issue and tell us that they're not aware of the problem, or that they are aware but can't reproduce it but hey, they really don't need my machine that would help them reproduce this.

I have a very cynical view of what I think might be going on with this, but since I don't have any proof I won't share it here. But I think you can guess what I'm thinking...

Here's something funny: when I called, I was told that it must be my template, then I explained that I'd been told last year that MOTU was aware of the problem, and Joe asked me the number of the technote where this was said, and then put me on hold while he was checking this out. Now, this technote has only a few lines and Joe was gone for like 6-7 minutes, maybe more. I guess I have a vivid imagination, but I can easily see my man Joe asking around the office "hey guys, are you aware of some MIDI outburst/MIDI bomb issue?" and being told by some of his co-workers "Ah crap, not the f&%*ing MIDI outburst/Bomb issue, fill in the blank", and then Joe coming back with "Yeah, it must be your template".

But seriously guys, we're never going to see this fixed, I think. Frodo, if you do indeed have a good contact inside MOTU, push him real hard about this, cause that's probably our only chance. As for me, I'll do what Joe asked, but in the mean time I'm going to make a little trip to my reseller and get a fresh copy of Logic. I am indeed afraid that it will have its own issues, and the prospect of learning a new DAW does not exactly enchant me, but honestly, it can't be worse than this....

[nadeama]

Hi,

I've had this problem since DP 5, and have done extensive testing on it, both last year and about when DP 6 came out. Please see this thread for more detail on that:

http://www.motunation.com/forum/viewtop ... 12#p250669

I have DP5 and I'm willing to run this test. Is it possible to publish it in the DP5 format?

The test file from the post linked in my previous message is now in DP 5.12 format, so that DP 5 users can test it if they want.

I've tried it earlier today before calling MOTU, and although the volume outburst problem was difficult to reproduce for some reason (I had been able to reproduce it relatively easily previously using this same sequence), the MIDI bomb issue should be fairly easy to replicate, if you follow my instructions from the thread linked above.

Thanks for your interest in this!

[magicd]

Frodo, if you do indeed have a good contact inside MOTU....

Martin,
Check your PM.

Dave

[nadeama]

I've got one friend who runs a studio and uses DP. He says that of all the things he's had to deal with that volume surges are not one of them. He doesn't do any VIs or MIDI at all-- just audio and fx. That's only one case of an audio-only person that I could find. But these threads are about gathering stats. If there's some audio-only person out there experiencing surges, I hope they chime in with their symptoms.

I don't work with audio as much as some other users, because the bulk of my work is with MIDI, but I still do dable in it quite a bit. I've never, ever had a volume outburst with audio tracks. This, in my own experience, seems to be truly a MIDI problem.

- I would contend that these issues are quite simply bugs introduced in DP 5, because I've never had them before, and I can play my test sequence (see link above) with no problem whatsoever in DP 4.61, but I get the issues both in DP 5 and 6. Except that.... some people say that they've started experiencing this as early as DP 4.x. This has been and remains the biggest puzzle for me. Why didn't we all start having these problems with the same version? I can't explain this.

This is my stumper as well. Could it have been some OSX firmware update that might be getting "in the way"? I just don't know. I didn't see it until 5.0 on my G5. That's a lot of updates ago.

- Inserting CC7 at the beginning of a track doesn't help (at least not in my case). I have numerous projects containing several sequences where all MIDI tracks have CC 7 inserted and automation turned on, and I still get the issues.

Exactly, a lot of updates ago. And because of just that, I've been a bit worried about users maybe not remembering correctly when these issues all started. Someone could think he remembers the problems starting with 4.5, when in fact they may have started with 5. I'm not saying that's the case, but there's a possibility there. In my case, I'm 100% certain that it started with 5.0, because I've tried the exact same sequence consecutively on 4.61 and 5.x, and only version 5 exhibited the problems.

With that said, I do know of one person who works for MOTU who is now going the extra mile to pursue the surge issue specifically. They may not be able to reproduce the issue (yet), but I'm encouraged to know that someone there genuinely cares.

If this is someone that you have personal contact with, it may very well be our only hope (please see previous post about my phone call to MOTU). If you talk to him, please feel free to mention my offer of leaving my computer with them for testing, if needed. It still stands, as long as I can find someone up there who is interested in looking at it.

I don't know if this is related or not, but I was working on a project tonight for about 8 hours. For the most part I was doing okay-- no VIs, just 4 stereo audio tracks. I solo'd one of the tracks and all was well. But when I MUTED the other three tracks, I heard them through the solo'd track. Very odd. This was in DP 6.01 and 10.5.5. I began to wonder about that dreaded audio leak issue that I've associated with this general surge issue. Hmm.

I have also had a similar issue with MIDI tracks, where a single track would be soloed and I would intermittently hear parts of other MIDI tracks, but I've never personally experienced this with audio tracks. Mmmmm..... very interesting indeed.

As far as Logic goes, the grass is not always greener-- or even the same shade of green-- on the other side. Logic has a whole different set of issues a world apart from DP. I'd spent half of today in Logic and wound up porting everything back into DP because some things just didn't make sense. There *are* some nice features, but I got tired of jumping through hoops and spinning plates on sticks.

Yeah, I'm quite worried about this. The truth is that I don't want to leave DP, but it's like MOTU are almost forcing me to. I really don't want to have to learn and get used to another DAW, but these issues (especially MIDI bombs) are so severe for me that I can't just "live with them". I have to find a solution.

On another note, Nadeama, I owe you big time on a test that you asked for a while back. I was traveling internationally at the time, and I think 6.01 was released right after you posted your test request. I've not forgotten your request, but I want you to know that I consider such tests very important when a user needs confirmation. Please accept my apologies for not following through. I've got a rough mix to finish tomorrow and a whole CD's worth of roughs to send out before the end of the day Tuesday, but I will delve into your test once I can get these rough mixes out of the way.

Actually, that's exactly on the same note, because that test was all about volume outburst/MIDI bomb issues.

No problem at all, I just figured you were very busy. If you do indeed find the time to run the test sequence, just make sure that you also read my original post (linked in a previous message above) for the steps to reproduce the problems. The volume outburst issue was more difficult to reproduce this afternoon (no idea why), but the MIDI bombs were easily triggered.

[HCMarkus]

Thank you for your efforts Martin. Your work is appreciated. For all our sakes, I hope Magic Dave and MOTU come thru for you.

[michkhol]

I've tried it earlier today before calling MOTU, and although the volume outburst problem was difficult to reproduce for some reason (I had been able to reproduce it relatively easily previously using this same sequence), the MIDI bomb issue should be fairly easy to replicate, if you follow my instructions from the thread linked above.

Thanks for your interest in this!

Running the test right now.

I have not been able to reproduce the volume bursts yet, but MIDI bomb is there. On my system after I hit "1" on the keypad, there is no sound until the wiper reaches the point (approximately) where I hit the "1" button and it spurts MIDI out and then continues normally. Sometimes it happens earlier. No such thing happens if I press stop first. The CPU usage (I use the command line "top" utility) drops to the half of the play value when there is no sound which probably means that at least half of the DP threads are blocked. I can guess how it happens but certainly I don't know why. I did a good share of multithreaded programming myself and I can assure you that it is a debugging nightmare if the design is not right. Here is the explanation: at some point the threads get out of sync. DP has very little control over which thread runs and when, namely it can only require that a certain piece of code would be run by a single thread at a time, that's it. But here is a tradeoff. While the thread is running this piece of code, all other threads should wait to run the same piece of code. It slows down the whole thing, so usually programmers try to avoid that sometimes at the risk of threads clashing or blocking each other. Now back to the MIDI bomb. It looks like that there is a buffer (or several buffers) which is fed by several threads with MIDI data. The buffer(s) has a flag, if it is reset, it is time to fill up the buffer. Another thread is responsible for reading data from the buffer(s). When it is empty, the flag is reset, it signals all other threads to fill it up again. Now imagine that that thread does not reset the flag in time. The buffer is empty, there is no sound, other threads think that the buffer(s) still has data and continue reading the MIDI tracks into their own buffers, which become overflown. Finally when the flag is reset, the sync is already lost, the buffer(s) get some random data and you get the MIDI bomb.
The audio volume burst mechanism is similar. All audio data must be summed synchronously in a summing buffer before it goes out. Now imagine that several threads supply the data and a dedicated thread does actual summing and sending data out. For each data frame the buffer must be zeroed out first, all samples corresponding to that frame must be added up and send out, than everything is repeated for the next frame. Audio burst happens when the data for the next frame comes before the previous frame is sent out, i.e. the synchronization between threads is lost. In theory it is possible to force the standard synchronization but it eats the precious CPU resources especially in such a critical piece of code. So most probably (and this is my wild guess) the synchronization is done by other more risky but more CPU-friendly ways but sometimes it breaks. The problem is that it is practically impossible to find the exact point where it breaks if you have so many threads running. The only way is to change the code design. The problem lies obviously within MAS and I suppose this is the piece of code too critical for MOTU to touch unless the whole redesign. It does not seem like DP6 changed anything in MAS implementation.

If I'm right, there is no 100% workaround, because there is no external control over the thread timing. One might try to change the timing by running something along, like Safari with flash-heavy web pages.

[nadeama]

I've tried it earlier today before calling MOTU, and although the volume outburst problem was difficult to reproduce for some reason (I had been able to reproduce it relatively easily previously using this same sequence), the MIDI bomb issue should be fairly easy to replicate, if you follow my instructions from the thread linked above.

Thanks for your interest in this!

Running the test right now.

Snip fascinating but long explanation

Michkhol,

Thank you for running the test and confirming that I'm not totally insane by reproducing the MIDI bombs! The volume outburst is indeed more difficult to reproduce. I have seen it in this same sequence but it happens less often, and today I was able to reproduce it only once in several minutes of testing. So it's not too surprising that you didn't experience it.

As for the MIDI bombs, I'm not 100% certain, but I think it can also happen even if you press stop before rewinding. I should really try this again to make sure, but I'm 99% certain that this has happened in my testing.

Thank you for the explanation on what may be happening. I can't honestly say that I understand every single detail, but I get the general idea. It seems like it might indeed be a nightmare to fix for the poor engineers.

What's interesting is that at some point the design was right, because I've never had those problems in DP 4.61. I tried the same test sequence on the same computer, with the same OS, and DP 4.61 runs it perfectly, while DP 5 exhibits the symptoms. This has to mean that something was changed in DP 5 (and the change maintained in 6). I can't understand how come some people started having these issues in DP 4.x. It makes no sense to me.

Again, thanks for running the test sequence.

Oh and.... for anyone else running the sequence, it's atonal music for a tense scene in a film. It really sounds better with strings. Honest...

[nadeama]

Frodo, if you do indeed have a good contact inside MOTU....

Martin,
Check your PM.

Dave

Hi Dave,

Thank you for your message. I saw your post only after replying to some others. I'll be sending an email your way shortly.

It seems that all hope may not be lost yet...

[Frodo]

Frodo, if you do indeed have a good contact inside MOTU....

Martin,
Check your PM.

Dave

Hi Dave,

Thank you for your message. I saw your post only after replying to some others. I'll be sending an email your way shortly.

It seems that all hope may not be lost yet...

"There's always hope." ~ Aragorn

[Tomas E]

What is actually the meaning of "being able to reproduce the problem"? Could it be that MOTU themselves have this problem but they haven't found a way to reproduce their own bursts? Just a thought, in that case they are not lying about it...

[tonwurm]

I´m able to reproduce (at least) the MIDI bombs with nadeama´s test sequence

every 2nd or 3rd hit of the "play from beginning" command results in playing those bombs. the delay where they occur is everytime a little different, sometimes earlier - after 2 or 3 seconds, sometimes after 5-6 sec.


made another test with hitting stop before hitting the play from beginning command. it then seens to occur only after every 6th or 7th hit, but it occurs