Sonar, latency, and dropouts

[ljc]

I'm running Sonar 2.2, an AMD 2000xp, a DMX 6fire 24/96 card and a motu midi express 8X8. I sync a digital board and a synth to the motu so that Sonar,board, and Synth are midi slaves to motu ( Sonar actually delivers mmc, but motu is clock ). I'm able to record clean audio on my board and even send to Sonar for archive purpose through the audio card. At the end of the day I have a Sonar project with several audio tracks and several midi tracks. During record of either midi or audio, I sometimes get dropoutss, where Sonar just stops everything. the audio is clear so I don't hear any latency problems ( card is at 7ms ), and I've turned off any background processing on the pc. I guess my question here is, how does one begin to troubleshoot this problem and what are the key ingredients that can cause a dropout ?

[moskus]

Originally posted by ljc
the audio is clear so I don't hear any latency problems ( card is at 7ms ), and I've turned off any background processing on the pc. I guess my question here is, how does one begin to troubleshoot this problem and what are the key ingredients that can cause a dropout ?

If you don't use Input Monitoring or any DXi's you'd be better off just turing the latency up a bit, say 50 ms. Sonar compensates for latency during normal playback and recording, so there's no problem with high latency unless you need realtime processing...

[dachay2tnr]

Originally posted by ljc
what are the key ingredients that can cause a dropout ?

Audio Card
Audio Card Driver
Buffers and Latency Settings
Processor speed
Amount of Memory
Hard Drive (speed of drive, dma enabled, separate audio drive)
Overhead on Processor (# of tracks, # of realtime effects, type of effects)
Overhead on Processor (programs running in background)
There's probably others.

Assuming you have already optimised your computer for audio work, then as moskus said, the easiest one to deal with is latency. You can simply raise the latency setting (assuming nothing else you are doing will suffer from the higher latency).

Are you having fun yet?

[ljc]

I have not yet attempted to monitor while recording audio in Sonar, but I've heard that this is where latency is a factor ( delay in the playback ). Is it common to set latency based on how you are recording. So during a session where I monitor I also need to set latency lower than 50ms. Of course if this is the case, here come the drop outs ? I guess what I'm getting at is that the coupling between different number of tracks, cpu usage etc. is not defined well and seems to be up to the user to tweak and figure out. A bit disturbing, because this also means that couplings can change based on the Sonar project and requirements ( monitoring etc. ). I bought Sonar second hand from a guy who was unhappy with the midi functionality when taxed with several synths and audio running . He later found similar problems using cubase ( maybe its him ), but my point here is that it seems to be a bit of an art to midi sync all of these machines and drive audio, Anybody know of a really good white paper on this topic ?

[tntkemp]

Depending on your sound card, the latency you are talking about is related to monitoring through Sonar settings. Most pro audio cards come with their own control panel, giving you control for monitoring directly through them. With my Dakota there is a very, very short latency produced this way. Not even enough to cause confusion with timing.

As for your dropout issues, have you disabled Microsofts Automatic Update feature, and disabled or uninstalled Microsofts Indexing service? How about system sound?...disabled? Installed in Standard PC mode if W2K or XP? Basically, what I'm asking is has your system been optimized for audio? As for the Automatic Updates, whether you're attached to the Internet or not, it will try to find an Internet connection and check for updates if successful about every 5 minutes or so....that will most often cause dropouts. In Sonar you'd see a sudden spike then dropout.

What about shared interrupts, especially with the audio card, but also with Video?

You haven' t given us your configuration...so...are you storing audio on it's own separate drive, while keeping the OS, programs and files stored on a separate drive? For Example:

C: = OS, Apps and files
D: = Audio files only


This is a pretty common issue, and these are pretty common answers. As you might have assumed, your system is powerful enough...so either you have an incompatibility, or mis-configuration involved here. If increasing the latency makes it function better, then there are probably some tweaks or updates that you can do that will improve the perfomance of your system.

[ljc]

Thanks, time to do a little homework. I've done a sort of old school optimization, meaning removed most backround software like virus programs and basic stuff, but I think the automatic update is kicking around. I've heard that using my SCSI drive for audio is a good idea as well . but I'm still using my C drive for the whole world. I will put all these tips into action and see if I can post details on my configuration. Oh, here's a good question - does the CPU and Disk usage at the bottom of Sonar really give an accurate sense of what is going on ? I've never seen either go above 8%, even during a dropout, so this leads me to think that it's only half the battle. It would be great to have a plugin or something that monitors and combines all of these items like buffers and latency setting into a monitor. Every time an interrupt occurs or how many and what device. Anybody hear of something like that ?

[mixsit]

Since driving the whole system with synced midi AND accurate audio recording seems to generally be at best a can of worms, might I suggest breaking the problem down a bit first: Load the shit out of your system with a whole bunch of good ol' fashioned 24/44.1 audio recording clocked to your converters. Now how's that part working?
Cheers, and good hunting.
Wayne

[mcnameep]

I had the same problem as you. I had a 233mhz with 64mb ram. You say you have an amd 2000xp. I do not know what that means. I assume it's an amd processor with xp as an operating system? What speed is the processor and how much ram do you have? I upgraded to a 2.8 ghz processor and 512 ram and the problem went away. I tried to break it for fun and record 32 tracks which I'll never do in real life and it was very happy with no dropouts. You can mess around with latency and buffers and whatever but if your processer can't handle it in realtime you are not going anywhere. You probably need to upgrade your PC. Besides being a musician I'm a computer systems analyst so I know a little bit about this stuff.

[ljc]

amd 2000xp is AMD 2000 XP processor with quantispeed arch.. AMD is intels major chip competitor, they try to rate their chip in comparison with the pentium4. 2000 is equivalent to 2 gig in a pentium and XP is their acronym for extreme processing, pretty clever because it causes buyers to think its got something to do with microsofts xp OS. As you can see I'm pretty well upgraded. I picked AMD for the number of instructions per clock cycle and some of the comparisons I read about for processing power. Anyhoo, enough about chips. Back to dropouts, so far I've caught a low hanging fruit in my task manager, the old MS updates was running every 5 minutes. Now for some tweaking .......

[dachay2tnr]

Originally posted by mcnameep
Besides being a musician I'm a computer systems analyst so I know a little bit about this stuff.

Then one might expect that you would know that an AMD Athlon XP 2000+ is a brand of processor with a clock speed of 1.67 GHz.

But then, what do I know.

[mixsit]

Wow. Drop outs (with as yet un-know causes), processor seeds, up grades, all thrown in to gether.
If we're talking good clean, uninterupted track count, then my experience with my first system -333 win 95 with 128 megs, says it's not processor speed. Did fine at 20, 30 tracks 24/48. No dropouts. Nada. But I wasn't mixing in the box, and wasn't flying with midi.
Now take a hot-rod of the day, you can still bog down with a similar load, just keep loading on the plugs.
I feel for you. Going through all of this just to get to music at the other end. (It happens to all of us from time to time dosn't it?
I'm sure when you break the problem down to it' basics, you will over come.
Lord knows I'm no computor techy. So it looks mire and more like the money I sent to Jim over at StudioCats does not sting badly at all...knock on wood.

(Hang in there, the music's comming...

[tntkemp]

Yeah, I got my Dakota from Jim. I've read alot of good things about him, his company and his talents. If you don't know what you're doing, then it is a good idea to let someone who does handle it for you.

However, I hope you have a good backup, like a ghosted or mirrored c: drive (assuming that your OS is on C Someday either something is going to go wrong with the drive, or something you install is going to make things crazy...it's inevitable in this imperfect computer world of ours! I ask myself regularly "Why did I ever get into fixing these things?" The answer, inevitably, is so that I can pay my bills...but computers can still be such a PITA sometimes!

On the other hand, there are tons of resources out here now, such as the one that we are using here now... and http://www.musicxp.net/MusicXP to help those of us that want to take it on our own, or at least maintain what we have. The majority of us should definitely thank God for that!

[ljc]

My system :
***************************************************
Sonar 2.2 with all updates
AMD 2000XP+ processor with FIC AN11 motherboard.
128 mb DDR RAM
tnt2 m64 32mb video graphics card ( nvidia )
DMX6fire audio card with break out box ( same chip as M-audio )
24/10/48 justlink CD burner ( Nero )
9 gig quantum Viking external SCSI drive ( 7200 rpm )
24X pioneer external SCSI CD Drive
USB printer
***************************************************

My gear :
***************************************************
Kurz k2600XS 88 key weighted sampler/Midi workstation
Motu 8X8 * 128 midi express midi patch bay
Aw4416 digital 24 track audio workstation
***************************************************


My Settings :
***************************************************
DMX 6fire :

DMA buffer size can be set to 1,3,5,7,10,15 ms
I am set to 10 ms which seems to clean up pops and clicks
I set "record digital In" to 44.1 Hz with masterclock to external
for recording ( spooling from digital board where board is master )


Windows :
multimedia is set to DMX card for record and playback
IRQ 5 is set up like this :
5 = DMX audio card
5 = VIA tech PCI to USB controller
5 = Win 9x Promise Ultra100 IDE Controller
5 = ACPI IRQ Holder for PCI IRQ steering

This is strange, because last I checked I only shared IRQ
5 between DMX and one other device, now I'm sharing 4 ?
It's been near impossible to force bios settings to isolate
an IRQ to DMX audio only, but I figured the video and most heavy interrupt devices are using their own. This does look like USB printer could interfere though.


Sonar :

I do not use plugins
In the project General settings I have audio driver bit depth set to 24 bit, Sample rate is set to 44.1. file bit depth is 24 bit

In the advanced tab Read and write caching are disabled
I/O buffer size is 128 kb
Apply Dither is set and Full chase lock is set since I chase Motu timecode. No timing offset is applied.

Sonar midi clock is set to smpte/MTC but no start/stop/continue
I do set transmit MMC and I use device 20.

AW4416 :

I set up a song in 24/44.1 mode and my midi settings are sync to MMC and MTC as slave. I set device 20 as my MMC device.

I typically send stereo master out the digital out which ends up input to DMX digital in. I also digital in to a selected pair of faders from the DMX to my AW. I have to be careful not to create a playback loop between Sonar and AW, but other than that my path is straight digital conversion to PC and AW ( actually no bit conversion at all since the DMX digital in/out is true bit for bit exchange.

Motu :

Set as master clock and although Sonar kicks things off with MMC, its the MOTU that generates clock/timecode ( AW and Sonar become slaved ).

****************************************************

[mixsit]

I basically asked the same questions in another thread here with what looks like the same (or similar) problems and setup, and I'm still unsure about one thing...
Are you actually suposed to be able to record audio with midi or SMPTE as the master clock? Midi for one, is not sample accurate.
Thanks in advance.
Wayne

[tntkemp]

Just like I believe I told you in the other post, it's not MIDI clock or SMTPE, rather it's MTC that you need. Your options in Sonar are Internal, MIDI Sync, SMPTE/MTC and Audio....choose MTC on both ends.

[mixsit]

Originally posted by tntkemp
Just like I believe I told you in the other post, it's not MIDI clock or SMTPE, rather it's MTC that you need. Your options in Sonar are Internal, MIDI Sync, SMPTE/MTC and Audio....choose MTC on both ends.

..to record audio? I have found in the manual where it say's it will record midi, or chase with audio, but not record audio.
Thanks again for being patient!
wayne

[ljc]

your quote : "Just like I believe I told you in the other post, it's not MIDI clock or SMTPE, rather it's MTC that you need. Your options in Sonar are Internal, MIDI Sync, SMPTE/MTC and Audio....choose MTC on both ends"

I do use MTC. What other post ? If you look at my spec it says I choose SMPTE/MTC in Sonar and slave to MTC on the AW. My midi rig is stable. The settings for Sonar are on the MIDI CLOCK tab and SMPTE/MTC are a combined choice there.

[tntkemp]

Sorry, ljc, that was in response to mixsit's interjection.

mixsit,
Yes, I will tell you I know little about the difference between MIDI sync and MTC Sync other then by experience. I did this once over a year ago, and it worked perfect, so I've been doing it ever since. Sonar, the sound card in use and the Master device...such as the BR8 or VS840, must all agree on one clock source, or you will get random clicks from the timing slipping. If you set the sound card and Sonar to slave to MTC, and set the master device to transmit MTC, it works great...almost too good to be true. I've been doing it this way ever since, and have discussed it with others that do the same.

[tntkemp]

oh, and ljc,
Your last comment was "The settings for Sonar are on the MIDI CLOCK tab and SMPTE/MTC are a combined choice there."

Yes, SMPTE and MTC do share a setting under Project Options/Clock. There is also a separate MIDI Sync option. I thought I understood that you were using the MIDI Sync option, rather than the SMPTE/MTC option. If you have the clock source set to SMPTE/MTC in Sonar, MTC selected in your sound card's clock source setup, and Generate MTC in any external device, you should be ok. Of course, you can use any capabled device as the clock source for MTC, but Sonar can't do it, because it's not capable of generating MTC....it's only capable of receiving it.

[Wurlitzer]

I personally have found that dropout problems in Sonar seem to have NO connection with the load on the system. I get them just as much running 3 midi tracks and a soft synth as I do with 20 audio tracks loaded to the hilt with effects.

I've optimised my system to the hilt, and I now get reasonable performance at about 12ms latency, which is fine so I get on with it. But my system should be capable of more.

The only time I really get dropouts now is during a lot of screen activity - scrolling during playback, or most especially if I minimize or maximize a window during playback. I have the most classic and known stable motherboard (Asus TUSL-2) and graphics card (Matrox G450), and a Mixtreme soundcard that is rock solid in other apps.

My point? Dunno really - just that I think there are some basic instability problems in the coding of Sonar, that have nothing to do with CPU load but probably are to do with MIDI/Audio sync, or over-sensitivity to the soundcard's access to the PCI buss, or something. (For example, I have found that when Vegas isn't entirely happy at low latency, I get clicks and pops but it doesn't drop out - it seems more robust).

I think Cakewalk still have some work to do on the solidity of the Sonar engine, and I'd rather see them attend to that than bring out loads of new features.

[ljc]

I probably should have stated that I do not share midi with my soundcard. I use a serial port with my motu and the motu drivers are very good ( this is standard for midi ). Sonar is set to recieve SMPTE/MTC as slave and generate MMC as master. The motu really takes control over all of this. I only use my card for audio although it is capable. I just went out and picked up another 512 MB RAM and am going to use my SCSI drive for audio from now on. So currently I have removed MS updates from background tasks, adding more RAM, offloading disk access to SCSI and finally I'm going to research my IRQ layout a bit more.


AW4416 [ MMC slave ] [ MTC slave ]


Motu [ MMC thru ] [ MTC midi master clock ]


Sonar [ MMC master ] [ MTC slave ] : ( no soundcard )


**************************************************

SONAR [ record incoming audio ]

DMX 6fire audio card
[ slave external clock 44.1 ]

AW4416 [ generate 44.1 master clock]


**************************************************

[tntkemp]

Yeah, I'd rather hear a few pops and clicks in the audio, rather than just dropout myself, and I'm not looking for new features, since I haven't use alot of the current features. I'm game for making it more user friendly, and more stable....then I'd be happy.

What it comes down to for me is this: It's not Cakewalks issue that causes dropouts, but it is their problem to figure out how to avoid them. The should be able to learn about what causes dropouts in the first place, then somehow make the software aware of these, and do what it has to to continue working. It may be a tall order...but I think it would make alot of people happy. Heck, if you got a great performance with a couple of Pop's and Clicks...you could always filter for the Pops and Clicks!

[mixsit]

Thanks again for the information and extra details Tmtkemp.
Wayne

[tntkemp]

No problem...I like to help when I can....

[ljc]

Just think if they came out with a monitor that let you know :

background tasks
IRQ conflicts
DISK status
interrupts that occur


Then when a dropout occurs, give a suggestion as to what the cause was and a solution. That would be cool.