PDA

View Full Version : PCI bus problem in ProTools/Panther



biggles
09-05-2004, 11:20 AM
This is the query I just sent to Digi Support, but that could take a while, so in the meantime does anybody have any idea what could be tying up the PCI bus? PTLE 6.4 & Digi001, Panther 10.3.4, G4 DA 733.

Always, within a few seconds of starting to record, the process stops and I get the following alert:

"DAE ran out of time while sending audio over the PCI bus. Try reducing the number of tracks, or using a playback engine with fewer voices per DSP. (-6042)"

I have been successfully using this configuration for 6 months with no problems at all until now.

This problem occurs with just 2 audio tracks, one of which has the Click insert. Of course, I don't have the "fewer voices" option, and the Playback Engine is set to use 85% CPU.

The problem always occurs between 1 and 5 seconds after starting to record, and may occur during playback.

When the problem first became apparent, it was actually during playback, and in addition the Left monitor audio would occasionally drop out, sometimes cleanly, and sometimes with a distortion as it came or went. The same audio file would playback cleanly through both channels, so the file is intact. After the troubleshooting I've done the audio drop-out is not occurring.

When a session is opened, the PCI bar in the System Usage window shows green to about 50% with the vertical green marker at about 80%. The usage bar doesn't move at all, the marker will flash if I click in the bar, but it remains fixed too. When recording stops with the 6042 error the PCI bar remains the same, and doesn't show a red clip. I have seen it clipped during playback, and playback continues with no 6042 error.

Trouble-shooting done so far;

Checked the Answerbase for -6042 error.

Ran DigiTest, all OK.

Trashed those Preferences in user/library that seem to pertain to PTLE and DAE, but of course I may have missed something!

Created another Admin user account and tried from there, same problem.

Re-installed OS and PTLE.

Pulled all but the 001 card, moved the 001 to a different slot.

Booted from a different drive.

Booted and recorded to a single drive on the native ATA bus.

Tried a different interface cable.

Repaired permissions.

Reset PRAM.

OpenFirmware Reset to defaults.

Installed OS 9.1 and PTLE 5.1.1, opened same session, and it records just fine, so I guess the hardware's OK. :)

I couldn't see anything unusual going on with Activity Monitor or top. Is the PCI bus choked, or does PT just "think" that it is? I've got PT in playback behind me as I type this, and the typing is slow. When I quit PT the typing speeds up, so I reckon the bus really is busy.

TZ
09-05-2004, 12:18 PM
Do you normally have PCI card for your add'l hard drives? OS on its own drive? Scratch? RAID? SATA? how much RAM (does it even help or matter? I assume you've maxed it out).

biggles
09-05-2004, 12:52 PM
TZ,

this strange gremlin just appeared out nowhere, it happened the first time I ran ProTools after 2 weeks on vacation. Nothing added or taken away, except (I've just remembered!) to install M$ Word. Hmmn!

The system has run without a hitch for the last 6 months since PTLE6. Now this.

1GB RAM, 2 PATA drives on the Acard host, one on the ATA bus. OS and App's on one drive, one for media, one for backup.

biggles.

TZ
09-05-2004, 01:05 PM
Clone your system over and de-install MS Office/Word (and anything Norton if you have those also). if you do clone and try a "test system" pop a 4GB file on the drive first, remove afterwards for a nice large "drive cache" at the front, and BE SURE that "Ignore Ownership" for the destination volume is NOT CHECKED. Then repair permissions once booted from the volume.

That is a nasty. which is why some production systems are kept lean, too. Check startupitems and background processes and see what gives - maybe it has too much interactive stuff. And see if top -au or Bigtop (part of CHUD) shows anything out of the ordinary.

biggles
09-05-2004, 01:15 PM
thanks TZ, I'll let you know how it goes.

biggles.

eric
09-05-2004, 02:10 PM
biggles,

When you de-install Office, use the Office uninstaller on the CD. That will remove the little bits of junks hidden here and there. If memory serves, it's what you need to uninstall the deamon which is checking the network for other copies with the same serial number.

biggles
09-05-2004, 05:07 PM
It would appear that the Acard ATA host is at the heart of my problem. I really thought that I'd covered that possibility early on when I'd stripped things down to just the Digidesign card and one drive on the native bus, but there you go. For the last 6 months I've been running the boot drive and the media drive off the card. Why it should go bad now I have no idea, and it's only a problem in OS X. It matters not whether the OS & App's drive is on the card or the logic board, but any attempt to record to either drive on the host card produces the error within a few seconds. In OS 9, which is on the native ATA drive, recording to a hosted drive works OK! Go figure. I don't know if this is a hardware or a software fault; is there any way of testing the Acard 6280M? BTW, Apple Hardware Test showed no faults on the computer. Acard do have a firmware update, which I did some time ago but just did again, just in case. There is also a recent driver for OS X which I just installed, just in case, but neither improved things. Oh well, I was going to get a SATA card and drives soon anyway. Thanks for all the suggestions.

biggles.

M.Brane
09-05-2004, 09:33 PM
Nothing added or taken away, except (I've just remembered!) to install M$ Word. Hmmn!

The system has run without a hitch for the last 6 months since PTLE6. Now this.

No M$ allowed on my DAW Mac. :crossbne:

PT runs just fine on a Tempo Trio. I'm using 5.01 on 9.1 though. I have noticed a lot of posts lately on the DUC that involve sudden bizarre issues, and Panther.

biggles
09-06-2004, 03:32 AM
M.Brane,

I couldn't agree more, but this was a case of "needs must when the Devil drives", and much as I despise that particular software I find it hard to credit that it could break the Acard, and it is the Acard causing the problem. Whether it's the hardware or firmware that's gone bad I can't tell.

biggles.

TZ
09-06-2004, 03:52 AM
In all the troubleshooting, have you reset-nvram at some point? That rebuilds the device tree quite nicely.

When updating firmware on ACard, be sure to not have any drives attached, or at the least, unmount them first. Otherwise, it is possible to result in corrupt firmware.

Just saw a post on XPF forum and how someone is using 8600 w/ new G4/800 and SCSI chain to get their audio work done. As horrible as the price of SCSI can appear, it DOES get the job done. Would SeriTek card get along and play nice?

biggles
09-06-2004, 05:05 AM
TZ,

OF reset was one of the first things I did, and yes I disconnected the drives from the Acard for the firmware update.

All is moot now though. SATA is more than fast enough for my needs. I just ordered the Seritek and cables from the Store and a pair of 120GB 'cudas from a local supplier (sorry 'gurus, they're actually cheaper here in the UK, and that don't happen often!).

I'd still like to fix the Acard if that's possible, partly out of curiosity, and partly because it should go back in the B&W which needs it on account of the flakey IDE thing.

biggles.

biggles
12-10-2004, 09:20 AM
Sorry guys, I should have closed this issue a long time ago. It turns out that the Digi001 PCI card had failed! I've been so chuffed with the 002R after Digi's hardware exchange that I completely forgot. Digidesign care not whether the 001 card is functional or not, so it really was a no-brainer.

biggles.