ASUS P5WDH, use sound card not the on-board audio
ASUS P5WDH on-board Realtek Sound controller seems to be causing “late interrupts” & audio dropouts when recording to tape.
No video problems; just audio drop outs. It occurs only when recording to tape and never any other time. It’s only a problem with the recorded tape. There are no problems with capture and no problems with playback from the timeline. Even more annoying, when I monitor the audio during record to tape, there are no detectable audio glitches. (However, I did discover that I can hear the glitches if I tell the system to record analog video out using the Avid Liquid Break-Out-Box; however, this occurs only when Avid Liquid is told to record-to-tape. There are no glitches if I simply click the timeline “play” button and don’t use the record function.)
The problem stops when I disable the on-board audio controller in the BIOS. Because it’s a pain to have to reboot anytime I want to print to tape, I installed an older Soundblaster Audigy 2 Platinum sound card in a PCI slot, and haven’t had any problems since.
Anybody else had problems with the Realtek Audio when working with vdeo on the P5WDH Deluxe.
I use Avid Liquid Pro 7.1 (using SP1 with the 3581 and the AL-Break-Out-Box (BOB)) as well as my old and heretofore dead-bang reliable SD editing standby, Cinestream 3.1 (the last version of the DV-SD editing program originally known as Radius EditDV.) The audio drop-out problem is not limited to these editing apps; it also occurred when I downloaded and ran tests using the trial versions of Sony Vegas 7.0 and GV Edius 4.0.
Cinestream (unlike the other apps) will generate an error report when any problems occur on printing to tape. The error report says: “0 frames were dropped due to inadequate disk performance, xx number of frames were dropped due to late interrupts, disk throughput average was 0.1 Mb/sec.” The number of dropped frames varied each time, but the dropouts were only in the audio and never in the video. I could recapture/redigitize the segment from the just-recorded tape, and then display the audio wave forms (in Cinestream, Sound Forge, or whatever). The drop outs were points where the waveform went totally flat for 1 or 2 frames. The number of frames in those flat spots equaled the number of dropped frames reported in the Cinestream error message box.
For a five minute segment, the number of dropouts varied from as few as 15 to as many as 579. The problem occurred even with clips that were captured from tape and then immediately recorded back. The original camera miniDV tapes did not have any audio dropouts nor did the clips captured/digitized from those tapes.
As noted above, the problem stopped when I disabled the Realtek audio in the BIOS. So, at least on my system. there seems to be some kind of glitch in the Realtek audio controller where it hogs interrupts on the PCI bus. I’m guessing that the editing programs are sending out two program streams when printing to tape: one stream goes to the computer monitor and speakers (from which the sound was fine) and the another stream goes out over Firewire to the recording deck or camera (where I got audio glitches). What I can’t figure out is why the late interrupts would only affect audio and only appear when recording to tape.
I’ve e-mailed ASUS about this oddity but haven’t yet heard back from them.
Until I discovered that the problem went away by disabling the Realtek Audio, my main workaround was to move the project files to an external firewire drive, then connect the drive and deck to an older P4/2.4g laptop from which everything recorded to tape without a hitch. So, that pretty clearly eliminates drive, deck, camera and cable problems.
Here’s the information that pertains to my system and what I went through in trying to diagnose tried so far in attempts to isolate this problem.
System specs: A slightly downmarket version of Videoguys’ suggested “DIY 5" system. See http://www.videoguys.com/DIY5.html.
ASUS P5DWH Deluxe mobo (wireless and remote not enabled);
Core 2 DUO 6400 CPU, stock CPU fan mounted with Arctic Silver;
2 gB Kingston DDR2 533 RAM,
ATI Radeon XT 1950 Pro PCIe-16 graphics card (256 Mb) running dual monitors,
Maxtor 200 gB IDE UltraATA 7500 rpm system drive on primary IDE channel w/ 80 pin cable
(approx. 25% of disk space occupied so far);
Pioneer DVR111 DVD burner on second IDE channel
slave IDE drive (older 30 gig 7500 rpm Maxtor ATA 100 IDE drive,
used only for page files – set to 4092 Mb per Dave LaBorde’s WinXP Tweaks
Internal 640 gB Raid 0 run off the SATA connectors using Intel Raid
(twin 320 Mb WD SATA Caviar 7200 rpm drives)
PC Power & Cooling 750EPS “Silencer” PSU
External Firewire drives: LaCie 120 gB HD, Maxtor “OneTouch” 300 gB HD,
& Sony DVD burner, chained and connected to rear Firewire port
AL Pro 7 break-out box (BOB) connected to a USB port
USB dongle for Canopus ProCoder
PS/2 keyboard and USB mouse.
Epson 900 Photo/DVD printer is fed by USB
1500VA TrippLite UPS.
WinXP Pro, SP2, with all current updates
PCI cards: Digital Origin/MotoDV/TI Firewire card (used for realtime I/O with Cinestream)
Because of the messages about Late Interrupts and greatly diminished disk throughput, I spent a lot of time looking for memory conflicts and hard drive problems. What I went through in attempting to diagnose the problem is a long list of things that might (or might not) be of interest to others having trouble with editing workstations.
1. Camera, cable and firewire drive problems were ruled out because I could immediately move the camera, cables and a firewire drive over to the laptop and get a good recording.
2. This was not a system optimization issue. Windows Device manager “Advanced” tab settings for the drives are all for “performance” and “programs.” The properties settings for the external Firewire drives are been set to “optimize for performance” and not quick removal. Win XP Device Manager Advanced System tab is configured to give foreground programs priority, etc. I implemented all 21 of the recommended system and registry tweaks for video editing listed and explained in Dave LaBorde’s articles on WinXP Tweaks on the Videoguys website, http://www.videogus.com/tweaksWINXPVE.html which I’ve used for years. Reversing those tweaks back to default settings had no apparent affect on the problem (but did make editing slower).
3. Windows Device Manager revealed no apparent memory or IRQ conflicts.
4. The problem was not caused by interactions between the various apps and their associated hardware. In trying to track down the source this glitch, I wound up testing everything separately. For example, I tested AL 7.1 after uninstalling the BOB as well as uninstalling Cinestream and its firewire card, rebooting a couple of times, editing the Registry to remove all references and rebooting a couple more times before running the tests. Except for the print-to-tape audio drop out problems, I had no problems with any of the applications. Disabling the Realtek audio caused no problems with any of the apps, either. All seem to be fine with Audigy 2 card, as well.
5. I ruled out problems with firewire drivers. The ASUS P5DWH mobo has two firewire channels, one connected to the back of the case and the other to the front. The Cinestream card (Digital Origin’s real time editing card) is a PCI firewire card that adds its own dedicated channel for Cinestream output. (It was the only PCI card used in the new system.) The AL PRO 7.1 break-out-box (BOB) connects to the computer via USB but also has a firewire connector for controlled I/O. I can run Cinestream using plain OHCI connections, which is what I do on the laptop. However, it made no difference to the problem on the workstation whether I’ve set up Cinestream for the dedicated Digital Origin card or set it up to use the OHCI Firewire ports. The Soundblaster Audigy 2 Platinum — which was installed after I discovered that the problem went away when the mobo’s Realtek audio controller is disabled — adds two more Firewire ports which I have used without a problem. Also, this problem was not caused by the WinXP SP2 firewire slowdown problem. That was fixed per instructions on the Microsoft website. See http://support.microsoft.com/kb/885222. I’ve checked the registry repeatedly to make sure the fix is still in place and hasn’t been overwritten by something else. Also, as noted, there is no problem with capturing video or editing.
6. The audio dropout problem was unrelated to the simplicity or complexity of what was on the timeline. The drop-outs occurred with projects as simple as a fifteen minute, cuts-only edit of a witness interview from a single tape with a single audio track. It occurred on some of my most complex projects (such as 1½ hour video of a theater production using 14 camera angles). I had the drop out problem even if I captured good video from tape and then tried to immediately record it back to tape. The input files were all fine. The re-recorded tape segments were not. (A comparison of the tape to a re-recaptured file showed that all dropouts were faithfully reproduced in the new capture file.)
7. Electrical power problems were ruled out. The system PSU is new and replacing it with the older PSU that I used in the predecessor editing workstation made no difference. Everything runs off new 1000 and 1500 VA UPS units that also function as line conditioners. Each UPS unit is on a separate electrical circuit (separate breakers. Power monitoring meters showed minimal fluctuations, anyway. Swapping to different UPS units and/or surge protectors had no impact.
8. The problem was unrelated to how full or empty the media drives were. (Drives are never allowed to get over 75% full). All drives were properly configured for NTFS, UDMA, etc. Repeated runs of Defrag, Chkdsk and DiskCleanup did not affect the print to tape problem.
9. Repeated runs of hard disk throughput testing utilities such as Canopus RAPTest --- l – showed more than ample throughput whenever the problem was encountered. (These utilities were run repeatedly because their results can vary a bit on large NTFS partitions). I did discover that I got much better throughput for the Raid when attached to the mobo’s Intel SATA-Raid connectors that when connected to the on-board ASUS/Silicon Image Ez-Backup Raid connectors. It turned out that the EZ-Raid runs on Sata Channel 2 and splits the channel between the two drives. RapTest reported throughput of about 75 mb/sec on the EZ-Raid. Moving the Raid 0 over to the onboard Intel connectors on the mobo (which gave each drive a full SATA channel to itself) raised the throughput rates up to 110 MB/sec for disk reads. This was interesting but made no difference to the recurrence of the audio drop-out problem.
10. Although I use an older, somewhat slower drive for the main WinXP paging file, I was able to rule it out as a problem. I physically removed the old drive from the system and relocated the paging file to another, faster drive (one not used for project media) and also tried moving the paging file all back to the system drive. This had no affect on the print-to-tape audio dropout problem.
11. Heat is not a problem. The Antec P180B case came with three 120 mm case fans. Those, plus the CPU and GPU fans, plus the very large PSU fan, plus another 120 mm fan for cooling the hard-drives keep temperatures below 41̊ C. (Temp readings from physical probes as well as the ASUS BIOS monitors that have a reputation being inaccurate on some motherboards.) At one point, I tried increasing the ATI card’s graphics speed using the clocking utility in Catalyst 6.x, which did raise the card’s operating temp to 51̊ C and sped up Cinestream’s visual display of audio waveforms, but didn’t have any effect on the print-to-tape problem.
12. There was no pattern to when the problems occurred or didn’t occur. Sometimes the problem would occur with one segment of a program but not with others. Sometimes it occurred right at start up in the morning (when the system presumably would be cool). Sometimes it occurred later in the day and never went away for the project but would stop if I started a new project. Also, it occurred with the both of the external Firewire drives. I could move all project files to single firewire drive (from which I would still have the drop-out problem) but could disconnect that firewire drive from the editing workstation, immediately hook it the P4/2.4 laptop, and print-to-tape/record without a hitch. If I then disconnected the firewire drive from the laptop, immediately reconnected it to the editing workstation, the drop out problem would recur.
13. Also, if I put all project files on the Raid, disabled all the firewire drives, uninstalled their software, rebooted twice, and then tried to print to tape, I still got audio dropouts.
14. Restarts, shut-downs and complete reboots usually had no effect. Sometimes, walking away from the project for a day would cure the problem but only long enough for me to record a couple of timeline segments. Generally, once the problem occurred with a project, however, it always occurred thereafter with that project on the new editing workstation.
15. One thing that sometimes worked with Cinestream but had no apparent effect under AL 7.1, was to set the application process priority to “high” or “realtime.” (Ctrl-Alt-Del to call the Windows Task Manager, click on the “Processes” tab, right click on the editing app’s “exe” file, chose priority, set it to “high” or “real time” and ignore the warning that this could make the system unstable. Doing this did not cause any stability problems except for me and my temper when the fix stopped working when I was close to a deadline.)
16. Cleaning out the C:WindowsPreFetch folder made no difference. Also, when the problem was occurring, it made no difference whether I the enabled “clean out the PageFiles at shutdown” or disabled it. (Enabling it only seemed to make for muuuucccchhhhhhhhh longer shutdown and reboot times.)
17. I used WinPatrol Plus (see http://www.winpatrol.com) to check on all operating processes. I turned off (or set to manual) anything not needed to run Windows and my editing apps. I’ve also used WinPAtrol to check for conflicting apps to make sure that Cinestream wasn’t running anything when AL was running and vice versa. Nothing is running that isn’t essential to running the system and the editing app that I am using.
18. I have never installed iTunes on the editing system, so none of its services are running. They do run on the laptop, however. Also, I fiddled with Norton Internet Security & Antivirus 2007 settings, then tried disabling them while recording to tape, and finally gave up and removed them and all components. Made no difference.
19. I used the Performance Window in the Windows Task Manager, as well as ASUS mobo tools, to watch processor usage when printing to tape. I saw no difference between CPU and memory usage when the problem was occurring and when it was not occurring.
20. Whenever I run the editing apps, I physically disconnect the network cable — so no network or Internet Access and no background updating was going on. All automatic updating and update notice functions are turned off, anyway.
21. Using WinPatrol Plus and the Win Task Manager applet (processes tab), I kept finding multiple instances of MsMsgs (Microsoft’s annoyingly persistent instant messenger applet). I have no use for this function and don’t do any e-mail from the editing workstation, anyway. No matter how many times I turned it off, it kept popping back up. There doesn’t seem to be any way to uninstall the Messenger app, so I just deleted all of its files. That reduced the number of instances of late interrupts but did not eliminate the problem.
22. Late Interrupts were sometimes a problem with sound cards and the early PC versions of EditDV (about ten years ago). Usually the problem showed up as snaps and noise in the audio during timeline playback, a problem that sometimes could be fixed by going into the BIOS and adjusting PCI bus frequency settings or adjusting the PCI latency setting. (PCI latency is a setting for how long functions get to use of the bus before being required to hand over to other apps.) Increasing the Latency setting often decreased the audio problems. However, with the P5DWH mobo, the default setting was 64 and going up or down from there did not seem to make any difference.
23 I’m aware of the problems with using Quicktime version 7.1.3 with many editing apps including Cinestream and AL 7.1. I have never even downloaded it to the editing workstation. I’ve been using QT version 18.104.22.168. I tried uninstalling it and using older versions going back to version Quicktime 6.0. The version of Quicktime had no effect on the problem.
24. This was not an overclocking problem. I don’t overclock. (Not that I’m against it, I just don’t want to spend all the time I’d need to properly tweak everything; it’s hard enough doing the work as it is.) The ASUS mobo comes with BIOS utilities that allow automated, system selected auomated overclocking and fan control. ASUS calls this “AI-NOS.” The problem occurred whether AI-NOS was enabled or disabled.
So, my conclusion is that there is something in the on-board RealTek Audio controller on my mobo that interferes with video editing.