View Full Version : Graphic weirdness - zoom, etc ends in BSOD
Twice today while working on a post project with a 30 minute video on the timeline, Nuendo's graphics freaked out - the arrange page turned into a larger blocky outlined version of itself.
Had to force quit Nuendo. Anyone seen this, or know of a possible resolution?
Running a Decklink card for external video playback, which was fine. The system has an NVidia GForce 9500 GT. I'll call the builder tomorrow.
I've increased the video thumbnail cache to 128 to see if that helps.
Thanks.
RiffWraith
07-01-2009, 09:53 PM
Saw your other post.
I have never seen that kind of wierdness here. Tell me more about the project - type of vid file, codec, how large, 32 or 64-bit, etc., and I'll see if I can dupe here in C4/5.
Cheers.
I somewhat doubt it can be duplicated, but I am running Nuendo 4.2.2 (32-bit of course).
Video files is PhotoJpeg, 70% quality. Roughly 30 minutes in length. OMF import of 48k audio.
The first time this happened, I had zoomed out to see the full video/project length - probably hit the max zoom out point.
I had never seen this before either.
RiffWraith
07-01-2009, 10:15 PM
"Video files is PhotoJpeg"
.mov file? .avi file?
How much RAM is being used?
"4.2.2 (32-bit of course)"
There is no 64-bit vers of N4?
Sorry - .mov. Not in the project anymore, but it had 1.6G free when I had to kill it the last time.
4G total, /3G switch.
RiffWraith
07-02-2009, 11:09 AM
I was going to check that here, but I do not have the option of converting anything to photojpeg :eusa_eh:
That's what I use Vegas for. Quicktime pro can do it as well, but it's painfully slow compared to Vegas. Funny, QT can't convert it's own formats in any reasonable amount of time but Vegas can zip right through it.... lol.
MJPeg Streamclip can convert to/from most anything. It's also faster than Quicktime.
RiffWraith
07-02-2009, 12:23 PM
Well, converted to Phot JPEG with that converter (cool little tool - thanks!) but I keep getting that farkin' error:
http://www.cubendo.com/showthread.php?t=982
No solution to that error yet? The only time I see this with applications is if I leave another app open that had saved to that file (even if not open). Apps will keep a file "open" even if it isn't in visible use by that app. Annoying little habit of some apps (Cubendo does this with audio files that aren't in the pool, but were written to - such as a bounce/export - same with Vegas and Sequoia).
If you've closed everything but Cubase, then it's a head scratcher. Be sure no processes are left hanging. Not common, but it can happen. I have to manually kill Dreamweaver here as it always leaves a ghost .exe running.
Yet again, zooming/moving around this project has caused the graphics to go awry and killing Nuendo in task manager caused a BSOD. I can't afford this. I assume too many BSODs and I'll eventually hose the OS/system drive... time for another backup image.
*&$(*#&@ Nuendo - costing me more time than it's worth.
Anyone know of anything that could contribute to this? Graphics card? Drivers? The system was built in December, so it is new.
TAFKAT
07-02-2009, 10:31 PM
Hey Dedric,
Send me the minidump, that will give us a better idea to what caused the BSOD.
I have a few ideas, but I'd like to see the minidump first
RiffWraith
07-02-2009, 11:34 PM
What's a minidump?
TAFKAT
07-03-2009, 12:00 AM
Hey Riff,
A minidump is a small debugging file that is generated when the system BSOD's , which I can analyse using some diagnostic tools..
In this instance the MOTU driver has brought the system down, we just need to sort why the video activity is conflicting with the MOTU.
Motu and windoze always make me a sad boy :(
Ditto Sam. I've always held the opinion that MOTU pretty much sucks. This confirmed it, though I was hoping it would not be the case. It saved me $2500 over a MADI system. So much for that.
Not surprising, but my MOTU USB midi interface also loses connection with the system if I restart it - have to unplug it to get it back. If only my Unitor 8 still worked. Never got Clockworks to work with it either. We really need more options, esp. for hardware midi interfaces.
although you pay the $$ for RME - it has to be said that you know what to expect from it....i turn on the machine and work and the audio inteface/drivers is the last thing on my mind these days....
Then again I only need 16 channels of AES so easier/cheaper for me to say such things....
My Tascam FW-1884 has worked since day one, and they even bothered to fix a problem with 64 bit OS and over 4 GB of memory. In addition to the many I/O ports, it also has a 4x4 MIDI patchbay. Works great.
If the driver loses comms with the unit, a simple hard restart gets everything back, nice and tidy. I've heard plenty of nightmare stories about MOTU hardware. Lots of manufacturers are having problems sourcing quality 64 bit driver talent, or even 32 bit talent.
Yeah, it's the I/O requirements that are the problem - I need 48 channels minimum - 64 preferred. That's $3500 for the RME MADI system. Not a trivial difference in the cost vs. performance arena. I have RME in the other systems already, and that's my preferred system usually. Pure cost vs. I/O decision this time.
I'll try moving the MOTU card and reinstalling drivers. Maybe that will cure the BSOD issue at least until I can replace it.
Other than this issue (hardly a minor one), the MOTU system does perform well at low latency. Comparable to my RME when it was running on the last system.
RiffWraith
07-03-2009, 10:44 AM
Hey Riff,
A minidump is a small debugging file that is generated when the system BSOD's , which I can analyse using some diagnostic tools.
I knew that! :eusa_whistle:
Ha - thanks for the clarification.
In this instance the MOTU driver has brought the system down, we just need to sort why the video activity is conflicting with the MOTU.
Not surprising, but my MOTU USB midi interface also loses connection with the system if I restart it - have to unplug it to get it back. If only my Unitor 8 still worked. Never got Clockworks to work with it either. We really need more options, esp. for hardware midi interfaces.
Ok, the MOTU driver that "brought the system down" - was it the MIDI interface driver, or do you have a MOTU soundcard as well, with that being the problem driver?
Well, I had this system built and hadn't looked at IRQs, assuming they should be optimized, but the MOTU PCI audio was sharing with the 9500GT - problem #1.
Also talked to their tech support and they think the graphics card may be bad given the graphic corruption that happens after working for a few hours (e.g. heat, etc). Working today to see if this is problem #2 (I'll have to reinstall its' drivers just in case they've become corrupt, then work next week to see if the card overheats/corrupts again).
I reinstalled the MOTU (wiped drivers, started clean) in a new PCIe slot and it is better - better response in setting latency in Nuendo.
And here's an interesting twist - it no longer switches sample rate to 44.1k when loading a 48k project. Grrr.... that pretty much confirms what I said on the Nuendo.com thread - it's a timing issue that Nuendo assumes 44.1k too quickly - there isn't another direct tie between the video and audio drivers that would cause that afaik. The bus contention may have delayed the MOTU driver just enough that Nuendo decided to set the rate itself.
Still really irritates me that Nuendo makes sample rate assumptions so quickly, and incorrectly - the fact that it doesn't happen for others just implies that their drivers are circumventing the problem within Nuendo.
Yeah, it's the I/O requirements that are the problem - I need 48 channels minimum - 64 preferred. That's $3500 for the RME MADI system. Not a trivial difference in the cost vs. performance arena. I have RME in the other systems already, and that's my preferred system usually. Pure cost vs. I/O decision this time.
I'll try moving the MOTU card and reinstalling drivers. Maybe that will cure the BSOD issue at least until I can replace it.
That pretty much excludes just about everyone else, excepting Digi.
paulwr
07-03-2009, 03:30 PM
Sorry - .mov. Not in the project anymore, but it had 1.6G free when I had to kill it the last time.
4G total, /3G switch.
Can you run a 64bit O/S now? I'd dump the 3GB switch. I great luck for over a year with it, but then with some new software, had goofy problems. Just overall, you'll get MUCH more stability from XP x64, as an example.
(I didn't have time to read all posts, in case this is covered already, sorry.)
-Paul
Too much of a transition at the moment to go Vista or XP64. Maybe in a few months.
I'm booting without the /3G switch for now, but that's causes other problems with Nuendo - when reaching 1.2-1.3G free, Nuendo usually starts freezing.
Not sure if it's related, but this morning video playback started freezing, then Nuendo froze. Had to kill it in task manager - no BSOD at least. Probably moved the conflicts to the Decklink card now, or without the /3G switch Nuendo ran out of memory trying to paint video thumbnails. I could shut those off, and heck, shut off video too and probably solve it - heck, who needs video for audio post anyway? :icon_rolleyes:
Somehow doubt that was the NVidia card since the GUI/screen graphics were fine - just video playback/streaming seemed to lockup.
Yet one more thing I'll have to offload to another PC to avoid compatibility/conflict issues.
With all of these problems, you would think I was some 14-yr old with a $399 Dell in my garage, instead of a fulltime post studio with years of studio, tech and computer experience, a custom built DAW from one of the best (next to Vin :-), and a config that was pretty much a lock for stability based on my research and many recommendations.
Just goes to show that with the modern DAW, there are no guarantees.
TAFKAT
07-03-2009, 06:24 PM
Hey Dedric,
Well at least we figured out why the MOTU was being taken out with the vid activity. This highlights the less than intelligent resource sharing that is a common issue on modern boards. I have seem many end users select boards simply because they had more expansion slots than others, whereas in many cases the board with less slots was the better option as there was a far better chance of getting a clear resource. A perfect example a few years back was the Intel XBX2 - i975 board ( I don't mind letting everyone in on the board now ) , that by far was the best board of that genre for DAW use, and I have numerous clients with those systems , with every slot chocked, Dual Vid cards, multiple UAD's, Decklinks, etc, etc, and they just rock.
Re the Decklink, what slot do you have it in out of interest ?
If its not in the second PCIe x16, give it a run there and let me know how you go.
Quick summary of PCIe Slots v Lanes
I have done some investigation in this area , although no-one has ever been able to conform this 100%, but IMO, one misconception of the PCIx16 slot is that it only directly feeds to the Northbridge/ Memory Buss, when in fact its the actual PCIex16 lanes that feed the Northbridge/Memory Buss, the PCIx1 lanes on the PCI x16 slots are still running thru the Southbridge , thats why you can actually run the UAD2 cards on spare PCIex16 slots, many do not realise that and have a spare slot staring at them which in most cases has a clear resource.
Worth a shot
Re the Vid card, it does sound like there is a problem there as well.
What size LCD's are running off the card , if you only require Single link DVI , I would dump the domestic card for a smaller , leaner workstation card.
That one I'll fill you in via PM.. :-)
THE APIC on many boards is the best solution for most of these type of conflicts. In fact, I have not had a problem with resource allocation on my last two DAW builds, both of which had all slots filled. The selection process for MB components was very narrow.
Back when I built my first DAW, around 1997 or so, it had to be a dual-purpose system: both video and audio (on a NT 4.0 and a Win98 platform, respectively).
It was not easy getting all of that old hardware to work nicely together. I spent a full 60 hour week trying different combinations of slot selection where I could get both systems to function flawlessly. Naturally, because of the registry issues with both OS, I had to start the install from scratch each time I moved a card. That was the biggest PITA...
TAFKAT
07-04-2009, 10:13 AM
Hey D,
The current APIC employed is showing its age, as is the 32 bit BIOS, it was never designed to handle so many devices and the amount of low level sharing that is being enforced due to the increasing number of USB/SATA channels. APIC by nature is simply extending the IRQ list from 16 physical to 24 virtual on single socket boards, there are more on multi socket boards if they employ multi APIC's, not all do.
The other problem with APIC virtual channels is that even tho it may look clear in XP it may not necessarily mean that its clear at the hardware resource level, and also, the opposite , that because they are sharing in the virtual domain that they are sharing at the hardware level. Confusing eh ?
Re the earlier PIC 98/NT/2000 Systems, interestingly I had the opposite experience as many of the earlier boards altho reserved to 16 physical IRQ's, would allow you to hard set the IRQ's per slot , and no sharing. Also whatever IRQ's that were reported by the O.S, is exactly the resource being used at the hardware level.
APIC in XP ( from memory in W2K it forced all virtual IRQ's to 16 in some instances ) initially worked very well, but there was nowhere near the resources being shared at the hardware level, so we got away with it, its no longer a given now.. :-(
The APIC on my MB looks like it's providing 190(!) virtual Interrupt lines. Obviously it is much harder to determine exactly what is happening at the hardware level. And obviously we also have many more devices now than were available then, especially onboard devices like USB and SATA controllers.
Then again, back in the days, these would ALL be externally bussed devices (PCI mostly), and the internal buss had far fewer issues to contend with. Now almost everything is on the internal buss, so most users aren't even aware of them. Also, unfortunately, we aren't allowed to decide what device gets what. This would really fix these issues, but the powers that be have never seen fit to trust us with that flexibility.
Animus
07-04-2009, 02:58 PM
Yet again, zooming/moving around this project has caused the graphics to go awry and killing Nuendo in task manager caused a BSOD. I can't afford this. I assume too many BSODs and I'll eventually hose the OS/system drive... time for another backup image.
*&$(*#&@ Nuendo - costing me more time than it's worth.
Anyone know of anything that could contribute to this? Graphics card? Drivers? The system was built in December, so it is new.
I am starting to think you have computer problems. Didn't you say in another thread your motu was sharing with your graphics card? Besides the obvious bugs my Nuendo system is rock solid stable and never have any corruption or crashes, and when I have in the past it has always been something to do with hardware (for example a capacitor broken on my UAD Quad and was causing bluescreens).
I think it's a couple of different issues. Moving the MOTU solved some audio problems and the BSOD after force-quitting Nuendo (though only 2 cases where I've had to do that, and one that's a known bug with a VSTi).
The graphics problem may just be the graphics card going south (after only 5 months). So much for buying a custom built system....
I really have very little on this system, so it's hard to fathom how it could be other apps/processes.
If it's hardware, well, let's just say I'll be building my next PC myself.... sigh. Already lost the advantage of outsourcing that one with the recent problems, my time lost on issue, etc. Frustrating doesn't begin to describe the last 6 months of trying to expand my farm PC system.
TAFKAT
07-04-2009, 05:54 PM
The APIC on my MB looks like it's providing 190(!) virtual Interrupt lines. Obviously it is much harder to determine exactly what is happening at the hardware level.
Hey D,
190 Virtual IRQ's..., right, someone pass me the cool aide... LOL
The most I have seen on some Dual Socket systems with True Multiple APIC controllers is around 48 from memory , which makes sense . i.e 2 x 24 Virtual.
The Intel Macs don't have a BIOS, they have EFI (http://en.wikipedia.org/wiki/Extensible_Firmware_Interface) , which interestingly was developed by Intel for the Itanium Platform - or the Itanic Platform as its more commonly knowm.. LOL. BTW: The initial Windows Hack for the Macs was basically some guys hacking in the EFI boot sector of the Itanium version of XP into a regular XP install, which Apple then unofficially authorised with BootCamp.
Now while the Macnoids get all teary about it being miles ahead of a BIOS as they would while they sip their cool aide, in reality it is in some ways more limiting as its far more closed system than the regular BIOS, not even allowing basic enabling/disabling of parameters we take for granted in the BIOS, like EIST/CI Halt, etc.
Quote from Wiki :EFI has been criticized for bringing more complexity to the system, without bringing significant advantages and disabling fully open-source BIOS replacements like OpenBIOS and Coreboot. It does not solve any of BIOS's long standing problems of requiring two different drivers for most hardware. It is not clear why it is beneficial to have two completely different operating systems running concurrently and doing similar things, or why a new operating system has to be created from scratch.
The Free Software Foundation has criticized Intel for pretending to open up their platform, and for preventing interoperability: "It won't run, and doesn't bring us any closer to a BIOS that does run. It is just a distraction."
@ Dedric,
If the card has vid gone south, then thats just an unfortunate luck of the draw, not really something that can be attributed to the system being custom built.., PITA either way tho.. :-(
Hey D,
190 Virtual IRQ's..., right, someone pass me the cool aide... LOL
Would you like to see the screenshots? :wink:
Technically, they are numbered to 190, but there are a bunch of numbers missing....35 to 79, so it isn't exactly 190 (most of those are allocated to ISA IRQ placeholders).
There also seems to be considerably less hogging of I/O ports and system memory address allocations. This is good.
@ Dedric,
If the card has vid gone south, then thats just an unfortunate luck of the draw, not really something that can be attributed to the system being custom built.., PITA either way tho.. :-(
Hey Vin - I know. It's just a combination of factors that's frustrating - nothing to slight or blame my builder for in the least, or builders in general - you guys are working at the top of the DAW tech world, and there's no way I would want to keep up with the level tech info you know.
I've just lost all the time I was hoping to save, and then some. The bottom line has suffered to some extent.
I guess I could get a Mac next time with some Apple-coolness Koolaid and pretend it works better. The logo is just so cool. :wink: (no, that's not directed at any Mac users here - use what works for you).
TAFKAT
07-04-2009, 10:43 PM
Would you like to see the screenshots? :wink:
Hey D,
Oh I believe you that they are listed, all I am saying is that is not an indication that you have any more resources. The comment wasn't meant to be against you in any way, more so Snapples smoke and mirrors.
@Dedric,
I know here you are coming from Mate , and you have every right to be frustrated that you went with a custom build to save some time and energy, and it hasn't panned out.
It is tough to stay on top of all of the changes at the tech level, and even a BIOS / driver update can move the posts substantially on a previously known configuration, so it is a constant learning curve.
Anyhow, anything that I can do to help you settle the rig, just give me a hoy.. :-)
Anyhow, anything that I can do to help you settle the rig, just give me a hoy.. :-)
Thanks Vin - your help has already been a lifesaver. I know you guys go through hell to stay up with the computer industry's fickle flip flop of architecture, designs and compatibility changes. And they are the stable, logical, and predictable compared to the DAW software industry. lol
Michael Tibes
07-05-2009, 05:59 AM
Yeah, it's the I/O requirements that are the problem - I need 48 channels minimum - 64 preferred. That's $3500 for the RME MADI system. Not a trivial difference in the cost vs. performance arena. I have RME in the other systems already, and that's my preferred system usually. Pure cost vs. I/O decision this time.
Sortof my situation here as well. I'd like to get beyond my 26 I/Os that the RME digiface offers, but haven't had the cash to dive into the RME madi world. Luckily I didn't follow the temptation to try motu, their stuff still seems to be suboptimal - concluding from this thread?
I have some thoughts regarding madi options to get out of the motumess: there are madi cards from SSL, they are not expensive and even offer 2 madi ports if I remember right. BUT they don't have asio direct monitoring. I need that, maybe you don't, so it might be an option. They also announced cards with it, but I think they are not released yet.
Definitely worth some thoughts are the SSL converters, including madi-adat conversion. They could replace the RME madi-adat converter for a similiar price (roughly), but offering 24 converters on top.
And I've sometimes seen the older pci-based RME madi cards on ebay go for little money. If you have pci slots and resources avilable, that might be an option.
I haven't followed the whole story in detail, but in a sense of 'always expect the unexpected' I'd as well consider a hardware failure. Failing caps are hard to find, but mostly can even be found visually with a close and detailed inspection. Other usual suspects obviously could be the psu or heat, insufficient power supply or cooling for the graphics card?
Michael
re MADI - the SSL card is an option but I don't know enough users to get a feel for its' reliability, driver performance, etc. For the same I/O, their Alpha links are quite a bit more than RME's 648 MADI-ADAT.
I don't know that I would tank the MOTU based on this experience, despite my earlier frustrations with it (which I have to recant now) - it does appear sharing IRQs, and perhaps the video card were/are the main problems. The MOTU's low latency performance seems quite good in general. I don't have any of my RME's in this system to compare, but I've had it on 128 sample latency since day 1, with plenty of system load on many different projects. It should perform even better now.
The BSOD probably listed the MOTU because of the driver state when I killed Nuendo in task manager (sound likely Vin?). After the move/reinstall I haven't had any problems with ASIO at all. Just the one freeze with Decklink video playback the other day. Who knows, the decklink could be sharing with the video SATA drive now....need to check that tomorrow.
smoke and mirrors.
We're on the same wavelength.
Powered by vBulletin® Version 4.1.11 Copyright © 2012 vBulletin Solutions, Inc. All rights reserved.