Dropped samples on Render

[Bulls Hit]

When I render a 2 track (24/48) containing UAD plugs (slowly, real time render) I'm still getting dropped samples. The dropouts occur in a different place each time I render.

The start of the dropout occurs (in the last mix) at sample 11880707 andit comes back again at 11881912. That's....quite a lot of samples

[Justin]

this probably belongs in bug reports / UAD.. (I should make a subforum, heh)

[Bulls Hit]

Yeah I did a bug report on this a couple of weeks back.

http://www.cockos.com/forum/showthread.php?t=6665

Even managed to get a screenshot in there.

Hmm a UAD bug subforum would be cool. I'd have thought you'd be heartily sick of UAD by now though!

[griz lee]

Bulls. I get this if I don't have Render Ahead turned on !!

Turning on UAD-1 synch mode turns it off. So i go and turn it back on for rendering.

Don't ask -- my system is completely out to lunch, but it could work for you.

[Bulls Hit]

Hmmm. Thanks Griz, it was worth a try. It still dropped samples though, and otherwise didn't behave any differently to FX render ahead off

[billybk1]

The start of the dropout occurs (in the last mix) at sample 11880707 andit comes back again at 11881912. That's....quite a lot of samples


I think we discussed this issue some time back. I had a similar problem a few years ago when I was using SONAR 4 & ACID Pro 5 and rendering with UAD-1 plugins, particularly UAD-1 plugins on the master bus. Nobody from UA, Cakewalk or Sony tech support had any idea what was causing it. This went on for months and was really getting on my nerves. Until, on a lark, I checked my IRQ settings and found to my surprise that a UAD-1 card, was sharing an IRQ. I say surprise, because when I initially installed the 2nd card, months earlier, I made sure it was not sharing and IRQ. What I learned was that Windows IRQ assignments are dynamically allocated. Which means whenever you install/un-install devices Windows can & will change exisiting IRQ assignments to acommodate the new devices. I had installed/uninstalled several other devices subsequent to the UAD-1 card install.
So it ending up being an IRQ contention issue. One of my UAD-1 cards was sharing an IRQ with a USB Host Controller (probably the worst device to share with because of the USB ports constant polling activity which can & does interrupt a render) and would cause gaps in my exports. When I zoomed into the 512 sample level, in Sound Forge, on the resulting render I could visually see an actual gap, the length of which was determined by my project sample latency. If I exported @ 1024 the gap would generally be in the 900-1200 sample range, if @ 128, the gap would be quite a bit smaller (80-150 samples).
I routinely check my rendered stereo mixes, including REAPER's, in Sound Forge (force of habit) for any inconsistances and I have not noticed any gaps, in any renders when using UAD-1 plugins. Of course, all (4) of my UAD-1 cards are on their own IRQ's or sharing with each other.
Having dropped samples on render is indicative of an IRQ sharing issue. Are you absolutely sure that none of your UAD-1 cards are sharing are IRQ with another device, particularly a USB Host Controller?

Cheers,

Billy Buck

[billybk1]

Here is a pic that shows the gap in renders when using UAD-1 plugins and having it's PCI card sharing an IRQ. This was taken from Sound Forge zoomed in @ 64 samples, back in 2003, when I was having this issue:


http://216.77.188.54/coDataImages/p/Groups/58/58781/folders/86478/573987SONARUAD-1Export.jpg

[Bulls Hit]

Yeah that screenshot is just what mine looks like.

I've checked the IRQ thing, but my card is using 22 all by itself.

I'd love to find the cause of this

[billybk1]

Have you tried switching to another PCI slot? Sometimes even though the UAD-1 is not sharing an actual IRQ, the PCI slot that it is in can be sharing the "PCI Bus" with another PCI slot (like a AGP video card). Try swapping the UAD-1 card to another PCI slot even if you have to swap with an exisiting occuppied PCI slot, if you have none available. Try it, that may indeed may solve the problem.


Good luck,

Billy Buck

[Bulls Hit]

Well it was worth a shot, but now it's sharing 23 with a usb thing. I did a render anyway for the hell of it, but it still fumbled the samples