It's not a Vegas issue. It's a Decklink issue. They update drivers, and break everything that precedes them. That's why Final Cut Users are a full 2 revs behind where Vegas is, because like Sony, Apple refuses to be writing new code every week when BMD changes out drivers, breaking the old drivers in the process.
Decklink broke their AMD drivers back at 4.5, and hasn't fixed that, either. Tweaks me, because I can't use my HDLinks in my fastest/biggest machines.

Interesting, I think AJA (or is it Kona) are now targeting the PC users, maybe a little more competition will get BMD a bit more focussed.
BMD seem to offer a lot of 'extras' that I doubt more than 1% of the user base would make much use of.
As I said, the currently supported drivers do the job, I mean a driver is a pretty ho hum bit of code, if it works there's no reason to fiddle with. Adding 'features' to it that mean serious changes in the SDK is one sure way to piss of your 3rd party developers. What's more it makes it harder for them to add the missing features that I've been jumping up and down for since day one.

Bob.

thank you for replay guys.
let's say, if i need to capture multichanel audio over sdi from digi or imx tape, i have to use premiere since vegas can't. premiere 1.51 does it so-so with 4.8.1 but export into stereo or mono waves isn't functional. premiere 2 with 5.4.2 does it much better.
we are in closed loop for shure.

but april isn't so far away.

regards,
gordan antic

Spot,

The Decklink Extreme card works fine with later BMD drivers in AMD machines, so I wonder if your HDLink driver problem could be related to a specific motherboard or PCI Express configuration rather than AMD processors per se? That has happened to many other pros.

BMD's drivers are busy and very sensitive creatures. The reason is that they are trying to bypass as much of Windows as possible to get sufficient performance to crunch huge video frames (data-wise). Their drivers seem to be running in kernel mode, where data transfer is run synchronously or nearly synchronously, and there is little margin for delays caused by other CPU tasks, or a PCI-E bus controller that is also used to shuffle data for something else.

The question is if they could write the drivers at a higher level if they raised the minimum motherboard performance required. Not so easy to achieve even today,
and it's not just a CPU issue.

BMD's cards seem to work perfectly when used with the exact workstations and motherboards they have on their Approved list.

For anything else, you really have to ask around if somebody else has used a particular configuration successfully.

Fortunately BMD has finally added more AMD configurations, saving them from being relegated to the dustbin of history :O).

I doubt AJA would do any better at this time, although it is possible that they have grown some driver talent in house (BMD used to write their drivers).

Writing reliable high performance drivers requires a much deeper level of understanding of the operating system than most programmers will ever acquire. We're not talking the top 1% here, or even the top 0.01%.

I have a close friend who is one of the top five people in the world when it comes to high performance drivers. After a team of 14 specialist programmers had failed to write a reliable and fast Internet message switch after two years of intense development, he wrote a replacement all by himself in three months that switched several million Internet messages per hour. On 1990s hardware... This software is still used at many telcos worldwide, because it is so reliable (I think two relatively minor bugs were found in the first six months).

Of course this is a guy who used to have a complete OS source code listing next to his bed, then read and understand his way through it one evening at a time, until it was all in his head...