Think and think alike...

[tbow::]>]

ATI has just released a new version of their Catalyst drivers, which look very nice by the way, and now have a backbone of .NET........very familiar. One thing I have always liked about Matrox is the drivers, wonder if ATI finally took notice of someone doing it right?

[Wombat]

If they were noticing someone doing it right, they wouldn't be using .NET.

[tbow::]>]

[Mikko]

IMO, the only problem when using .NET is that you don't have it in Linux. This means, of course, that you'll have to write completely different type of drivers for Linux and this seems to be a big trouble for Matrox at the moment...

[Kruzin]

When Matrox starting using .NET, everyone bashed it.
I predicted it was only a matter of time before others started doing the same. Nay-sayers (nVidiots and fanATIcs) quickly denied it, and told me I was crazy....nobody would do that.

So forgive me if I'm not surprized that I WAS RIGHT

[spadnos]

Originally posted by Mikko
IMO, the only problem when using .NET is that you don't have it in Linux. This means, of course, that you'll have to write completely different type of drivers for Linux and this seems to be a big trouble for Matrox at the moment...

The main problem with .NET as far as drivers are concerned is that it's there to insulate the programmer from the specifics of creating user interface, especially when that interface might include network components or the like! Anything that insulates the programmer from anything has no place in a hardware driver.

Obviously the reason they're using it is for the user control panels and such (plus the fact that anything which isn't .net probably won't be allowed to run on Longhorn (just my pessimistic opinion, not based on facts)), but it still seems counter to all good software practices to require high level libraries for low level device drivers

- Steve

[schmosef]

Just a guess but I think that the .Net component is also for integration with GDI+.

@Kruzin, don't get confused. People thinking that you're crazy and your predictions about video drivers are two separate issues!

[dwood]

Originally posted by spadnos
it still seems counter to all good software practices to require high level libraries for low level device drivers

Well said.....

[Peter Aragon]

Originally posted by Mikko
IMO, the only problem when using .NET is that you don't have it in Linux. This means, of course, that you'll have to write completely different type of drivers for Linux and this seems to be a big trouble for Matrox at the moment...

It's a wise choice however to do some experience with .Net if you develop for Windows. Sooner or later you would need to do that anyway, Microsoft's next operating system will natively support it. .Net is the next gen API, MFC will become a thing of the past.

On a P3 .Net is as fast as traditional code, on a P4 it's around 85%, but that figure is steadily increasing as they keep improving and optimizing the back-end.

But you never heard of Project Mono?
It's a ported .Net 1.0, running under Linux, OSX and FreeBSD. Your executable runs on all of these operating systems without recompiling, as long as you don't use not-yet-implemented functions from the framework.

[Virgulino]

Originally posted by Mikko
IMO, the only problem when using .NET is that you don't have it in Linux. This means, of course, that you'll have to write completely different type of drivers for Linux and this seems to be a big trouble for Matrox at the moment...

Just to be technically correct, kernel drivers are always in native code, there is no way to run managed code in kernelland.

What is in managed code is just the fancy little graphical applet used to configure/talk to the driver.

Anyways, drivers are the most platform-specific type of code, there is very very little code that can be shared among different OSs.

Also, drivers are one of the most hard, messy, ugly and expensive pieces of software to code, test and support.

I don't think we will see 64-bit Windows and *nix (100 flavours and counting...) drivers from Matrox unless they have a very, very strong business oportunity for that.

Regards.

* 64bit Windows should be easy to get. If the 32-bit driver is well coded, it just needs to be recompiled. But then you have to test, test, test, and support, support, support...

[Wombat]

Matrox had an IA-64 driver for the G series for a long time.

[leech]

Peter beat me to it, I was going to say that they could impliment the .Net using the Mono project. I think pretty much any new distribution should have packages for it by now. nVidia I know has a GTK interface for their graphics cards. And I believe ATI uses QT for their control panel under linux.

Leech