Our minimum target system is unpatched W2K embedded, so I've been using DX7
for our application. (I keep telling my bosses I'm a programmer, not an
archeologist, but they never listen).

Recently, our artists got bought new machines. But they've started randomly
getting bluescreens when running the application. The "Windows has
recovered from a serious error" screen on reboot blames a device driver, and
the only major difference I could think of was the newer cards. Every other
machine in the office, which is a pretty whide selection of hardware from
the last five years, runs the app fine.

I'm not after a solution as such, but I was wondering if there's any known
issues with older versions of DX on the newer cards. One machine has dual
SLI NVidia 9600s, and the other a NVidia 9800, both running XP SP3.

Cheers,

- SamB

The DX7 rendering path on modern cards is probably not well tested by
anyone. For Windows Vista, the older APIs were all "lofted" onto DX9, so
they should work better on Vista than on XP.

--
Chuck Walbourn
SDE, XNA Developer Connection

This posting is provided "AS IS" with no warranties, and confers no rights.

"Sam Brown" dslIMEANIT.pipexREALLY.com> wrote in message

"Chuck Walbourn [MSFT]" online.microsoft.com> wrote in message
news:%%23P6Ei0xAJHA.5160@TK2MSFTNGP04.phx.gbl...

Yeah, that's kind of what I thought. I just wondered if there was anything
official beyond "DX7 is not supported any more".

Cheers,

- SamB

I did a bit of further looking, and it's not just NVidia - My
four-year-old Radeon 9800 does it as well. But all of them only
exhibit the problem when using a TnL device. A normal HAL is fine,
as is the reference rasterizer.

It _seems_ to be when I'm uploading texture data. Unfortunately,
the drivers on all the systems I have that let me create a TnL
device are all DX9, so no chance of getting any debug output from
them for DX7 calls!

Also, every so often (and usually the first time drawing a new
set of meshes) DrawIndexedPrimitiveVB returns either DDERR_GENERIC
or DDERR_INVALIDPARAMS. However, calling it again immediately
after the failure with the same params results in success.