Daryll Strauss wrote:
>
> If this happens at random times, then I'd be more suspicious of hardware
> failures than anything else. Either the board overheating or just dieing.
>
It happens very soon after the program starts, the same
board can run (glide?) games happily for long times.
> If it is reproducable, then you can help track it down. Get the Glide
> sources from http://linux.3dfx.com. You should be able to get them to
> build under Windows, hopefully without too much trouble. If you look in
> swlibs/include/<whichever make product you use>/3dfx.mak for the line
> that says # DEBUG = xx. Remove the comment and turn on deubgging.
>
> Glide with debugging on will report a number of bad types of data being
> sent to the board. See where it hangs. It will most likely be hung
> waiting for the FIFO. That means the board itself has hung. The
> instruction that caused the hang will have happened earlier.
>
> The easiest way to hang the fifo is to send a triangle with bad
> values. That means values > 2048 or < -2048. For the V1 and Rush it also
> means triangles that aren't snapped to floating point values of 1/16.
>
The thing that non snapped values could crash it was news to
me, but as Neal mentioned it happens even when Mesa is
compiled for snapping. Also I think I have
seen it freeze once during my "texture load"
phase, before drawing.
Anyways I will download the source and see if I can compile
it.
Eero
_______________________________________________
Mesa-dev maillist - [EMAIL PROTECTED]
http://lists.mesa3d.org/mailman/listinfo/mesa-dev
