Pipedream Demo running on NV34

[Zephyr]

Next week, maybe i can get a NV35 card with these drivers, then I eventually have the chance to test the card with my own ideas and program

BTW, I just test the R300's truform implementation today, it is absolutely a soft one and can eat up all of your cpu resource easily if enabling it it is the reason why the default option for R300 is disable it, IMO.

 

[demalion]

Yep, that's well established.

There's a whole gang of us who bug ATI people about it, with varying levels of courtesy.

Want to join?

 

[Zephyr]

Yep, that's well established.

There's a whole gang of us who bug ATI people about it, with varying levels of courtesy.

Want to join?

If it is an old news, sorry for that I cannot visit b3d forum frequently for recent months, so I think I must miss many interesting things

 

[apoch003]

I would hardly call that "running" the pipe dream demo... even the bar supporting the marimba line is missing...although I can see it's shadow.

Does anyone else notice that the patterns on the animals on the dx9ref test are completely different from the normal rendering of the FX?

Is that a bug?

 

[Xmas]

Does anyone else notice that the patterns on the animals on the dx9ref test are completely different from the normal rendering of the FX?

Is that a bug?

That's the reason I asked about the noise function. DaveBaumann answered by quoting from the 3DMark03 whitepaper, but there seems to be no explanation what "smoothpulsetrain()" really is, and "turbulence" seems to be a load-time generated 3D noise map, that could be different on different CPUs.

But the marble shader has a sine function which could lead to significantly different results in different implementations.

 

[Humus]

A smooth pulse train would be a 1D texture containing a remapping function remapping the texture coord [0..1] to a new number [0..1] through some kind of arbitrary curve. In most cases it's preferrably tileable.