My site is up again and with a new volumetric fogging demo

[Humus]

I'm still on a 8500. I have ordered a 9700 directly from ATi with the 40% developer discount one and a half month ago, it still hasn't arrived even though I've reminded them several times. I was told like a week or two ago when I reminded them the last time that it should ship that day, but apparently it didn't. Can't see what's holding it back ...

... Goddammit ATI! Which reminds me: Are you still set for that job at ATI in Canada? There was a work permit problem I believe? (crazy Canadians!)

Heh, I've just got my 9700 and it's running perfectly fine in my system. I'm really impressed by the performance this thing has Has been playing Q3 in 2048x1536x32 with everything on max and still got 100fps I get roughly the same performance in 1600x1200x32 with 6xAA and 16xAniso. Similar in UT. UT2003 demanded a little more though, but it's still smooth at 1152x864x32 with all details and 6xAA + 16xAniso.

I'm still set for the ATi job, unless it turns out I can't get a work permit for the next year either.

 

[Humus]

I think the problem with multimonitor should be solved now. I've uploaded a new version that should work, tell me if it does.

 

[Brent]

I think the problem with multimonitor should be solved now. I've uploaded a new version that should work, tell me if it does.

Nope

it starts fullscreen but still says i am missing the ogl support...

 

[Humus]

Ack. Oh, well. Guess that means I'll have to borrow my roommates monitor and do some debugging some day. The hard part will be to get him far enough from the computer first

 

[Tonyo]

Re: The demo doesn't get a retained-alpha pixelformat

Humus, AFAIK the problem is that for some reason you don't get a pixelformat with retained alpha (i.e. with destination alpha).
I tested your demo in a Wildcat VP and I had to force the driver to return a pixelformat with alpha in order to make it work, otherwise I would get the same image as the screenshot posted here.
What are the flags you pass to ChoosePixelformat?.

Nice demo, though

	static int iAtt[] = {
		WGL_DRAW_TO_WINDOW_ARB, GL_TRUE,
		WGL_ACCELERATION_ARB,   WGL_FULL_ACCELERATION_ARB,
		WGL_RED_BITS_ARB,       8,
		WGL_GREEN_BITS_ARB,     8,
		WGL_BLUE_BITS_ARB,      8,
		WGL_ALPHA_BITS_ARB,     8,
		WGL_DEPTH_BITS_ARB,     24,
		WGL_STENCIL_BITS_ARB,   0,
		WGL_DOUBLE_BUFFER_ARB,  GL_TRUE,
		0, 0
	};

I saw that code, but I guess that's the code you use if you support WGL_ARB_pixel_format, which the Wildcat VP does not (it doesn't export WGL_ARB_extensions_string), so in the Wildcat VP case the demo is using a different pixelformat selection path, which is the one causing not to request alpha.

I don't think it's the alpha, without dest alpha the first fog volume wouldn't work either.

Then it could be that the problem in the Wildcat VP was a different one from the one in the Radeon, but the images look exactly the same.

 

[Humus]

It doesn't support WGL_ARB_pixel_format? Hmm, that's unexpected. I've looked through the fallback code and I can see why it doesn't get any alpha. I've fixed that. But the 9700 isn't taking that path, but it might still be something with the alpha though.

 

[Humus]

Alright, I've uploaded a new version now. The multimonitor bug is fixed. You can now set the volume on the console too.

 

[Xmas]

Doesn't the R9700 also support 10:10:10:2 RGBA? Maybe it's about alpha precision.

 

[Humus]

Yes, but it's not that. If I request 8 bit alpha the driver can't give me anything less, the WGL_ARB_pixel_format specifies that. If I were to ask for 2 bit alpha they could give me the 10:10:10:2 mode.
Anyway, I tracked down the bug to be a certain texture environment setup which didn't work. It's supposed to do a alpha replace but let the RGB pass through, it returned all 1 on the alpha though. However, this bug should be fixed in the next driver. There already exists nonreleased drivers for which is works.