PDA

View Full Version : testing p-buffers, RTT,... plenty of bugs



V-man
10-22-2003, 10:55 AM
I was having some problems with creating p-buffers and RTT so I created this little app.
http://www.geocities.com/vmelkon/pbuffercreator.html

I think there may be a few bugs in Catalyst 3.8

For example, if I ask for a [WGL_ACCELERATION_ARB, TRUE]
or
[WGL_ACCELERATION_ARB, FALSE]

it wglChoosePixelFormatARB returns 0 pixelformats, but if I remove that from my list, it succeeds. huh?

Other than that, RTT cubemaps doesn't work. All my textures dissapear. As if texturing is disabled.

glGetTexImage doesnt work right for cubemaps.

glCopyTexSubImage goes into an infinite loop in one case.

fudge!

nonick
10-23-2003, 01:50 AM
WGL_FULL_ACCELERATION_ARB, WGL_GENERIC_ACCELERATION_ARB and WGL_NO_ACCELERATION_ARB are the only correct values for WGL_ACCELERATION_ARB (TRUE/FALSE are erroneous)

V-man
10-24-2003, 07:27 AM
Yea, sorry about that. The code I have is actually *correct*.
It asks for FULL_ACCEL and NO_ACCEL if you use the (bug) items from the menu. The code just keeps on trying until it's time to give up.

I wanted to hear from people if it was working OK. Specially people on ATI with Catalyst drivers.

elias
10-24-2003, 08:16 AM
It seems to work here - all 4 menu points state that pbuffer allocation succeeds 2 times.

I have catalyst 3.8 drivers on a radeon 9800 pro.

- elias

[This message has been edited by elias (edited 10-24-2003).]

SFRuckus
10-24-2003, 08:24 AM
Try changing the swap method. I had problems on a 9800pro with "swap exchange" and full acceleration. "DontCare" worked.

- SFRuckus

V-man
10-24-2003, 09:08 AM
I dont ask for double buffering so that's not it.

I ask for 32 color, 32/24 depth, 8 stencil,
8 red, 8 green, 8 blue, 8 alpha, single buffer and other generic ones like (support opengl, ...)

It's suppose to succeed on the 2nd try normally, because the first asks for 32 bpp z and the second asks for 24 bpp z-buffer.

I have Catalyst 3.8 and strangely enough, if I ask for (ACCELERATION, FULL or NONE) for 2D RTT, it fails (0 pixel formats), but for cubemaps it works fine.

PS: That would be menu item #4

GKW
10-24-2003, 10:06 AM
I was under the impression that pbuffers are always hardware accelerated and there is no need to ask for it.