PDA

View Full Version : accum + GeForce 5900 FX + linux


A027298
10-03-2003, 05:05 AM
Hey Guys!!!

I'm just checking out the accumlation buffer with a 5900 FX ultra..it is sooo slow!!! As far as I thought is that floating point buffers on FX are hardware support. Could it be a driver issue?

Thanks!
Tzupy
10-03-2003, 06:43 AM
Hi,
The accumulation buffer is very slow on all software implementations. The 'floating point buffers' are for the fragment programs.
Does anyone know how to perform antialiasing through multiple passes into a pbuffer, in the same way as it's working with the accumulation buffer, just much faster?
Korval
10-03-2003, 08:38 AM
From what I understood, the 9500+ and FX series of cards had hardware accumulation buffer support, though it was limitted to certain operations.
azazello
10-03-2003, 09:59 AM
just wait for Detonator 50 for Linux.
Under Windowsin 51.xx& 52.xx it's exist and fully hardware implemented.
A027298
10-03-2003, 05:24 PM
Originally posted by Korval:
From what I understood, the 9500+ and FX series of cards had hardware accumulation buffer support, though it was limitted to certain operations.
Yes I have the same understanding in that. I get such a frame rate drop even I am rendering only one sample in to the accumulation buffer.
A027298
10-03-2003, 05:30 PM
Originally posted by ayaromenok:
just wait for Detonator 50 for Linux.
Under Windowsin 51.xx& 52.xx it's exist and fully hardware implemented.

Cool but how long would that take? - Days, weeks, months, ....? :-)
azazello
10-05-2003, 11:52 PM
>>Days, weeks, months, ....? :-)
weeks.. Looks like it will be in time of presentation of NV38/36.

Now you can check your algorithm under Win32 with existing det50.

[This message has been edited by ayaromenok (edited 10-06-2003).]