Jeff Garzik wrote:
James Richard Tyrer wrote:
Jeff Garzik wrote:
For many functions, swrast is faster on modern CPUs, particularly
current multi-core CPUs, so its not an easy question with a "does it
have NV 3D accel?" checkbox.
I really don't see your point. If I buy a graphics board with
hardware 3D acceleration and the driver doesn't use it, then I would
say that the driver isn't 'fully functional' -- I might as well stick
with my old Rage 64 based card. This had been my position all along.
My point is that you don't know what you need, if you haven't measured
your current graphics subsystem. It is pointless to support 3D
acceleration if its only going to slow down your OpenGL.
You might have a point. It is possible that with a 3GHz CPU that a
frame buffer or 2D accelerator could beat an older Radeon with 3D
acceleration. But with a < 1GHz system the 3D acceleration would
probably speed things up since that is what it was designed to do.
Regarding "stick with old $x card": newer nv cards have _much_ higher
memory bandwidth, support current AGP and PCI-Express topologies, and
would blow any old ATI Rage card out of the water.
Being sarcastic!
I'll agree with the statement "nv doesn't contain 3D accel", but without
knowing your needs, and measuring on a relevant system, its a useless
statement.
Perhaps what we need to do is clone the ATI R200 chip! :-D
Cloning 3D chips is a complete waste of time. It takes a long time, and
your end result is an outdated chip.
I was being sarcastic. I do that a lot and you tend to miss it. :-) You
missed it even with a ":-D" after it. ARGH!
IAC, what I was referring to would be a black box clone, a chip that
used the same hardware interface although it was implemented completely
different and could probably be faster.
BTW R200 docs are not published either... Nor R300+
IIUC, they did provide some information to assist in writing the driver.
Unfortunately, they have now stopped doing this.
--
JRT
_______________________________________________
Open-graphics mailing list
[email protected]
http://lists.duskglow.com/mailman/listinfo/open-graphics
List service provided by Duskglow Consulting, LLC (www.duskglow.com)
