From: Roland S. <rsc...@sw...> - 2003-04-01 02:23:01
|
Frank Van Damme wrote: > On Monday 31 March 2003 19:18, Roland Scheidegger wrote: > >>you could try without agpgart, you can force pci gart >>(http://www.xfree86.org/4.3.0/radeon.4.html) . Not sure if this works >>reasonably, IIRC there were some other issues with that igp apart from >>the agp bridge. > > > Wait a minute. Are you saying that I might be able to use an agp card reliably > in pci mode even if he agp driver for my motherboard sucks? Or doesn't it > work for dri? In prinicple, yes. As for "reliable" I'm not sure - there might be a reason it isn't enabled as default. I've just noticed pcigart is still #ifdefed out in radeon_cp.c, at least in the dri trunk (and if you use the new forcepcimode option XFree86 will just hang at startup filling its logfile rapidly...). But just out of curiousity, I've changed the #ifdef and enabled it. Seems to work but the performance hit is _HUGE_ (ut, quake3 are about a factor of 3 slower). That's with a via kt133a (insmoded radeon manually to be certain agpgart is not loaded), radeon 9000 pro. Looks like this isn't really an option for those poor nforce2 owners... And the driver needs to support this - the only ones I'm aware of which do are the radeon and r128. Nvidias binary driver also should work without agp support (at least you can disable it). But, as others have pointed out, it won't help the OP since the IGP still won't work, though I don't know what the differences to a standard radeon 7000 are (might need agp texturing?). Roland btw what board do you have which agp drivers sucks? I'm only aware of some boards which don't have an agp driver at all, or which don't run stable with all features (4x, sba, fw) enabled... |