Xfree86 4.3 and ATI 340M

[Cristiano De Michele]

Does anyone know if new driver for radeon boards of Xfree86 4.3 works with ATI IGP 340M?
thanks
Cristiano

[Bruce Copeland]

I tested 4.3.0 with my Radeon IGP 340M last night. It doesn't appear that 4.3.0 recognizes or enables anything better than previous versions of XFree86 did. This is true at both the functional and the source code levels. The list of PCI Ids in the source code DOES now include the 0x4336 (320M) and 0x4337 (340M) video adapter chips, but none of the drivers do anything with these Ids.

So basically you can use the VESA driver for your IGP 340M, or you can do like I do and force the radeon driver to use ChipID 0x4c59 (an ATI M6) with the Option "noaccel". The noaccel option is necessary because there aren't any kernels (including 2.5.x) that recognize the ATI host bridge (agpgart) used in Radeon IGP 320M or 340M systems. This is the real problem holding back the Radeon IGP systems on Linux. According to the kernel developers, ATI simply hasn't provided the specs necessary to support their host bridges.

Bruce

[Cristiano De Michele]

Hi Bruce,
me again,
I'm about to order the compaq N1020V with P4 1.8-M (I took a decision...maybe :) and I saw
that XFree86 4.3 at least with
some patches can work with IGP 340M (see http://www.wsu.edu/~ice124/)
Anyway I'd like to know if AGP
will work or not.
Looking at config file you can find at URL above
(http://www.wsu.edu/~ice124/XF86Config.radeon.xf4.3) you can see that there is no "noaccel" option in the config file, so I can deduce that AGP will work properly?
Moreover is radeon driver without AGP (if it shouldn't work) in any case faster than VESA driver (at least
acceptably)?

best regards Cristiano

[Bruce Copeland]

Cristiano:

Actually the config file at the site you mention DOES have the Option "noaccel" line in its Device section. So AGP is NOT functional. The author of the site is basically doing the same thing for the RadeonIGP 320M (U1) that I do for the RadeonIGP 340M (See http://www.cybersym.com/pages/linux-ze4100.html
for my XF86Config.)

As far as your question about relative performance of the unaccelerated Radeon vs. the VESA driver. I don't think there's much difference. The Radeon driver has a cleaner startup without all the VESA framebuffer 'residual' image from last use. If you want to play video games under Linux, forget it. It is however plenty fast enough for all routine work (software developement, graphical AI training simulations under java, etc.); it is not fast enough for good molecular graphics simulation. You get the idea.

If you do decide to get the 1020v, make sure and look at

http://www.bandaancha.st/documentos.php?docid=58

Bruce