Date: Fri, 5 Dec 2008 14:36:54 +0000 From: Matt Dawson <matt@chronos.org.uk> To: freebsd-current@freebsd.org Subject: Re: i give up Message-ID: <200812051436.59308.matt@chronos.org.uk> In-Reply-To: <20081204221511.254FE1065758@hub.freebsd.org> References: <20081204221511.254FE1065758@hub.freebsd.org>
next in thread | previous in thread | raw e-mail | index | archive | help
--nextPart1672594.PDyGr4r73Q Content-Type: text/plain; charset="iso-8859-15" Content-Transfer-Encoding: quoted-printable Content-Disposition: inline On Thursday 04 December 2008 22:15:11 freebsd-current-request@freebsd.org=20 wrote: > > I'm not referring to ATI/AMDs proprietary drivers. =A0They are providing > > open documentation now, as well as information that isn't yet public > > when I ask for it. =A0So, I'm speaking about the open drm/mesa/Xorg > > driver, which afaik is working pretty well on r500 and below now. > > > > robert. > > Well, yeah. I was referring to the proprietary drivers though, because > all you get is 2D acceleration with the open ATI/nVidia drivers. > Given the fact that r500 is rather old too though... that lags a lot > more behind nVidia -- but I'm sure that's due to volunteering and lack > of man-hours because nVidia has a few devs dedicated towards > maintaining their driver on *BSD. Actually, no. The subset of ATi cards Robert refers to has direct rendering= =20 support on the to-be-committed (after the ports repo is thawed) Xorg 7.4, D= RM=20 kernel update (which is already in -HEAD) and new Mesa. I have personally=20 tested R200-R480 cards here with good results, after some time messing abou= t=20 with nVidia hardware under the same illusion that they are better supported= =2E=20 Even the troublesome RS48x (Radeon XPress 200/1100M IGP) which I had abando= ned=20 all hope of ever seeing DRM working upon now works with the open driver,=20 although there are still rendering issues with some 3D workloads using port= s'=20 Xorg and Mesa. nVidia's team consists of Christian Zander and another guy (whose name elud= es=20 me right now) as far as I can tell. "Our" team consists of work already don= e=20 by Eric Anholt, by whose own admission no longer uses FreeBSD and most of i= t=20 over 24 months old, and Robert. Also, the open ATi driver/DRI/DRM works=20 (again, verified here) to the same standard on amd64 as it does on i386. Th= is=20 is not the case with nVidia's proprietary driver, so quite how nVidia's=20 offering leads the OSS ATi driver escapes me right now. Robert's work in th= is=20 area over the past few months has brought this hardware, along with many of= =20 the Intel IGPs, almost bang up to date. Only the R6/700 cards lack 3D/2D=20 accel/possibly Xv (I could be wrong on that last item, but the man page sti= ll=20 says no support) but this work is ongoing. One more little niggle in nVidia's direction: The OSS ATi driver integrates= =20 with the base system and ports' Xorg without having to faff about. nVidia's= =20 screws up the system config with odd .so files and symlinks to the point th= at=20 manual intervention is required to upgrade xorg-server and get the linuxula= tor=20 working. Not optimal, IMHO. Bottom line is we have rather good support in -CURRENT for ATi cards and=20 similar in -STABLE with Robert's patches applied. This is probably set to g= et=20 a whole lot better when 7.1-RELEASE shows up (I doubt the DRM kernel stuff= =20 will be MFC'd to 7.1, but I'd like to be wrong) and we get the new stuff in= to=20 ports. When one compares the resource gap between Robert's dedicated volunt= eer=20 efforts and two full-time paid nVidia devs with full docs, this is nothing= =20 short of amazing and I, for one, am very grateful. =2D-=20 Matt Dawson ku.gro.sonorhc@ttam MTD15-RIPE --nextPart1672594.PDyGr4r73Q Content-Type: application/pgp-signature; name=signature.asc Content-Description: This is a digitally signed message part. -----BEGIN PGP SIGNATURE----- Version: GnuPG v2.0.9 (FreeBSD) iEYEABECAAYFAkk5PIsACgkQAmT9uY8euiJKewCfRQP6BLObg1GcVlVVFYEP/3C7 7I8AnA+Q47qHO0HN5EgxlSlde2kVz4C4 =t/cN -----END PGP SIGNATURE----- --nextPart1672594.PDyGr4r73Q--
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200812051436.59308.matt>