Not logged in
Log in now
Create an account
Subscribe to LWN
LWN.net Weekly Edition for May 23, 2013
An "enum" for Python 3
An unexpected perf feature
LWN.net Weekly Edition for May 16, 2013
A look at the PyPy 2.0 release
Is there any news at all on R300 and/ or R400?
AMD to open up graphics specs
Posted Sep 5, 2007 16:33 UTC (Wed) by tialaramex (subscriber, #21167)
On a Radeon X600 mobile variant in a laptop I'm using to post this for instance I seem to have an Xv scaler and overlay (ie accelerated video playback) and enough 3D to play Quake 3 Arena adequately, though I don't have a mouse plugged in so I got my backside handed to me when I just tried it. The card should be capable of more, but not a huge amount more (mostly more features rather than better performance).
If you have an R3xx or R4xx and thought it didn't work with Free drivers, upgrade to the latest available and try it out for yourself.
Posted Sep 5, 2007 17:02 UTC (Wed) by cathectic (subscriber, #40543)
I would disagree with you on 3D performance though - I believe there is a
lot more that can still be done there on the free side, especially if AMD
are prepared to release the specs on these cards (also, full specs for
things like controlling PowerPlay - I know about "DynamicClocks", but
DynamicClocks != PowerPlay by a long shot).
My question was more to do with the fact that - is there any reason why
AMD cannot open up the R300 and R400 specs as well (this announcement only
says "R500 and newer" - is this an oversight, or is there some other
problem with releasing the older specs?), rather than continue to make
life more difficult that it already is for the reverse engineers and
Posted Sep 5, 2007 17:25 UTC (Wed) by vmole (guest, #111)
Probably because they're not selling r300/400 hardware anymore? Or at least not enough to justify the effort of providing the specs? Remember, it's likely that there is no existing formal specification, just lots of internal e-mail between the hardware and driver developers. One can't really blame them for spending the effort on future hardware rather than "old stuff".
Posted Sep 5, 2007 19:42 UTC (Wed) by AWJ (guest, #47173)
Posted Sep 5, 2007 16:48 UTC (Wed) by Duncan (guest, #6647)
FWIW, I'm running an old Radeon 92xx series, the last of the R200s with
the long time native/freedomware driver based on ATI's sponsored work
before they up and closed everything. As I don't do closed source and am
running an AMD system so Intel's integrated solution isn't an option
unless I upgraded and switched platforms (which I've been planning to do
unless AMD opened up), I've been following the R300/R400 developments with
the intention of upgrading to them when the support was in mainline, thus
my knowledge on the subject, tho it's not direct as I'm not running them
The problem with the R500 and beyond is that ATI killed the old 2D VESA
framebuffer portions of the chip -- it's all 3D, or so I've read. Thus,
even 2D was a major challenge. There has been a recent reverse
engineering project, but it was slow going. Still, ATI hasn't always been
so dead set against freedomware -- as I mentioned they actually sponsored
a third party to do the R200 series drivers.
AMD has been quite open as well, and in fact did very well selling AMD64
kit to Linux and other FLOSS users long before MS got out of beta on the
platform, so when they bought ATI, many of us had high hopes that we'd at
least get specs. Boosting them even farther was AMD's declared intent in
the graphics arena and why they bought ATI in the first place -- an intent
with teeth recently added when they announced their SSE5 stuff --
basically, they want to fully standardize a graphics/physics oriented
instruction set as an extension to x86, much as they did with
AMD64/x86_64. There are whole areas they are targeting based on this,
merging aspects of the GPU/PPU into the CPU or at least into
interpluggable sockets (using Hypertransport), and having the whole
graphics thing continue to function on closed code and closed specs just
didn't seem to jive with where they were going. Thus, many watchers,
certainly including me, were highly optimistic that AMD would open specs
eventually. It just seemed they were sure taking their time, and it was
getting rather harder to maintain optimism, at least here. OTOH, I
suppose they had a lot of internal resistance to overcome on the ATI
Anyway... I've had a convenient budgeting excuse on the graphics card side
for some time, since there hasn't been a lot of freedomware driver
supporting hardware, and what there has been has been bargain bin.
However, it was getting a bit /too/ far behind for my tastes, and I was
looking to update to at least an R300 based card... but reluctantly as
reverse engineering is great, but I'd still be supporting a company that
wasn't cooperating. Now that it seems they may be cooperating, so even if
the generation I may be upgrading to is reverse engineered, I'll at least
be supporting a company cooperating with us once again.
That's certainly a good thing, as it's likely to be another couple years
before I upgrade main platform again (I'm still planning on upgrading this
thing to dual Opteron 290s, and will keep that for at least a couple
years), but I wasn't particularly liking the idea of being forced to Intel
after being an AMD guy since the K6 era, nor of supporting, even if
stop-gap, an uncooperative AMD/ATI in the mean time. It looks like AMD
may have decided to clear up those problems for me (and others like me)
now. As I said, that's a good thing! =8^)
Posted Sep 5, 2007 20:29 UTC (Wed) by joib (guest, #8541)
At the moment I'd say the old trusty radeon 92x0 is a better choice.
Well, one can always hope that the upcoming R500 driver will kick butt.
Posted Sep 6, 2007 5:54 UTC (Thu) by daenzer (✭ supporter ✭, #7050)
Try enabling the 'Disable Low-impact fallback' option in driconf.
Copyright © 2013, Eklektix, Inc.
Comments and public postings are copyrighted by their creators.
Linux is a registered trademark of Linus Torvalds