| Weekly edition | Kernel | Security | Distributions | Search |
| Archives | Calendar | Subscribe | Write for LWN | LWN.net FAQ |
AMD's open GPU documentation site
AMD's open GPU documentation site
Posted Feb 12, 2008 17:40 UTC (Tue) by jwb (subscriber, #15467)In reply to: AMD's open GPU documentation site by kraftcheck
Parent article: AMD's open GPU documentation site
It took Intel a year and a half to produce documentation for the G965. Perhaps it will take AMD about the same amount of time.
(Log in to post comments)
AMD's open GPU documentation site
Posted Feb 12, 2008 20:05 UTC (Tue) by tajyrink (subscriber, #2750) [Link]
The rumor is specs for 2D acceleration and basic 3D should be just around the corner (really). They are aiming for "basic" (read: Compiz, Nexuiz etc.) 3D support sooner and then eventually a kicking open source 3D support via the emerging Gallium architecture. XAA and EXA 2D accelerations are now already working on R500, R600 is the one that's a really different beast. Best source for news seems to be http://www.radeonhd.org/ that is related to the Phoronix site.
AMD's open GPU documentation site
Posted Feb 13, 2008 17:26 UTC (Wed) by jmayer (subscriber, #595) [Link]
> XAA and EXA 2D accelerations are now already working on R500 Well, not really or not in the way most readers would expect: They have been implemented in the driver, so yes, they run, but currently run without any hardware support. This means that anyone running XAA or EXA get *less* performance than someone running just ShadowFB. See http://lists.opensuse.org/radeonhd/2008-01/msg00233.html for details.
AMD's open GPU documentation site
Posted Feb 23, 2008 17:50 UTC (Sat) by daenzer (subscriber, #7050) [Link]
Where did you read 'without any hardware support' there? Both XAA and EXA would be totally pointless in that case of course. The benchmarks where they're slower than ShadowFB exercise paths that the driver doesn't accelerate yet (mostly related to the RENDER extension - that can be fixed now that AMD released 3D documentation), so the acceleration architectures have to pay a price for trying to accelerate things instead of just doing them with the CPU in system memory in the first place. It shouldn't be hard to find other benchmarks where hardware acceleration is much faster than ShadowFB. Which of the options works best in practice depends on the usage pattern, so everybody has to find out for themselves.
Time is not on their side
Posted Feb 13, 2008 9:17 UTC (Wed) by alex (subscriber, #1355) [Link]
It does seem to be that it takes a long time for the docs to get out there. At least I'm pretty confident that it's coming. However having the docs come out long after the chipset has debuted isn't going to help with design wins that want to know a open driver will be ready when the hardware ships. I still find it odd that the Dell Linux machines come with nVidia graphics because that's the most stable state of the art for the time being.