Weekly Edition
Return to the Announcements pageSponsored link
Vyatta –
Linux & Open Source
Alternative to Cisco –
Advanced Routing,
Firewall, VPN, QoS..
Free Download ->
| |||||||||||
AMD releases initial GPU specs
As noted by David
Airlie, AMD has made an initial set of
specifications for ATI graphics processors available. These are 2D
specifications, so they are not all that is needed to write a complete
graphics driver, but they are a good start.
(Log in to post comments)
AMD releases initial GPU specs Posted Sep 12, 2007 18:58 UTC (Wed) by ofeeley (subscriber, #36105) [Link] It has been argued repeatedly that this would never happen because ATI and Nvidia probably had patent-infringing implementations which would make it foolish for them to open the specs. So what's changed?
AMD releases initial GPU specs Posted Sep 12, 2007 19:43 UTC (Wed) by brouhaha (subscriber, #1698) [Link] That's been argued, but never officially confirmed by either ATI or NVidia. It's possible that register-level documentation does not directly expose anything covered by competitors' patents, and that any such potential patent infringement (if it exists at all) is entirely within the closed-source drivers.
AMD releases initial GPU specs Posted Sep 12, 2007 21:41 UTC (Wed) by khim (subscriber, #9252) [Link] Nope. It's in parts related to DRM. AMD is not opening THIS part and will probably NEVER open it - but the rest of stuff can be opened without any problems...
AMD releases initial GPU specs Posted Sep 12, 2007 23:07 UTC (Wed) by jmorris42 (subscriber, #2203) [Link] > It's in parts related to DRM. AMD is not opening THIS part and> will probably NEVER open it.
Which is just fine by us, what Free Software would enforce the reduced quality output, macrovision, etc flags even if the driver exposed the functionality?
Granted there ARE some potential users who would want that, settop box makers who would be legally obligated, but there is always the closed driver for them to go along with other closed binaries such a product will involve.
AMD releases initial GPU specs Posted Sep 13, 2007 2:35 UTC (Thu) by Richard_J_Neill (subscriber, #23093) [Link] Unfortunately, you're muddling DRM and DRM. In the context of graphics, it means Direct Rendering Manager: http://dri.freedesktop.org/wiki/DRM
AMD releases initial GPU specs Posted Sep 13, 2007 3:43 UTC (Thu) by jmorris42 (subscriber, #2203) [Link] > Unfortunately, you're muddling DRM and DRM.
Nope, I RTFA. AMD/ATI only released 2D info today, but are have promised the 3D stuff will follow. That means DRM (Direct Rendering Manager) stuff will be included. DRM (Data Restriction Managment/Digital Rights "Management") was not mentioned and probably won't be disclosed for the reasons I mentioned above. Plus if they disclosed the details it would make it trivial for Windows hax0rs to use that info to disable the nasty bits which would really annoy powerful interests. Not that they won't do it anyway via reverse engineering....
AMD releases initial GPU specs Posted Sep 12, 2007 19:46 UTC (Wed) by rez (guest, #43474) [Link] "So what's changed?"
Management. :)
AMD releases initial GPU specs Posted Sep 12, 2007 20:52 UTC (Wed) by jengelh (subscriber, #33263) [Link] So what's changed? I suppose it is the same as if Microsoft says Linux violated 235 patents [=vapor]. Or perhaps it is because someone wants to win the extra 178 customers from http://www.pledgebank.com/open3d :-) You know, there is this concept of first strike (not always nuclear-related like WP seems suggesting, can also work for kung fu or anything else).
AMD releases initial GPU specs Posted Sep 13, 2007 13:12 UTC (Thu) by NAR (subscriber, #1313) [Link] I think the patents would affect the algorithms in the (closed source) driver, not the interface of the card.
AMD releases initial GPU specs Posted Sep 13, 2007 8:51 UTC (Thu) by tialaramex (subscriber, #21167) [Link] One of the most hilarious comments I've seen so far was on OS News I think, where a poster wrote “500+ pages of detailed specs is certainly enough to write excellent OpenGL support.” which I suppose tangentially makes the point that this is a lot of paperwork for just the 2D side of the chip.
I'm slightly worried that a skim through the PDFs reveals whole pages with no commentary at all (ie just lists of opaque register names with the comment field left blank). It may be that to video driver developers these registers are either superfluous or their purpose is so obvious that commentary would be redundant but I'd guess it'll be a few days until we know ?
Can anyone say whether the extent of the micro-management in this hardware is usual? e.g. you can control what RGB colour is used for the "black" sent when there is no colour signal, but not just with one register, there's a separate register for overscan, and for blanking, and its duplicated for each CRTC. That's just one example of the sort of thing I'd not expected to find on consumer hardware.
|
|||||||||||