On Wed, Oct 14, 2009 at 2:33 AM, Eric Anholt <e...@anholt.net> wrote: > On Tue, 2009-10-13 at 13:30 +1000, Dave Airlie wrote: >> From: Dave Airlie <airl...@redhat.com> >> >> Some GPUs (mostly in the lower server end of the market) have memory >> bandwidth limitations that hit in the real world, however memory >> bw calcs require bit depth to be worked out and at the moment, the >> KMS interface doesn't know about bit depth until we set the mode. >> >> This "overloads" the connector_type_id in the get connector ioctl >> to pass a depth value to the kernel code. A new libdrm API is added >> to utilise this. libdrm has always set this to 0 up until now. >> >> This depth value is what the kms user proposed to use, the actual >> API in libdrm is drmModeGetConnectorWithDepth and will follow >> along. >> >> This patch also adds RN50 support to the radeon kms to check >> the memory bandwidth calcs and refuse the higher modes on these >> chips. > > So, is the assertion that on systems that have memory bandwidth > limitations that get hit in practice, there's only one output? >
I hadn't thought much about this but you are right it seems like a limited assertion to make although at this point I'm guessing its true. We have cards with multiple outputs but a single crtc in this case. The two cards we hit this on with regularity are ATI RN50 and MGA G200SE. If we have multi-crtc cards with limitations like this then we really need the enhanced try before you buy modeset interface, as I'm not sure how we could expose this limitation. Dave. ------------------------------------------------------------------------------ Come build with us! The BlackBerry(R) Developer Conference in SF, CA is the only developer event you need to attend this year. Jumpstart your developing skills, take BlackBerry mobile applications to market and stay ahead of the curve. Join us from November 9 - 12, 2009. Register now! http://p.sf.net/sfu/devconference -- _______________________________________________ Dri-devel mailing list Dri-devel@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/dri-devel
