On Wed, 2009-08-12 at 01:58 +0300, Pauli Nieminen wrote: > This patch series predicts sizes of emit for next rendering operation. Which > prevents flush while emiting rendering commands. > Improved prediction gives mesa chance to use larger part of ring buffer before > flushing. This should give small performance boost when rendering has. > > Code changes are still only working in r200 driver but all radeon drivers > should > be updated to take advantage of this idea. Patches are breaking others radeon > drivers untill I add patches to update others too. > Code doesn't update max_state_size because I tough removed it would be good > idea. > But then i noticed that it is used to limit possible sizes for ring buffer so > I > didn't remove it yet. > Can someone spot if I forgot something that will get emited? Is there some > call > path that can emit to buffer without going tough pipeline? All comments are > wellcome. >
In IniState you should use = and not +=, for rcommonEnsureCmdBufSpace with AOS we need them because iirc AOS are emited inside the cs cmd stream and last i don't see how this prevent flush while emiting rendering command (my guess being that by removing EnsureCmdBuf in few place you end up seeing that but i think we need them for AOS). Cheers, Jerome ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Mesa3d-dev mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mesa3d-dev
