On Wed, Sep 23, 2015 at 11:12:26PM +0100, Neil Roberts wrote: > Ben Widawsky <b...@bwidawsk.net> writes: > > > Hmm. As I read it, it sounded like you didn't have to send LOD it's > > implied to be 0 if you don't send it. If I am wrong about that, then I > > agree with you completely. > > I'm a bit lost. You're right that it's not necessary to send the LOD > when it's zero. In fact Mesa never sends it for the TXF_MCS instruction > presumably because OpenGL doesn't provide a way to make a texture that > is both multisampled and mipmapped so the LOD will always be zero. > However I don't see how that relates to deciding whether to always use > the ld2dms_w instruction or not. The reason I am saying we shouldn't > always do it is because it makes the sampler message longer and thus > requires at least one extra instruction to initialise the extra argument > in the message. The extra argument is due to the second MCS argument > (mcsh), not the LOD. I don't understand the point you are trying to make > about the LOD. > > Even without the LOD, the arguments are potentially up to 6 > (si,mcsl,mcsh,u,v,r). On SIMD16 this takes up 12 registers which is > greater than the maximum of 11 so it has to be broken up into two SIMD8 > messages. This only matters for sampler2DMSArray samplers because > otherwise the r argument is not needed. > > Apologies if I'm being dumb. > > Regards, > - Neil
Hmm. For some reason I was thinking we could use 12 registers, and LOD was the cause of us going over. My bad. _______________________________________________ mesa-dev mailing list mesa-dev@lists.freedesktop.org http://lists.freedesktop.org/mailman/listinfo/mesa-dev
