[Mesa-dev] [PATCH 26/26] i965: Enable fast clears for multi-lod

[Pohjolainen, Topi]

On Wed, Oct 12, 2016 at 10:14:44PM -0700, Ben Widawsky wrote:
> On 16-10-12 14:21:08, Matt Turner wrote:
> > On Tue, Oct 11, 2016 at 12:26 PM, Topi Pohjolainen
> > <topi.pohjolainen at gmail.com> wrote:
> > > From: Ben Widawsky <ben at bwidawsk.net>
> > > 
> > > Signed-off-by: Ben Widawsky <benjamin.widawsky at intel.com>
> > > ---
> > >  src/mesa/drivers/dri/i965/intel_mipmap_tree.c | 6 +++---
> > >  1 file changed, 3 insertions(+), 3 deletions(-)
> > > 
> > > diff --git a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > > index 932220e..5769bee 100644
> > > --- a/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > > +++ b/src/mesa/drivers/dri/i965/intel_mipmap_tree.c
> > > @@ -256,13 +256,13 @@ intel_miptree_supports_non_msrt_fast_clear(struct brw_context *brw,
> > >        if (arrayed)
> > >           assert(mt->num_samples <= 1);
> > > 
> > > -      if (mip_mapped && brw->gen >= 8) {
> > > +      if (mip_mapped && brw->gen < 8) {
> > >           perf_debug("Multi-LOD fast clear - giving up (%dx%dx%d).\n",
> > >                      mt->logical_width0, mt->logical_height0, mt->last_level);
> > >           return false;
> > 
> > The perf_debugs indicate something that could be done better, which is
> > the condition was gen >= 8: gen8+ can do multi-LOD fast clears. If we
> > begin supporting that, we should just remove the perf_debug, not start
> > emitting the message on incapable hardware.
> > 
> > >        }
> > > 
> > > -      if (arrayed && brw->gen >= 8) {
> > > +      if (arrayed && brw->gen < 8) {
> > >           perf_debug("Layered fast clear - giving up. (%dx%d%d)\n",
> > >                      mt->logical_width0, mt->logical_height0,
> > >                      mt->physical_depth0);
> > 
> > Same thing here.
> > 
> > > @@ -1637,7 +1637,7 @@ intel_miptree_alloc_non_msrt_mcs(struct brw_context *brw,
> > >        ALIGN(mt->logical_width0, width_divisor) / width_divisor;
> > >     unsigned mcs_height =
> > >        ALIGN(mt->logical_height0, height_divisor) / height_divisor;
> > > -   assert(mt->logical_depth0 == 1);
> > > +   assert(brw->gen >= 8 || mt->logical_depth0 == 1);
> > >     uint32_t layout_flags = MIPTREE_LAYOUT_TILING_Y;
> > > 
> > >     if (brw->gen >= 8) {
> 
> I agree with Matt. I don't actually remember writing this patch (I suspect it
> changed a lot since I wrote the original and Topi was trying to be nice by
> giving me credit).

Right. That perf_debug was supposed to be removed (original from Ben did that
correctly). I must have falsely kept it in one of the many rebases.