Commit 236f17a9 authored by Pohjolainen, Topi's avatar Pohjolainen, Topi Committed by Jason Ekstrand

intel/isl/gen7: Use stencil vertical alignment of 8 instead of 4

The reasoning Chad gave in the comment for choosing a valign of 4 is
entirely bunk.  The fact that you have to multiply pitch by 2 is
completely unrelated to the halign/valign parameters used for texture
layout.  (Not completely unrelated.  W-tiling is just Y-tiling with a
bit of extra swizzling which turns 8x8 W-tiled chunks into 16x4 y-tiled
chunks so it makes everything easier if miplevels are always aligned to
8x8.)  The fact that RENDER_SURFACE_STATE::SurfaceVerticalAlignmet
doesn't have a VALIGN_8 option doesn't matter since this is gen7 and you
can't do stencil texturing anyway.

v2 (Jason Ekstrand):
 - Delete most of Chad's comment and add a more descriptive commit
   message.
Signed-off-by: Topi Pohjolainen's avatarTopi Pohjolainen <topi.pohjolainen@intel.com>
Cc: "17.0 17.1" <mesa-stable@lists.freedesktop.org>
Reviewed-by: Jason Ekstrand's avatarJason Ekstrand <jason@jlekstrand.net>
Reviewed-by: chadversary's avatarChad Versace <chadversary@chromium.org>
parent dafc2f18
......@@ -352,30 +352,12 @@ gen7_choose_valign_el(const struct isl_device *dev,
if (isl_surf_usage_is_stencil(info->usage)) {
/* The Ivybridge PRM states that the stencil buffer's vertical alignment
* is 8 [Ivybridge PRM, Volume 1, Part 1, Section 6.18.4.4 Alignment
* Unit Size]. However, valign=8 is outside the set of valid values of
* RENDER_SURFACE_STATE.SurfaceVerticalAlignment, which is VALIGN_2
* (0x0) and VALIGN_4 (0x1).
*
* The PRM is generally confused about the width, height, and alignment
* of the stencil buffer; and this confusion appears elsewhere. For
* example, the following PRM text effectively converts the stencil
* buffer's 8-pixel alignment to a 4-pixel alignment [Ivybridge PRM,
* Volume 1, Part 1, Section
* 6.18.4.2 Base Address and LOD Calculation]:
*
* For separate stencil buffer, the width must be mutiplied by 2 and
* height divided by 2 as follows:
*
* w_L = 2*i*ceil(W_L/i)
* h_L = 1/2*j*ceil(H_L/j)
*
* The root of the confusion is that, in W tiling, each pair of rows is
* interleaved into one.
*
* FINISHME(chadv): Decide to set valign=4 or valign=8 after isl's API
* is more polished.
* Unit Size]. valign=8 is outside the set of valid values of
* RENDER_SURFACE_STATE.SurfaceVerticalAlignment, but that's ok because
* a stencil buffer will never be used directly for texturing or
* rendering on gen7.
*/
require_valign4 = true;
return 8;
}
assert(!require_valign2 || !require_valign4);
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment