[Mesa-dev] [PATCH 3/3] radv: enable TC-compat HTILE for 16-bit depth surfaces on GFX8

[Bas Nieuwenhuizen]

Reviewed-by: Bas Nieuwenhuizen <bas at basnieuwenhuizen.nl>

for the series.

On Wed, Mar 21, 2018 at 9:30 PM, Samuel Pitoiset
<samuel.pitoiset at gmail.com> wrote:
> The hardware only supports 32-bit depth surfaces, but we can
> enable TC-compat HTILE for 16-bit depth surfaces if no Z planes
> are compressed.
>
> The main benefit is to reduce the number of depth decompression
> passes. Also, we don't need to implement DB->CB copies which is
> fine.
>
> This improves Serious Sam 2017 by +4%. Talos and F12017 are also
> affected but I don't see a performance difference.
>
> This also improves the shadowmapping Vulkan demo by 10-15%
> (FPS is now similar to AMDVLK).
>
> No CTS regressions on Polaris10.
>
> Signed-off-by: Samuel Pitoiset <samuel.pitoiset at gmail.com>
> ---
>  src/amd/vulkan/radv_device.c | 22 ++++++++++++++++------
>  src/amd/vulkan/radv_image.c  | 20 ++++++++------------
>  2 files changed, 24 insertions(+), 18 deletions(-)
>
> diff --git a/src/amd/vulkan/radv_device.c b/src/amd/vulkan/radv_device.c
> index 22500bfc13..9c82fd059f 100644
> --- a/src/amd/vulkan/radv_device.c
> +++ b/src/amd/vulkan/radv_device.c
> @@ -3615,12 +3615,22 @@ radv_calc_decompress_on_z_planes(struct radv_device *device,
>
>                 max_zplanes = max_zplanes + 1;
>         } else {
> -               if (iview->image->info.samples <= 1)
> -                       max_zplanes = 5;
> -               else if (iview->image->info.samples <= 4)
> -                       max_zplanes = 3;
> -               else
> -                       max_zplanes = 2;
> +               if (iview->vk_format == VK_FORMAT_D16_UNORM) {
> +                       /* Do not enable Z plane compression for 16-bit depth
> +                        * surfaces because isn't supported on GFX8. Only
> +                        * 32-bit depth surfaces are supported by the hardware.
> +                        * This allows to maintain shader compatibility and to
> +                        * reduce the number of depth decompressions.
> +                        */
> +                       max_zplanes = 1;
> +               } else {
> +                       if (iview->image->info.samples <= 1)
> +                               max_zplanes = 5;
> +                       else if (iview->image->info.samples <= 4)
> +                               max_zplanes = 3;
> +                       else
> +                               max_zplanes = 2;
> +               }
>         }
>
>         return max_zplanes;
> diff --git a/src/amd/vulkan/radv_image.c b/src/amd/vulkan/radv_image.c
> index 6e5f3e7ad0..dd3189c67d 100644
> --- a/src/amd/vulkan/radv_image.c
> +++ b/src/amd/vulkan/radv_image.c
> @@ -91,18 +91,14 @@ radv_image_is_tc_compat_htile(struct radv_device *device,
>             pCreateInfo->format == VK_FORMAT_D32_SFLOAT_S8_UINT)
>                 return false;
>
> -       if (device->physical_device->rad_info.chip_class >= GFX9) {
> -               /* GFX9 supports both 32-bit and 16-bit depth surfaces. */
> -               if (pCreateInfo->format != VK_FORMAT_D32_SFLOAT_S8_UINT &&
> -                   pCreateInfo->format != VK_FORMAT_D32_SFLOAT &&
> -                   pCreateInfo->format != VK_FORMAT_D16_UNORM)
> -                       return false;
> -       } else {
> -               /* GFX8 only supports 32-bit depth surfaces. */
> -               if (pCreateInfo->format != VK_FORMAT_D32_SFLOAT_S8_UINT &&
> -                   pCreateInfo->format != VK_FORMAT_D32_SFLOAT)
> -                       return false;
> -       }
> +       /* GFX9 supports both 32-bit and 16-bit depth surfaces, while GFX8 only
> +        * supports 32-bit. Though, it's possible to enable TC-compat for
> +        * 16-bit depth surfaces if no Z planes are compressed.
> +        */
> +       if (pCreateInfo->format != VK_FORMAT_D32_SFLOAT_S8_UINT &&
> +           pCreateInfo->format != VK_FORMAT_D32_SFLOAT &&
> +           pCreateInfo->format != VK_FORMAT_D16_UNORM)
> +               return false;
>
>         return true;
>  }
> --
> 2.16.2
>
> _______________________________________________
> mesa-dev mailing list
> mesa-dev at lists.freedesktop.org
> https://lists.freedesktop.org/mailman/listinfo/mesa-dev