From 016e357c752e3aed1094c664e7387ebed9738f03 Mon Sep 17 00:00:00 2001
From: Subv <subv2112@gmail.com>
Date: Wed, 4 Jul 2018 10:42:33 -0500
Subject: [PATCH] GPU: Implemented the 32 bit float depth buffer format.

---
 src/video_core/renderer_opengl/gl_rasterizer_cache.cpp | 6 ++++--
 src/video_core/renderer_opengl/gl_rasterizer_cache.h   | 9 +++++++++
 src/video_core/textures/decoders.cpp                   | 2 ++
 3 files changed, 15 insertions(+), 2 deletions(-)

diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
index 50469c05c..57d7763ff 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.cpp
@@ -108,7 +108,8 @@ static constexpr std::array<FormatTuple, SurfaceParams::MaxPixelFormat> tex_form
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
      false}, // Z24S8
     {GL_DEPTH24_STENCIL8, GL_DEPTH_STENCIL, GL_UNSIGNED_INT_24_8, ComponentType::UNorm,
-     false}, // S8Z24
+     false},                                                                            // S8Z24
+    {GL_DEPTH_COMPONENT32F, GL_DEPTH_COMPONENT, GL_FLOAT, ComponentType::Float, false}, // Z32F
 }};
 
 static const FormatTuple& GetFormatTuple(PixelFormat pixel_format, ComponentType component_type) {
@@ -191,7 +192,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<true, PixelFormat::DXT1>,         MortonCopy<true, PixelFormat::DXT23>,
         MortonCopy<true, PixelFormat::DXT45>,        MortonCopy<true, PixelFormat::DXN1>,
         MortonCopy<true, PixelFormat::ASTC_2D_4X4>,  MortonCopy<true, PixelFormat::Z24S8>,
-        MortonCopy<true, PixelFormat::S8Z24>,
+        MortonCopy<true, PixelFormat::S8Z24>,        MortonCopy<true, PixelFormat::Z32F>,
 };
 
 static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
@@ -213,6 +214,7 @@ static constexpr std::array<void (*)(u32, u32, u32, u8*, Tegra::GPUVAddr),
         MortonCopy<false, PixelFormat::ABGR8>,
         MortonCopy<false, PixelFormat::Z24S8>,
         MortonCopy<false, PixelFormat::S8Z24>,
+        MortonCopy<false, PixelFormat::Z32F>,
 };
 
 // Allocate an uninitialized texture of appropriate size and format for the surface
diff --git a/src/video_core/renderer_opengl/gl_rasterizer_cache.h b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
index 8005a81b8..b4d7f8ebe 100644
--- a/src/video_core/renderer_opengl/gl_rasterizer_cache.h
+++ b/src/video_core/renderer_opengl/gl_rasterizer_cache.h
@@ -42,6 +42,7 @@ struct SurfaceParams {
         // DepthStencil formats
         Z24S8 = 13,
         S8Z24 = 14,
+        Z32F = 15,
 
         MaxDepthStencilFormat,
 
@@ -94,6 +95,7 @@ struct SurfaceParams {
             4, // ASTC_2D_4X4
             1, // Z24S8
             1, // S8Z24
+            1, // Z32F
         }};
 
         ASSERT(static_cast<size_t>(format) < compression_factor_table.size());
@@ -120,6 +122,7 @@ struct SurfaceParams {
             32,  // ASTC_2D_4X4
             32,  // Z24S8
             32,  // S8Z24
+            32,  // Z32F
         }};
 
         ASSERT(static_cast<size_t>(format) < bpp_table.size());
@@ -135,6 +138,8 @@ struct SurfaceParams {
             return PixelFormat::S8Z24;
         case Tegra::DepthFormat::Z24_S8_UNORM:
             return PixelFormat::Z24S8;
+        case Tegra::DepthFormat::Z32_FLOAT:
+            return PixelFormat::Z32F;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
@@ -235,6 +240,8 @@ struct SurfaceParams {
             return Tegra::DepthFormat::S8_Z24_UNORM;
         case PixelFormat::Z24S8:
             return Tegra::DepthFormat::Z24_S8_UNORM;
+        case PixelFormat::Z32F:
+            return Tegra::DepthFormat::Z32_FLOAT;
         default:
             UNREACHABLE();
         }
@@ -284,6 +291,8 @@ struct SurfaceParams {
         case Tegra::DepthFormat::S8_Z24_UNORM:
         case Tegra::DepthFormat::Z24_S8_UNORM:
             return ComponentType::UNorm;
+        case Tegra::DepthFormat::Z32_FLOAT:
+            return ComponentType::Float;
         default:
             LOG_CRITICAL(HW_GPU, "Unimplemented format={}", static_cast<u32>(format));
             UNREACHABLE();
diff --git a/src/video_core/textures/decoders.cpp b/src/video_core/textures/decoders.cpp
index 7b06fea3e..d5ab4e4f9 100644
--- a/src/video_core/textures/decoders.cpp
+++ b/src/video_core/textures/decoders.cpp
@@ -78,6 +78,7 @@ static u32 DepthBytesPerPixel(DepthFormat format) {
     switch (format) {
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
+    case DepthFormat::Z32_FLOAT:
         return 4;
     default:
         UNIMPLEMENTED_MSG("Format not implemented");
@@ -132,6 +133,7 @@ std::vector<u8> UnswizzleDepthTexture(VAddr address, DepthFormat format, u32 wid
     switch (format) {
     case DepthFormat::S8_Z24_UNORM:
     case DepthFormat::Z24_S8_UNORM:
+    case DepthFormat::Z32_FLOAT:
         CopySwizzledData(width, height, bytes_per_pixel, bytes_per_pixel, data,
                          unswizzled_data.data(), true, block_height);
         break;