vulkan: Implement FidelityFX Super Resolution

src/common/settings.h

@@ -64,6 +64,7 @@ enum class ScalingFilter : u32 {
     Bilinear = 0,
     Bicubic = 1,
     ScaleForce = 2,
+    Fsr = 3,
 };
 
 struct ResolutionScalingInfo {

src/video_core/CMakeLists.txt

@@ -132,6 +132,8 @@ add_library(video_core STATIC
     renderer_vulkan/vk_descriptor_pool.h
     renderer_vulkan/vk_fence_manager.cpp
     renderer_vulkan/vk_fence_manager.h
+    renderer_vulkan/vk_fsr.cpp
+    renderer_vulkan/vk_fsr.h
     renderer_vulkan/vk_graphics_pipeline.cpp
     renderer_vulkan/vk_graphics_pipeline.h
     renderer_vulkan/vk_master_semaphore.cpp

src/video_core/host_shaders/CMakeLists.txt

@@ -1,3 +1,11 @@
+set(FIDELITYFX_INCLUDE_DIR ${CMAKE_SOURCE_DIR}/externals/FidelityFX-FSR/ffx-fsr)
+
+set(GLSL_INCLUDES
+    fidelityfx_fsr.comp
+    ${FIDELITYFX_INCLUDE_DIR}/ffx_a.h
+    ${FIDELITYFX_INCLUDE_DIR}/ffx_fsr1.h
+)
+
 set(SHADER_FILES
     astc_decoder.comp
     block_linear_unswizzle_2d.comp
@@ -13,6 +21,8 @@ set(SHADER_FILES
     present_bicubic.frag
     vulkan_blit_color_float.frag
     vulkan_blit_depth_stencil.frag
+    vulkan_fidelityfx_fsr_easu.comp
+    vulkan_fidelityfx_fsr_rcas.comp
     vulkan_present.frag
     vulkan_present.vert
     vulkan_quad_indexed.comp
@@ -78,7 +88,7 @@ foreach(FILENAME IN ITEMS ${SHADER_FILES})
             OUTPUT
                 ${SPIRV_HEADER_FILE}
             COMMAND
-                ${GLSLANGVALIDATOR} -V ${QUIET_FLAG} ${GLSL_FLAGS} --variable-name ${SPIRV_VARIABLE_NAME} -o ${SPIRV_HEADER_FILE} ${SOURCE_FILE}
+                ${GLSLANGVALIDATOR} -V ${QUIET_FLAG} -I"${FIDELITYFX_INCLUDE_DIR}" ${GLSL_FLAGS} --variable-name ${SPIRV_VARIABLE_NAME} -o ${SPIRV_HEADER_FILE} ${SOURCE_FILE}
             MAIN_DEPENDENCY
                 ${SOURCE_FILE}
         )
@@ -86,9 +96,12 @@ foreach(FILENAME IN ITEMS ${SHADER_FILES})
     endif()
 endforeach()
 
+set(SHADER_SOURCES ${SHADER_FILES})
+list(APPEND SHADER_SOURCES ${GLSL_INCLUDES})
+
 add_custom_target(host_shaders
     DEPENDS
         ${SHADER_HEADERS}
     SOURCES
-        ${SHADER_FILES}
+        ${SHADER_SOURCES}
 )

src/video_core/host_shaders/fidelityfx_fsr.comp

@@ -0,0 +1,114 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+//!#version 460 core
+#extension GL_ARB_separate_shader_objects : enable
+#extension GL_ARB_shading_language_420pack : enable
+#extension GL_GOOGLE_include_directive : enable
+#extension GL_EXT_shader_explicit_arithmetic_types : require
+
+// FidelityFX Super Resolution Sample
+//
+// Copyright (c) 2021 Advanced Micro Devices, Inc. All rights reserved.
+// Permission is hereby granted, free of charge, to any person obtaining a copy
+// of this software and associated documentation files(the "Software"), to deal
+// in the Software without restriction, including without limitation the rights
+// to use, copy, modify, merge, publish, distribute, sublicense, and / or sell
+// copies of the Software, and to permit persons to whom the Software is
+// furnished to do so, subject to the following conditions :
+// The above copyright notice and this permission notice shall be included in
+// all copies or substantial portions of the Software.
+// THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+// IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+// FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT.IN NO EVENT SHALL THE
+// AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+// LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+// OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
+// THE SOFTWARE.
+
+layout( push_constant ) uniform constants {
+    u32vec2 input_size;
+};
+
+uvec4 Const0;
+uvec4 Const1;
+uvec4 Const2;
+uvec4 Const3;
+
+#define A_GPU 1
+#define A_GLSL 1
+#define A_HALF
+
+#include "ffx_a.h"
+
+f16vec4 LinearToSRGB(f16vec4 linear) {
+    bvec4 selector = greaterThan(linear, f16vec4(0.00313066844250063));
+    f16vec4 low = linear * float16_t(12.92);
+    f16vec4 high = float16_t(1.055) * pow(linear, f16vec4(1 / 2.4)) - float16_t(0.055);
+    return mix(low, high, selector);
+}
+
+f16vec4 SRGBToLinear(f16vec4 srgb) {
+    bvec4 selector = greaterThan(srgb, f16vec4(0.0404482362771082));
+    f16vec4 low = srgb * float16_t(1.0 / 12.92);
+    f16vec4 high = pow((srgb + float16_t(0.055)) * float16_t(1.0 / 1.055), f16vec4(2.4));
+    return mix(low, high, selector);
+}
+
+#if USE_EASU
+    #define FSR_EASU_H 1
+    f16vec4 FsrEasuRH(vec2 p) { f16vec4 res = f16vec4(textureGather(InputTexture, p, 0)); return res; }
+    f16vec4 FsrEasuGH(vec2 p) { f16vec4 res = f16vec4(textureGather(InputTexture, p, 1)); return res; }
+    f16vec4 FsrEasuBH(vec2 p) { f16vec4 res = f16vec4(textureGather(InputTexture, p, 2)); return res; }
+#endif
+#if USE_RCAS
+    #define FSR_RCAS_H 1
+    f16vec4 FsrRcasLoadH(ASW2 p) { return f16vec4(texelFetch(InputTexture, ASU2(p), 0)); }
+    void FsrRcasInputH(inout float16_t r, inout float16_t g, inout float16_t b) {}
+#endif
+
+#include "ffx_fsr1.h"
+
+void CurrFilter(u32vec2 pos) {
+    // For debugging
+#if USE_BILINEAR
+    vec2 pp = (vec2(pos) * vec2_AU2(Const0.xy) + vec2_AU2(Const0.zw)) * vec2_AU2(Const1.xy) + vec2(0.5, -0.5) * vec2_AU2(Const1.zw);
+    imageStore(OutputTexture, ivec2(pos), textureLod(InputTexture, pp, 0.0));
+#endif
+#if USE_EASU
+    f16vec3 c;
+    FsrEasuH(c, pos, Const0, Const1, Const2, Const3);
+    imageStore(OutputTexture, ivec2(pos), f16vec4(c, 1));
+#endif
+#if USE_RCAS
+    f16vec3 c;
+    FsrRcasH(c.r, c.g, c.b, pos, Const0);
+    imageStore(OutputTexture, ivec2(pos), f16vec4(c, 1));
+#endif
+
+}
+
+layout(local_size_x=64) in;
+void main() {
+
+#if USE_EASU || USE_BILINEAR
+    vec2 ires = vec2(input_size);
+    vec2 tres = textureSize(InputTexture, 0);
+    vec2 ores = imageSize(OutputTexture);
+    FsrEasuCon(Const0, Const1, Const2, Const3, ires.x, ires.y, tres.x, tres.y, ores.x, ores.y);
+#endif
+#if USE_RCAS
+    FsrRcasCon(Const0, 0.25f);
+#endif
+
+    // Do remapping of local xy in workgroup for a more PS-like swizzle pattern.
+    AU2 gxy = ARmp8x8(gl_LocalInvocationID.x) + AU2(gl_WorkGroupID.x << 4u, gl_WorkGroupID.y << 4u);
+    CurrFilter(gxy);
+    gxy.x += 8u;
+    CurrFilter(gxy);
+    gxy.y += 8u;
+    CurrFilter(gxy);
+    gxy.x -= 8u;
+    CurrFilter(gxy);
+}

src/video_core/host_shaders/vulkan_fidelityfx_fsr_easu.comp

@@ -0,0 +1,13 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 460 core
+#extension GL_GOOGLE_include_directive : enable
+
+layout(set=0,binding=0) uniform sampler2D InputTexture;
+layout(set=0,binding=1,rgba16f) uniform image2D OutputTexture;
+
+#define USE_EASU 1
+
+#include "fidelityfx_fsr.comp"

src/video_core/host_shaders/vulkan_fidelityfx_fsr_rcas.comp

@@ -0,0 +1,13 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#version 460 core
+#extension GL_GOOGLE_include_directive : enable
+
+layout(set=0,binding=0) uniform sampler2D InputTexture;
+layout(set=0,binding=1,rgba16f) uniform image2D OutputTexture;
+
+#define USE_RCAS 1
+
+#include "fidelityfx_fsr.comp"

src/video_core/renderer_vulkan/vk_blit_screen.cpp

@@ -23,6 +23,7 @@
 #include "video_core/host_shaders/vulkan_present_vert_spv.h"
 #include "video_core/renderer_vulkan/renderer_vulkan.h"
 #include "video_core/renderer_vulkan/vk_blit_screen.h"
+#include "video_core/renderer_vulkan/vk_fsr.h"
 #include "video_core/renderer_vulkan/vk_master_semaphore.h"
 #include "video_core/renderer_vulkan/vk_scheduler.h"
 #include "video_core/renderer_vulkan/vk_shader_util.h"
@@ -147,8 +148,12 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer,
     scheduler.Wait(resource_ticks[image_index]);
     resource_ticks[image_index] = scheduler.CurrentTick();
 
-    UpdateDescriptorSet(image_index,
-                        use_accelerated ? screen_info.image_view : *raw_image_views[image_index]);
+    const VkImageView source_image_view =
+        use_accelerated ? screen_info.image_view : *raw_image_views[image_index];
+
+    if (!fsr) {
+        UpdateDescriptorSet(image_index, source_image_view);
+    }
 
     BufferData data;
     SetUniformData(data, layout);
@@ -225,9 +230,26 @@ VkSemaphore VKBlitScreen::Draw(const Tegra::FramebufferConfig& framebuffer,
                                    read_barrier);
             cmdbuf.CopyBufferToImage(*buffer, image, VK_IMAGE_LAYOUT_GENERAL, copy);
             cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_TRANSFER_BIT,
-                                   VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, write_barrier);
+                                   VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT |
+                                       VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                                   0, write_barrier);
         });
     }
+
+    if (fsr) {
+        auto crop_rect = framebuffer.crop_rect;
+        if (crop_rect.GetWidth() == 0) {
+            crop_rect.right = framebuffer.width;
+        }
+        if (crop_rect.GetHeight() == 0) {
+            crop_rect.bottom = framebuffer.height;
+        }
+        crop_rect = crop_rect.Scale(Settings::values.resolution_info.up_factor);
+        VkImageView fsr_image_view =
+            fsr->Draw(scheduler, image_index, source_image_view, crop_rect);
+        UpdateDescriptorSet(image_index, fsr_image_view);
+    }
+
     scheduler.Record(
         [this, host_framebuffer, image_index, size = render_area](vk::CommandBuffer cmdbuf) {
             const f32 bg_red = Settings::values.bg_red.GetValue() / 255.0f;
@@ -325,6 +347,13 @@ void VKBlitScreen::CreateDynamicResources() {
     CreateRenderPass();
     CreateFramebuffers();
     CreateGraphicsPipeline();
+    fsr.reset();
+    if (Settings::values.scaling_filter.GetValue() == Settings::ScalingFilter::Fsr) {
+        const auto& layout = render_window.GetFramebufferLayout();
+        fsr = std::make_unique<FSR>(
+            device, memory_allocator, image_count,
+            VkExtent2D{.width = layout.screen.GetWidth(), .height = layout.screen.GetHeight()});
+    }
 }
 
 void VKBlitScreen::RefreshResources(const Tegra::FramebufferConfig& framebuffer) {
@@ -716,13 +745,14 @@ void VKBlitScreen::CreateGraphicsPipeline() {
 }
 
 void VKBlitScreen::CreateSampler() {
+    bool linear = Settings::values.scaling_filter.GetValue() != Settings::ScalingFilter::Fsr;
     const VkSamplerCreateInfo ci{
         .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
         .pNext = nullptr,
         .flags = 0,
-        .magFilter = VK_FILTER_LINEAR,
-        .minFilter = VK_FILTER_LINEAR,
-        .mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR,
+        .magFilter = linear ? VK_FILTER_LINEAR : VK_FILTER_NEAREST,
+        .minFilter = linear ? VK_FILTER_LINEAR : VK_FILTER_NEAREST,
+        .mipmapMode = VK_SAMPLER_MIPMAP_MODE_NEAREST,
         .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
         .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
         .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_BORDER,
@@ -905,10 +935,11 @@ void VKBlitScreen::SetVertexData(BufferData& data, const Tegra::FramebufferConfi
     UNIMPLEMENTED_IF(framebuffer_crop_rect.top != 0);
     UNIMPLEMENTED_IF(framebuffer_crop_rect.left != 0);
 
-    // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering
-    // (e.g. handheld mode) on a 1920x1080 framebuffer.
     f32 scale_u = 1.0f;
     f32 scale_v = 1.0f;
+    // Scale the output by the crop width/height. This is commonly used with 1280x720 rendering
+    // (e.g. handheld mode) on a 1920x1080 framebuffer.
+    if (!fsr) {
         if (framebuffer_crop_rect.GetWidth() > 0) {
             scale_u = static_cast<f32>(framebuffer_crop_rect.GetWidth()) /
                       static_cast<f32>(screen_info.width);
@@ -917,6 +948,7 @@ void VKBlitScreen::SetVertexData(BufferData& data, const Tegra::FramebufferConfi
             scale_v = static_cast<f32>(framebuffer_crop_rect.GetHeight()) /
                       static_cast<f32>(screen_info.height);
         }
+    }
 
     const auto& screen = layout.screen;
     const auto x = static_cast<f32>(screen.left);

src/video_core/renderer_vulkan/vk_blit_screen.h

@@ -38,6 +38,8 @@ class RasterizerVulkan;
 class VKScheduler;
 class VKSwapchain;
 
+class FSR;
+
 struct VKScreenInfo {
     VkImageView image_view{};
     u32 width{};
@@ -132,6 +134,8 @@ private:
     std::vector<MemoryCommit> raw_buffer_commits;
     u32 raw_width = 0;
     u32 raw_height = 0;
+
+    std::unique_ptr<FSR> fsr;
 };
 
 } // namespace Vulkan

src/video_core/renderer_vulkan/vk_fsr.cpp

@@ -0,0 +1,375 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#include "common/div_ceil.h"
+#include "video_core/host_shaders/vulkan_fidelityfx_fsr_easu_comp_spv.h"
+#include "video_core/host_shaders/vulkan_fidelityfx_fsr_rcas_comp_spv.h"
+#include "video_core/renderer_vulkan/vk_fsr.h"
+#include "video_core/renderer_vulkan/vk_scheduler.h"
+#include "video_core/renderer_vulkan/vk_shader_util.h"
+#include "video_core/vulkan_common/vulkan_device.h"
+
+namespace Vulkan {
+
+FSR::FSR(const Device& device, MemoryAllocator& memory_allocator, size_t image_count,
+         VkExtent2D output_size)
+    : device{device}, memory_allocator{memory_allocator}, image_count{image_count},
+      output_size{output_size} {
+
+    CreateImages();
+    CreateSampler();
+    CreateShaders();
+    CreateDescriptorPool();
+    CreateDescriptorSetLayout();
+    CreateDescriptorSets();
+    CreatePipelineLayout();
+    CreatePipeline();
+}
+
+VkImageView FSR::Draw(VKScheduler& scheduler, size_t image_index, VkImageView image_view,
+                      const Common::Rectangle<int>& crop_rect) {
+
+    UpdateDescriptorSet(image_index, image_view);
+
+    scheduler.Record([this, image_index, crop_rect](vk::CommandBuffer cmdbuf) {
+        const VkImageMemoryBarrier base_barrier{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_MEMORY_BARRIER,
+            .pNext = nullptr,
+            .srcAccessMask = 0,
+            .dstAccessMask = 0,
+            .oldLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .newLayout = VK_IMAGE_LAYOUT_GENERAL,
+            .srcQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .dstQueueFamilyIndex = VK_QUEUE_FAMILY_IGNORED,
+            .image = {},
+            .subresourceRange =
+                {
+                    .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
+                    .baseMipLevel = 0,
+                    .levelCount = 1,
+                    .baseArrayLayer = 0,
+                    .layerCount = 1,
+                },
+        };
+
+        // TODO: Support clear color
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *easu_pipeline);
+        cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT,
+                             VkExtent2D{
+                                 .width = static_cast<u32>(crop_rect.GetWidth()),
+                                 .height = static_cast<u32>(crop_rect.GetHeight()),
+                             });
+
+        {
+            VkImageMemoryBarrier fsr_write_barrier = base_barrier;
+            fsr_write_barrier.image = *images[image_index],
+            fsr_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+
+            cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_ALL_COMMANDS_BIT,
+                                   VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, fsr_write_barrier);
+        }
+
+        cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0,
+                                  descriptor_sets[image_index * 2], {});
+        cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u),
+                        Common::DivCeil(output_size.height, 16u), 1);
+
+        cmdbuf.BindPipeline(VK_PIPELINE_BIND_POINT_COMPUTE, *rcas_pipeline);
+        cmdbuf.PushConstants(*pipeline_layout, VK_SHADER_STAGE_COMPUTE_BIT, output_size);
+
+        {
+            std::array<VkImageMemoryBarrier, 2> barriers;
+            auto& fsr_read_barrier = barriers[0];
+            auto& blit_write_barrier = barriers[1];
+
+            fsr_read_barrier = base_barrier;
+            fsr_read_barrier.image = *images[image_index];
+            fsr_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT;
+            fsr_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT;
+
+            blit_write_barrier = base_barrier;
+            blit_write_barrier.image = *images[image_count + image_index];
+            blit_write_barrier.oldLayout = VK_IMAGE_LAYOUT_UNDEFINED;
+            blit_write_barrier.newLayout = VK_IMAGE_LAYOUT_GENERAL;
+
+            cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                                   VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT, 0, {}, {}, barriers);
+        }
+
+        cmdbuf.BindDescriptorSets(VK_PIPELINE_BIND_POINT_COMPUTE, *pipeline_layout, 0,
+                                  descriptor_sets[image_index * 2 + 1], {});
+        cmdbuf.Dispatch(Common::DivCeil(output_size.width, 16u),
+                        Common::DivCeil(output_size.height, 16u), 1);
+
+        {
+            std::array<VkImageMemoryBarrier, 1> barriers;
+            auto& blit_read_barrier = barriers[0];
+
+            blit_read_barrier = base_barrier;
+            blit_read_barrier.image = *images[image_count + image_index];
+            blit_read_barrier.srcAccessMask = VK_ACCESS_SHADER_WRITE_BIT;
+            blit_read_barrier.dstAccessMask = VK_ACCESS_SHADER_READ_BIT;
+
+            cmdbuf.PipelineBarrier(VK_PIPELINE_STAGE_COMPUTE_SHADER_BIT,
+                                   VK_PIPELINE_STAGE_FRAGMENT_SHADER_BIT, 0, {}, {}, barriers);
+        }
+    });
+
+    return *image_views[image_count + image_index];
+}
+
+void FSR::CreateDescriptorPool() {
+    const std::array<VkDescriptorPoolSize, 2> pool_sizes{{
+        {
+            .type = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+            .descriptorCount = static_cast<u32>(image_count * 2),
+        },
+        {
+            .type = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
+            .descriptorCount = static_cast<u32>(image_count * 2),
+        },
+    }};
+
+    const VkDescriptorPoolCreateInfo ci{
+        .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_POOL_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = VK_DESCRIPTOR_POOL_CREATE_FREE_DESCRIPTOR_SET_BIT,
+        .maxSets = static_cast<u32>(image_count * 2),
+        .poolSizeCount = static_cast<u32>(pool_sizes.size()),
+        .pPoolSizes = pool_sizes.data(),
+    };
+    descriptor_pool = device.GetLogical().CreateDescriptorPool(ci);
+}
+
+void FSR::CreateDescriptorSetLayout() {
+    const std::array<VkDescriptorSetLayoutBinding, 2> layout_bindings{{
+        {
+            .binding = 0,
+            .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+            .descriptorCount = 1,
+            .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+            .pImmutableSamplers = sampler.address(),
+        },
+        {
+            .binding = 1,
+            .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
+            .descriptorCount = 1,
+            .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+            .pImmutableSamplers = sampler.address(),
+        },
+    }};
+
+    const VkDescriptorSetLayoutCreateInfo ci{
+        .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_LAYOUT_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .bindingCount = static_cast<u32>(layout_bindings.size()),
+        .pBindings = layout_bindings.data(),
+    };
+
+    descriptor_set_layout = device.GetLogical().CreateDescriptorSetLayout(ci);
+}
+
+void FSR::CreateDescriptorSets() {
+    const u32 sets = static_cast<u32>(image_count * 2);
+    const std::vector layouts(sets, *descriptor_set_layout);
+
+    const VkDescriptorSetAllocateInfo ai{
+        .sType = VK_STRUCTURE_TYPE_DESCRIPTOR_SET_ALLOCATE_INFO,
+        .pNext = nullptr,
+        .descriptorPool = *descriptor_pool,
+        .descriptorSetCount = sets,
+        .pSetLayouts = layouts.data(),
+    };
+
+    descriptor_sets = descriptor_pool.Allocate(ai);
+}
+
+void FSR::CreateImages() {
+    images.resize(image_count * 2);
+    image_views.resize(image_count * 2);
+    buffer_commits.resize(image_count * 2);
+
+    for (size_t i = 0; i < image_count * 2; ++i) {
+        images[i] = device.GetLogical().CreateImage(VkImageCreateInfo{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_CREATE_INFO,
+            .pNext = nullptr,
+            .flags = 0,
+            .imageType = VK_IMAGE_TYPE_2D,
+            .format = VK_FORMAT_R16G16B16A16_SFLOAT,
+            .extent =
+                {
+                    .width = output_size.width,
+                    .height = output_size.height,
+                    .depth = 1,
+                },
+            .mipLevels = 1,
+            .arrayLayers = 1,
+            .samples = VK_SAMPLE_COUNT_1_BIT,
+            .tiling = VK_IMAGE_TILING_OPTIMAL,
+            .usage = VK_IMAGE_USAGE_TRANSFER_SRC_BIT | VK_IMAGE_USAGE_STORAGE_BIT |
+                     VK_IMAGE_USAGE_SAMPLED_BIT,
+            .sharingMode = VK_SHARING_MODE_EXCLUSIVE,
+            .queueFamilyIndexCount = 0,
+            .pQueueFamilyIndices = nullptr,
+            .initialLayout = VK_IMAGE_LAYOUT_UNDEFINED,
+        });
+        buffer_commits[i] = memory_allocator.Commit(images[i], MemoryUsage::DeviceLocal);
+        image_views[i] = device.GetLogical().CreateImageView(VkImageViewCreateInfo{
+            .sType = VK_STRUCTURE_TYPE_IMAGE_VIEW_CREATE_INFO,
+            .pNext = nullptr,
+            .flags = 0,
+            .image = *images[i],
+            .viewType = VK_IMAGE_VIEW_TYPE_2D,
+            .format = VK_FORMAT_R16G16B16A16_SFLOAT,
+            .components =
+                {
+                    .r = VK_COMPONENT_SWIZZLE_IDENTITY,
+                    .g = VK_COMPONENT_SWIZZLE_IDENTITY,
+                    .b = VK_COMPONENT_SWIZZLE_IDENTITY,
+                    .a = VK_COMPONENT_SWIZZLE_IDENTITY,
+                },
+            .subresourceRange =
+                {
+                    .aspectMask = VK_IMAGE_ASPECT_COLOR_BIT,
+                    .baseMipLevel = 0,
+                    .levelCount = 1,
+                    .baseArrayLayer = 0,
+                    .layerCount = 1,
+                },
+        });
+    }
+}
+
+void FSR::CreatePipelineLayout() {
+    VkPushConstantRange push_const{
+        .stageFlags = VK_SHADER_STAGE_COMPUTE_BIT,
+        .offset = 0,
+        .size = sizeof(std::array<u32, 2>),
+    };
+    VkPipelineLayoutCreateInfo ci{
+        .sType = VK_STRUCTURE_TYPE_PIPELINE_LAYOUT_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .setLayoutCount = 1,
+        .pSetLayouts = descriptor_set_layout.address(),
+        .pushConstantRangeCount = 1,
+        .pPushConstantRanges = &push_const,
+    };
+
+    pipeline_layout = device.GetLogical().CreatePipelineLayout(ci);
+}
+
+void FSR::UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const {
+    const auto fsr_image_view = *image_views[image_index];
+    const auto blit_image_view = *image_views[image_count + image_index];
+
+    const VkDescriptorImageInfo image_info{
+        .imageView = image_view,
+        .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+    };
+    const VkDescriptorImageInfo fsr_image_info{
+        .imageView = fsr_image_view,
+        .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+    };
+    const VkDescriptorImageInfo blit_image_info{
+        .imageView = blit_image_view,
+        .imageLayout = VK_IMAGE_LAYOUT_GENERAL,
+    };
+
+    VkWriteDescriptorSet sampler_write{
+        .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
+        .pNext = nullptr,
+        .dstSet = descriptor_sets[image_index * 2],
+        .dstBinding = 0,
+        .dstArrayElement = 0,
+        .descriptorCount = 1,
+        .descriptorType = VK_DESCRIPTOR_TYPE_COMBINED_IMAGE_SAMPLER,
+        .pImageInfo = &image_info,
+        .pBufferInfo = nullptr,
+        .pTexelBufferView = nullptr,
+    };
+
+    VkWriteDescriptorSet output_write{
+        .sType = VK_STRUCTURE_TYPE_WRITE_DESCRIPTOR_SET,
+        .pNext = nullptr,
+        .dstSet = descriptor_sets[image_index * 2],
+        .dstBinding = 1,
+        .dstArrayElement = 0,
+        .descriptorCount = 1,
+        .descriptorType = VK_DESCRIPTOR_TYPE_STORAGE_IMAGE,
+        .pImageInfo = &fsr_image_info,
+        .pBufferInfo = nullptr,
+        .pTexelBufferView = nullptr,
+    };
+
+    device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {});
+
+    sampler_write.dstSet = descriptor_sets[image_index * 2 + 1];
+    sampler_write.pImageInfo = &fsr_image_info;
+    output_write.dstSet = descriptor_sets[image_index * 2 + 1];
+    output_write.pImageInfo = &blit_image_info;
+
+    device.GetLogical().UpdateDescriptorSets(std::array{sampler_write, output_write}, {});
+}
+
+void FSR::CreateSampler() {
+    const VkSamplerCreateInfo ci{
+        .sType = VK_STRUCTURE_TYPE_SAMPLER_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .magFilter = VK_FILTER_LINEAR,
+        .minFilter = VK_FILTER_LINEAR,
+        .mipmapMode = VK_SAMPLER_MIPMAP_MODE_LINEAR,
+        .addressModeU = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE,
+        .addressModeV = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE,
+        .addressModeW = VK_SAMPLER_ADDRESS_MODE_CLAMP_TO_EDGE,
+        .mipLodBias = 0.0f,
+        .anisotropyEnable = VK_FALSE,
+        .maxAnisotropy = 0.0f,
+        .compareEnable = VK_FALSE,
+        .compareOp = VK_COMPARE_OP_NEVER,
+        .minLod = 0.0f,
+        .maxLod = 0.0f,
+        .borderColor = VK_BORDER_COLOR_FLOAT_OPAQUE_BLACK,
+        .unnormalizedCoordinates = VK_FALSE,
+    };
+
+    sampler = device.GetLogical().CreateSampler(ci);
+}
+
+void FSR::CreateShaders() {
+    easu_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_COMP_SPV);
+    rcas_shader = BuildShader(device, VULKAN_FIDELITYFX_FSR_EASU_COMP_SPV);
+}
+
+void FSR::CreatePipeline() {
+    VkPipelineShaderStageCreateInfo shader_stage{
+
+        .sType = VK_STRUCTURE_TYPE_PIPELINE_SHADER_STAGE_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .stage = VK_SHADER_STAGE_COMPUTE_BIT,
+        .pName = "main",
+        .pSpecializationInfo = nullptr,
+    };
+
+    VkComputePipelineCreateInfo pipeline_ci{
+        .sType = VK_STRUCTURE_TYPE_COMPUTE_PIPELINE_CREATE_INFO,
+        .pNext = nullptr,
+        .flags = 0,
+        .layout = *pipeline_layout,
+        .basePipelineIndex = 0,
+    };
+
+    shader_stage.module = *easu_shader;
+    pipeline_ci.stage = shader_stage;
+    easu_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci);
+
+    shader_stage.module = *rcas_shader;
+    pipeline_ci.stage = shader_stage;
+    rcas_pipeline = device.GetLogical().CreateComputePipeline(pipeline_ci);
+}
+
+} // namespace Vulkan

src/video_core/renderer_vulkan/vk_fsr.h

@@ -0,0 +1,54 @@
+// Copyright 2021 yuzu Emulator Project
+// Licensed under GPLv2 or any later version
+// Refer to the license.txt file included.
+
+#pragma once
+
+#include "common/math_util.h"
+#include "video_core/vulkan_common/vulkan_memory_allocator.h"
+#include "video_core/vulkan_common/vulkan_wrapper.h"
+
+namespace Vulkan {
+
+class Device;
+class VKScheduler;
+
+class FSR {
+public:
+    explicit FSR(const Device& device, MemoryAllocator& memory_allocator, size_t image_count,
+                 VkExtent2D output_size);
+    VkImageView Draw(VKScheduler& scheduler, size_t image_index, VkImageView image_view,
+                     const Common::Rectangle<int>& crop_rect);
+
+private:
+    void CreateDescriptorPool();
+    void CreateDescriptorSetLayout();
+    void CreateDescriptorSets();
+    void CreateImages();
+    void CreateSampler();
+    void CreateShaders();
+    void CreatePipeline();
+    void CreatePipelineLayout();
+
+    void UpdateDescriptorSet(std::size_t image_index, VkImageView image_view) const;
+
+    const Device& device;
+    MemoryAllocator& memory_allocator;
+    size_t image_count;
+    VkExtent2D output_size;
+
+    vk::DescriptorPool descriptor_pool;
+    vk::DescriptorSetLayout descriptor_set_layout;
+    vk::DescriptorSets descriptor_sets;
+    vk::PipelineLayout pipeline_layout;
+    vk::ShaderModule easu_shader;
+    vk::ShaderModule rcas_shader;
+    vk::Pipeline easu_pipeline;
+    vk::Pipeline rcas_pipeline;
+    vk::Sampler sampler;
+    std::vector<vk::Image> images;
+    std::vector<vk::ImageView> image_views;
+    std::vector<MemoryCommit> buffer_commits;
+};
+
+} // namespace Vulkan

src/yuzu/configuration/configure_graphics.ui

@@ -402,6 +402,11 @@
                <string>ScaleForce</string>
               </property>
              </item>
+             <item>
+              <property name="text">
+               <string>FidelityFX Super Resolution</string>
+              </property>
+             </item>
             </widget>
            </item>
           </layout>