From 76ca2a5f82f4df64cb839af42c93acb6705411ae Mon Sep 17 00:00:00 2001
From: ReinUsesLisp <reinuseslisp@airmail.cc>
Date: Sat, 2 Nov 2019 04:08:31 -0300
Subject: gl_rasterizer: Upload constant buffers with glNamedBufferSubData

Nvidia's OpenGL driver maps gl(Named)BufferSubData with some requirements
to a fast. This path has an extra memcpy but updates the buffer without
orphaning or waiting for previous calls. It can be seen as a better
model for "push constants" that can upload a whole UBO instead of 256
bytes.

This path has some requirements established here:
http://on-demand.gputechconf.com/gtc/2014/presentations/S4379-opengl-44-scene-rendering-techniques.pdf#page=24

Instead of using the stream buffer, this commits moves constant buffers
uploads to calls of glNamedBufferSubData and from my testing it brings a
performance improvement. This is disabled when the vendor is not Nvidia
since it brings performance regressions.
---
 src/video_core/renderer_opengl/gl_buffer_cache.h | 20 ++++++++++++++++++--
 1 file changed, 18 insertions(+), 2 deletions(-)

(limited to 'src/video_core/renderer_opengl/gl_buffer_cache.h')

diff --git a/src/video_core/renderer_opengl/gl_buffer_cache.h b/src/video_core/renderer_opengl/gl_buffer_cache.h
index 022e7bfa9..8c7145443 100644
--- a/src/video_core/renderer_opengl/gl_buffer_cache.h
+++ b/src/video_core/renderer_opengl/gl_buffer_cache.h
@@ -4,10 +4,12 @@
 
 #pragma once
 
+#include <array>
 #include <memory>
 
 #include "common/common_types.h"
 #include "video_core/buffer_cache/buffer_cache.h"
+#include "video_core/engines/maxwell_3d.h"
 #include "video_core/rasterizer_cache.h"
 #include "video_core/renderer_opengl/gl_resource_manager.h"
 #include "video_core/renderer_opengl/gl_stream_buffer.h"
@@ -18,12 +20,14 @@ class System;
 
 namespace OpenGL {
 
+class Device;
 class OGLStreamBuffer;
 class RasterizerOpenGL;
 
 class CachedBufferBlock;
 
 using Buffer = std::shared_ptr<CachedBufferBlock>;
+using GenericBufferCache = VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer>;
 
 class CachedBufferBlock : public VideoCommon::BufferBlock {
 public:
@@ -38,14 +42,18 @@ private:
     OGLBuffer gl_buffer{};
 };
 
-class OGLBufferCache final : public VideoCommon::BufferCache<Buffer, GLuint, OGLStreamBuffer> {
+class OGLBufferCache final : public GenericBufferCache {
 public:
     explicit OGLBufferCache(RasterizerOpenGL& rasterizer, Core::System& system,
-                            std::size_t stream_size);
+                            const Device& device, std::size_t stream_size);
     ~OGLBufferCache();
 
     const GLuint* GetEmptyBuffer(std::size_t) override;
 
+    void Acquire() noexcept {
+        cbuf_cursor = 0;
+    }
+
 protected:
     Buffer CreateBlock(CacheAddr cache_addr, std::size_t size) override;
 
@@ -61,6 +69,14 @@ protected:
 
     void CopyBlock(const Buffer& src, const Buffer& dst, std::size_t src_offset,
                    std::size_t dst_offset, std::size_t size) override;
+
+    BufferInfo ConstBufferUpload(const void* raw_pointer, std::size_t size) override;
+
+private:
+    std::size_t cbuf_cursor = 0;
+    std::array<GLuint, Tegra::Engines::Maxwell3D::Regs::MaxConstBuffers *
+                           Tegra::Engines::Maxwell3D::Regs::MaxShaderProgram>
+        cbufs;
 };
 
 } // namespace OpenGL
-- 
cgit v1.2.3