Implement shader storage buffer operations using new Load/Store instructions (#4993)

* Implement storage buffer operations using new Load/Store instruction

* Extend GenerateMultiTargetStorageOp to also match access with constant offset, and log and comments

* Remove now unused code

* Catch more complex cases of global memory usage

* Shader cache version bump

* Extend global access elimination to work with more shared memory cases

* Change alignment requirement from 16 bytes to 8 bytes, handle cases where we need more than 16 storage buffers

* Tweak preferencing to catch more cases

* Enable CB0 elimination even when host storage buffer alignment is > 16 (for Intel)

* Fix storage buffer bindings

* Simplify some code

* Shader cache version bump

* Fix typo

* Extend global memory elimination to handle shared memory with multiple possible offsets and local memory

src/Ryujinx.Graphics.Gpu/Constants.cs

@@ -80,11 +80,6 @@ namespace Ryujinx.Graphics.Gpu
         /// </summary>
         public const int GobAlignment = 64;
 
-        /// <summary>
-        /// Expected byte alignment for storage buffers
-        /// </summary>
-        public const int StorageAlignment = 16;
-
         /// <summary>
         /// Number of the uniform buffer reserved by the driver to store the storage buffer base addresses.
         /// </summary>

src/Ryujinx.Graphics.Gpu/Engine/Compute/ComputeClass.cs

@@ -187,30 +187,6 @@ namespace Ryujinx.Graphics.Gpu.Engine.Compute
                 info = cs.Shaders[0].Info;
             }
 
-            for (int index = 0; index < info.CBuffers.Count; index++)
-            {
-                BufferDescriptor cb = info.CBuffers[index];
-
-                // NVN uses the "hardware" constant buffer for anything that is less than 8,
-                // and those are already bound above.
-                // Anything greater than or equal to 8 uses the emulated constant buffers.
-                // They are emulated using global memory loads.
-                if (cb.Slot < 8)
-                {
-                    continue;
-                }
-
-                ulong cbDescAddress = _channel.BufferManager.GetComputeUniformBufferAddress(0);
-
-                int cbDescOffset = 0x260 + (cb.Slot - 8) * 0x10;
-
-                cbDescAddress += (ulong)cbDescOffset;
-
-                SbDescriptor cbDescriptor = _channel.MemoryManager.Physical.Read<SbDescriptor>(cbDescAddress);
-
-                _channel.BufferManager.SetComputeUniformBuffer(cb.Slot, cbDescriptor.PackAddress(), (uint)cbDescriptor.Size);
-            }
-
             _channel.BufferManager.SetComputeBufferBindings(cs.Bindings);
 
             _channel.TextureManager.SetComputeBindings(cs.Bindings);

src/Ryujinx.Graphics.Gpu/Memory/BufferManager.cs

@@ -222,7 +222,7 @@ namespace Ryujinx.Graphics.Gpu.Memory
         /// <param name="gpuVa">Start GPU virtual address of the buffer</param>
         private void RecordStorageAlignment(BuffersPerStage buffers, int index, ulong gpuVa)
         {
-            bool unaligned = (gpuVa & (Constants.StorageAlignment - 1)) != 0;
+            bool unaligned = (gpuVa & ((ulong)_context.Capabilities.StorageBufferOffsetAlignment - 1)) != 0;
 
             if (unaligned || HasUnalignedStorageBuffers)
             {

src/Ryujinx.Graphics.Gpu/Shader/DiskCache/DiskCacheHostStorage.cs

@@ -22,7 +22,7 @@ namespace Ryujinx.Graphics.Gpu.Shader.DiskCache
         private const ushort FileFormatVersionMajor = 1;
         private const ushort FileFormatVersionMinor = 2;
         private const uint FileFormatVersionPacked = ((uint)FileFormatVersionMajor << 16) | FileFormatVersionMinor;
-        private const uint CodeGenVersion = 5027;
+        private const uint CodeGenVersion = 4992;
 
         private const string SharedTocFileName = "shared.toc";
         private const string SharedDataFileName = "shared.data";

src/Ryujinx.Graphics.Gpu/Shader/ShaderInfoBuilder.cs

@@ -92,7 +92,7 @@ namespace Ryujinx.Graphics.Gpu.Shader
             int imageBinding = stageIndex * imagesPerStage * 2;
 
             AddDescriptor(stages, ResourceType.UniformBuffer, UniformSetIndex, uniformBinding, uniformsPerStage);
-            AddArrayDescriptor(stages, ResourceType.StorageBuffer, StorageSetIndex, storageBinding, storagesPerStage);
+            AddDescriptor(stages, ResourceType.StorageBuffer, StorageSetIndex, storageBinding, storagesPerStage);
             AddDualDescriptor(stages, ResourceType.TextureAndSampler, ResourceType.BufferTexture, TextureSetIndex, textureBinding, texturesPerStage);
             AddDualDescriptor(stages, ResourceType.Image, ResourceType.BufferImage, ImageSetIndex, imageBinding, imagesPerStage);
 
@@ -133,19 +133,6 @@ namespace Ryujinx.Graphics.Gpu.Shader
             AddDescriptor(stages, type2, setIndex, binding + count, count);
         }
 
-        /// <summary>
-        /// Adds an array resource to the list of descriptors.
-        /// </summary>
-        /// <param name="stages">Shader stages where the resource is used</param>
-        /// <param name="type">Type of the resource</param>
-        /// <param name="setIndex">Descriptor set number where the resource will be bound</param>
-        /// <param name="binding">Binding number where the resource will be bound</param>
-        /// <param name="count">Number of resources bound at the binding location</param>
-        private void AddArrayDescriptor(ResourceStages stages, ResourceType type, int setIndex, int binding, int count)
-        {
-            _resourceDescriptors[setIndex].Add(new ResourceDescriptor(binding, count, type, stages));
-        }
-
         /// <summary>
         /// Adds buffer usage information to the list of usages.
         /// </summary>