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

4 files changed, 9 insertions, 17 deletions

diff --git a/src/Ryujinx.Graphics.Shader/StructuredIr/HelperFunctionsMask.cs b/src/Ryujinx.Graphics.Shader/StructuredIr/HelperFunctionsMask.cs
index d45f8d4e..c348b5d9 100644
--- a/src/Ryujinx.Graphics.Shader/StructuredIr/HelperFunctionsMask.cs
+++ b/src/Ryujinx.Graphics.Shader/StructuredIr/HelperFunctionsMask.cs
@@ -6,7 +6,6 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
     enum HelperFunctionsMask
     {
         AtomicMinMaxS32Shared  = 1 << 0,
-        AtomicMinMaxS32Storage = 1 << 1,
         MultiplyHighS32        = 1 << 2,
         MultiplyHighU32        = 1 << 3,
         Shuffle                = 1 << 4,
@@ -14,7 +13,6 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
         ShuffleUp              = 1 << 6,
         ShuffleXor             = 1 << 7,
         StoreSharedSmallInt    = 1 << 8,
-        StoreStorageSmallInt   = 1 << 9,
         SwizzleAdd             = 1 << 10,
         FSI                    = 1 << 11
     }
diff --git a/src/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs b/src/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs
index 44f0fad9..6e201350 100644
--- a/src/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs
+++ b/src/Ryujinx.Graphics.Shader/StructuredIr/InstructionInfo.cs
@@ -90,10 +90,8 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
             Add(Instruction.ImageAtomic,              AggregateType.S32);
             Add(Instruction.IsNan,                    AggregateType.Bool,   AggregateType.Scalar);
             Add(Instruction.Load,                     AggregateType.FP32);
-            Add(Instruction.LoadGlobal,               AggregateType.U32,    AggregateType.S32,     AggregateType.S32);
             Add(Instruction.LoadLocal,                AggregateType.U32,    AggregateType.S32);
             Add(Instruction.LoadShared,               AggregateType.U32,    AggregateType.S32);
-            Add(Instruction.LoadStorage,              AggregateType.U32,    AggregateType.S32,     AggregateType.S32);
             Add(Instruction.Lod,                      AggregateType.FP32);
             Add(Instruction.LogarithmB2,              AggregateType.Scalar, AggregateType.Scalar);
             Add(Instruction.LogicalAnd,               AggregateType.Bool,   AggregateType.Bool,    AggregateType.Bool);
@@ -123,14 +121,10 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
             Add(Instruction.Sine,                     AggregateType.Scalar, AggregateType.Scalar);
             Add(Instruction.SquareRoot,               AggregateType.Scalar, AggregateType.Scalar);
             Add(Instruction.Store,                    AggregateType.Void);
-            Add(Instruction.StoreGlobal,              AggregateType.Void,   AggregateType.S32,     AggregateType.S32,     AggregateType.U32);
             Add(Instruction.StoreLocal,               AggregateType.Void,   AggregateType.S32,     AggregateType.U32);
             Add(Instruction.StoreShared,              AggregateType.Void,   AggregateType.S32,     AggregateType.U32);
             Add(Instruction.StoreShared16,            AggregateType.Void,   AggregateType.S32,     AggregateType.U32);
             Add(Instruction.StoreShared8,             AggregateType.Void,   AggregateType.S32,     AggregateType.U32);
-            Add(Instruction.StoreStorage,             AggregateType.Void,   AggregateType.S32,     AggregateType.S32,     AggregateType.U32);
-            Add(Instruction.StoreStorage16,           AggregateType.Void,   AggregateType.S32,     AggregateType.S32,     AggregateType.U32);
-            Add(Instruction.StoreStorage8,            AggregateType.Void,   AggregateType.S32,     AggregateType.S32,     AggregateType.U32);
             Add(Instruction.Subtract,                 AggregateType.Scalar, AggregateType.Scalar,  AggregateType.Scalar);
             Add(Instruction.SwizzleAdd,               AggregateType.FP32,   AggregateType.FP32,    AggregateType.FP32,    AggregateType.S32);
             Add(Instruction.TextureSample,            AggregateType.FP32);
@@ -166,7 +160,7 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
             {
                 return AggregateType.FP32;
             }
-            else if (inst == Instruction.Call || inst == Instruction.Load || inst == Instruction.Store)
+            else if (inst == Instruction.Call || inst == Instruction.Load || inst == Instruction.Store || inst.IsAtomic())
             {
                 return AggregateType.S32;
             }
diff --git a/src/Ryujinx.Graphics.Shader/StructuredIr/ShaderProperties.cs b/src/Ryujinx.Graphics.Shader/StructuredIr/ShaderProperties.cs
index 061c89ed..157c5937 100644
--- a/src/Ryujinx.Graphics.Shader/StructuredIr/ShaderProperties.cs
+++ b/src/Ryujinx.Graphics.Shader/StructuredIr/ShaderProperties.cs
@@ -5,17 +5,25 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
     class ShaderProperties
     {
         private readonly Dictionary<int, BufferDefinition> _constantBuffers;
+        private readonly Dictionary<int, BufferDefinition> _storageBuffers;
 
         public IReadOnlyDictionary<int, BufferDefinition> ConstantBuffers => _constantBuffers;
+        public IReadOnlyDictionary<int, BufferDefinition> StorageBuffers => _storageBuffers;
 
         public ShaderProperties()
         {
             _constantBuffers = new Dictionary<int, BufferDefinition>();
+            _storageBuffers = new Dictionary<int, BufferDefinition>();
         }
 
         public void AddConstantBuffer(int binding, BufferDefinition definition)
         {
             _constantBuffers[binding] = definition;
         }
+
+        public void AddStorageBuffer(int binding, BufferDefinition definition)
+        {
+            _storageBuffers[binding] = definition;
+        }
     }
 }
\ No newline at end of file
diff --git a/src/Ryujinx.Graphics.Shader/StructuredIr/StructuredProgram.cs b/src/Ryujinx.Graphics.Shader/StructuredIr/StructuredProgram.cs
index 4405c07a..a8f13276 100644
--- a/src/Ryujinx.Graphics.Shader/StructuredIr/StructuredProgram.cs
+++ b/src/Ryujinx.Graphics.Shader/StructuredIr/StructuredProgram.cs
@@ -280,10 +280,6 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
                     {
                         context.Info.HelperFunctionsMask |= HelperFunctionsMask.AtomicMinMaxS32Shared;
                     }
-                    else if (operation.StorageKind == StorageKind.StorageBuffer)
-                    {
-                        context.Info.HelperFunctionsMask |= HelperFunctionsMask.AtomicMinMaxS32Storage;
-                    }
                     break;
                 case Instruction.MultiplyHighS32:
                     context.Info.HelperFunctionsMask |= HelperFunctionsMask.MultiplyHighS32;
@@ -307,10 +303,6 @@ namespace Ryujinx.Graphics.Shader.StructuredIr
                 case Instruction.StoreShared8:
                     context.Info.HelperFunctionsMask |= HelperFunctionsMask.StoreSharedSmallInt;
                     break;
-                case Instruction.StoreStorage16:
-                case Instruction.StoreStorage8:
-                    context.Info.HelperFunctionsMask |= HelperFunctionsMask.StoreStorageSmallInt;
-                    break;
                 case Instruction.SwizzleAdd:
                     context.Info.HelperFunctionsMask |= HelperFunctionsMask.SwizzleAdd;
                     break;