Optimize RendererAllocation by reducing fragmentation with operating on power-of-2 blocks

Source/Engine/Renderer/RenderList.cpp

@@ -220,6 +220,7 @@ void RenderFogData::Init(const RenderView& view, IFogRenderer* renderer)
 void* RendererAllocation::Allocate(uintptr size)
 {
     PROFILE_CPU();
+    size = AllocationUtils::AlignToPowerOf2((int32)size); // Reduce fragmentation by operating on power-of-2 blocks
     void* result = nullptr;
     MemPoolLocker.Lock();
     for (int32 i = 0; i < MemPool.Count(); i++)
@@ -240,6 +241,7 @@ void* RendererAllocation::Allocate(uintptr size)
 void RendererAllocation::Free(void* ptr, uintptr size)
 {
     PROFILE_CPU();
+    size = AllocationUtils::AlignToPowerOf2((int32)size); // Reduce fragmentation by operating on power-of-2 blocks
     MemPoolLocker.Lock();
     MemPool.Add({ ptr, size });
     MemPoolLocker.Unlock();

Source/Engine/Renderer/RendererAllocation.h

@@ -4,6 +4,9 @@
 
 #include "Engine/Core/Memory/SimpleHeapAllocation.h"
 
+/// <summary>
+/// Shared pool allocator for the renderer used for large chunks of memory.
+/// </summary>
 class RendererAllocation : public SimpleHeapAllocation<RendererAllocation, 64>
 {
 public: