diff --git a/Source/Core/Core/Src/HW/GPFifo.cpp b/Source/Core/Core/Src/HW/GPFifo.cpp
index 333f52b789..28ba18da1f 100644
--- a/Source/Core/Core/Src/HW/GPFifo.cpp
+++ b/Source/Core/Core/Src/HW/GPFifo.cpp
@@ -55,7 +55,8 @@ void Init()
 	ResetGatherPipe();
 }
 
-bool IsEmpty() {
+bool IsEmpty()
+{
 	return m_gatherPipeCount == 0;
 }
 
@@ -66,27 +67,34 @@ void ResetGatherPipe()
 
 void STACKALIGN CheckGatherPipe()
 {
-	// HyperIris: Memory::GetPointer is an expensive call, call it less, run faster
-	// ector: Well not really - this loop will only rarely go more than one lap.
-	while (m_gatherPipeCount >= GATHER_PIPE_SIZE)
-	{	
-		// copy the GatherPipe
-		memcpy(Memory::GetPointer(ProcessorInterface::Fifo_CPUWritePointer), m_gatherPipe, GATHER_PIPE_SIZE);
+	if (m_gatherPipeCount >= GATHER_PIPE_SIZE)
+	{
+		u32 cnt;
+		u8* curMem = Memory::GetPointer(ProcessorInterface::Fifo_CPUWritePointer);
+		for (cnt = 0; m_gatherPipeCount >= GATHER_PIPE_SIZE; cnt += GATHER_PIPE_SIZE)
+		{	
+			// copy the GatherPipe
+			memcpy(curMem, m_gatherPipe + cnt, GATHER_PIPE_SIZE);
+			m_gatherPipeCount -= GATHER_PIPE_SIZE;
 
+			// increase the CPUWritePointer
+			if (ProcessorInterface::Fifo_CPUWritePointer + GATHER_PIPE_SIZE >= ProcessorInterface::Fifo_CPUEnd)
+			{
+				curMem -= ProcessorInterface::Fifo_CPUWritePointer - ProcessorInterface::Fifo_CPUBase;
+				ProcessorInterface::Fifo_CPUWritePointer = ProcessorInterface::Fifo_CPUBase;
+			}
+			else
+			{
+				curMem += GATHER_PIPE_SIZE;
+				ProcessorInterface::Fifo_CPUWritePointer += GATHER_PIPE_SIZE;
+			}
+
+			// TODO store video plugin pointer
+			CPluginManager::GetInstance().GetVideo()->Video_GatherPipeBursted();
+		}
+		
 		// move back the spill bytes
-		m_gatherPipeCount -= GATHER_PIPE_SIZE;
-		
-		// HyperIris: dunno why, but I use memcpy. TODO: See if a custom copy can be faster, like 4x MOVAPD
-		memmove(m_gatherPipe, m_gatherPipe + GATHER_PIPE_SIZE, m_gatherPipeCount);
-		
-		// increase the CPUWritePointer
-		if (ProcessorInterface::Fifo_CPUWritePointer == ProcessorInterface::Fifo_CPUEnd)
-			ProcessorInterface::Fifo_CPUWritePointer = ProcessorInterface::Fifo_CPUBase;
-        else
-            ProcessorInterface::Fifo_CPUWritePointer += GATHER_PIPE_SIZE;
-
-        // TODO store video plugin pointer
-		CPluginManager::GetInstance().GetVideo()->Video_GatherPipeBursted();
+		memmove(m_gatherPipe, m_gatherPipe + cnt, m_gatherPipeCount);
 	}
 }
 
diff --git a/Source/Core/VideoCommon/Src/CommandProcessor.cpp b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
index 4e8bcfb849..f8f0f4bb1c 100644
--- a/Source/Core/VideoCommon/Src/CommandProcessor.cpp
+++ b/Source/Core/VideoCommon/Src/CommandProcessor.cpp
@@ -594,7 +594,7 @@ void STACKALIGN GatherPipeBursted()
 	if (g_VideoInitialize.bOnThread)
 	{
 		// update the fifo-pointer
-        if (fifo.CPWritePointer >= fifo.CPEnd)
+        if (fifo.CPWritePointer + GATHER_PIPE_SIZE >= fifo.CPEnd)
 			fifo.CPWritePointer = fifo.CPBase;
         else
 		    fifo.CPWritePointer += GATHER_PIPE_SIZE;
@@ -634,7 +634,7 @@ void STACKALIGN GatherPipeBursted()
 	}
 	else
 	{
-		if (fifo.CPWritePointer >= fifo.CPEnd)
+		if (fifo.CPWritePointer + GATHER_PIPE_SIZE >= fifo.CPEnd)
 			fifo.CPWritePointer = fifo.CPBase;
         else
 		    fifo.CPWritePointer += GATHER_PIPE_SIZE;
@@ -687,11 +687,11 @@ void CatchUpGPU()
 
 			fifo.CPReadWriteDistance -= 32;
 			// increase the ReadPtr
-			if (fifo.CPReadPointer >= fifo.CPEnd)
+			if (fifo.CPReadPointer + 32 >= fifo.CPEnd)
 			{
 				fifo.CPReadPointer = fifo.CPBase;
 				// adjust, take care
-				ptr = Memory_GetPtr(fifo.CPReadPointer);
+				ptr -= fifo.CPReadPointer - fifo.CPBase;
 				DEBUG_LOG(COMMANDPROCESSOR, "Fifo Loop");
 			}
             else
diff --git a/Source/Core/VideoCommon/Src/Fifo.cpp b/Source/Core/VideoCommon/Src/Fifo.cpp
index 099084fae5..a24a96b3ad 100644
--- a/Source/Core/VideoCommon/Src/Fifo.cpp
+++ b/Source/Core/VideoCommon/Src/Fifo.cpp
@@ -118,12 +118,12 @@ void Fifo_SendFifoData(u8* _uData, u32 len)
     if (size + len >= FIFO_SIZE)
     {
 		int pos = (int)(g_pVideoData - videoBuffer);
-        if (size - pos > pos)
+		size -= pos;
+        if (size + len > FIFO_SIZE)
         {
             PanicAlert("FIFO out of bounds (sz = %i, at %08x)", size, pos);
         }
-        memmove(&videoBuffer[0], &videoBuffer[pos], size - pos);
-        size -= pos;
+        memmove(&videoBuffer[0], &videoBuffer[pos], size);
 		g_pVideoData = videoBuffer;
     }
 	// Copy new video instructions to videoBuffer for future use in rendering the new picture
@@ -180,7 +180,7 @@ void Fifo_EnterLoop(const SVideoInitialize &video_initialize)
 				}
 				distToSend = 32;
 
-				if ( readPtr >= _fifo.CPEnd) 
+				if (readPtr + 32 >= _fifo.CPEnd) 
 					readPtr = _fifo.CPBase;
 				else
 					readPtr += 32;
@@ -192,9 +192,9 @@ void Fifo_EnterLoop(const SVideoInitialize &video_initialize)
 				// send 1024B chunk max length to have better control over PeekMessages' period
 				distToSend = distToSend > 1024 ? 1024 : distToSend;
 				// add 32 bytes because the cp end points to the start of the last 32 byte chunk
-				if ((distToSend + readPtr) >= (_fifo.CPEnd + 32)) // TODO: better?
+				if (readPtr + distToSend >= _fifo.CPEnd)
 				{
-					distToSend =(_fifo.CPEnd + 32) - readPtr;
+					distToSend = _fifo.CPEnd - readPtr;
 					readPtr = _fifo.CPBase;
 				}
 				else