diff --git a/pcsx2/Gif.cpp b/pcsx2/Gif.cpp
index dd6056effa..59ef287d07 100644
--- a/pcsx2/Gif.cpp
+++ b/pcsx2/Gif.cpp
@@ -28,6 +28,7 @@
 
 using std::min;
 
+#define gifsplit 128
 enum gifstate_t
 {
 	GIF_STATE_EMPTY = 0,
@@ -71,7 +72,7 @@ __forceinline void gsInterrupt() {
 		/*if (!(vif1Regs->mskpath3 && (vif1ch->chcr & 0x100)) || (psHu32(GIF_MODE) & 0x1))
 			CPU_INT( 2, 64 );*/
 #endif
-		if (gspath3done == 0) return;
+		if(gspath3done == 0 || gif->qwc > 0) return;
 	}
 
 	gspath3done = 0;
@@ -83,6 +84,7 @@ __forceinline void gsInterrupt() {
 	psHu32(GIF_STAT)&= ~0xE00; // OPH=0 | APATH=0
 	psHu32(GIF_STAT)&= ~0x1F000000; // QFC=0
 	hwDmacIrq(DMAC_GIF);
+	GIF_LOG("GIF DMA end");
 
 }
 
@@ -125,7 +127,7 @@ static void WRITERING_DMA(u32 *pMem, u32 qwc)
 
 int  _GIFchain() {
 #ifdef GSPATH3FIX
-	u32 qwc = ((psHu32(GIF_MODE) & 0x4) && (vif1Regs->mskpath3)) ? min(8, (int)gif->qwc) : gif->qwc;
+	u32 qwc = ((psHu32(GIF_MODE) & 0x4) && (vif1Regs->mskpath3)) ? min(8, (int)gif->qwc) : min( gifsplit, (int)gif->qwc );
 #else
 	u32 qwc = gif->qwc;
 #endif
@@ -161,7 +163,7 @@ static __forceinline void dmaGIFend()
 	if ((psHu32(GIF_MODE) & 0x4) && gif->qwc != 0)
 		CPU_INT(2, min( 8, (int)gif->qwc ) /** BIAS*/);
 	else
-		CPU_INT(2, gif->qwc /** BIAS*/);
+		CPU_INT(2, min( gifsplit, (int)gif->qwc )  /** BIAS*/);
 }
 
 // These could probably be consolidated into one function,
@@ -172,7 +174,7 @@ static __forceinline void GIFdmaEnd()
 	if (psHu32(GIF_MODE) & 0x4)
 		CPU_INT(2, min( 8, (int)gif->qwc ) /** BIAS*/);
 	else
-		CPU_INT(2, gif->qwc /** BIAS*/);
+		CPU_INT(2, min( gifsplit, (int)gif->qwc ) /** BIAS*/);
 }
 
 void GIFdma() 
@@ -187,7 +189,7 @@ void GIFdma()
 		return;
 	}
 
-	GIF_LOG("dmaGIFstart chcr = %lx, madr = %lx, qwc  = %lx\n tadr = %lx, asr0 = %lx, asr1 = %lx", gif->chcr, gif->madr, gif->qwc, gif->tadr, gif->asr0, gif->asr1);
+	
 
 #ifndef GSPATH3FIX
 	if ( !(psHu32(GIF_MODE) & 0x4) ) {
@@ -266,14 +268,17 @@ void GIFdma()
 		
 		if (((gif->qwc == 0) && (gif->chcr & 0xc) == 0)) 
 			gspath3done = 1;
-		else
+		else if(gif->qwc > 0)
+		{
 			GIFdmaEnd();
 		return;
 		
 	}
-	else {
+	}
+	if ((gif->chcr & 0xc) == 0x4 && gspath3done == 0)
+	{
 		// Chain Mode
-		while ((gspath3done == 0) && (gif->qwc == 0)) {		//Loop if the transfers aren't intermittent
+		//while ((gspath3done == 0) && (gif->qwc == 0)) {		//Loop if the transfers aren't intermittent
 			ptag = (u32*)dmaGetAddr(gif->tadr);  //Set memory pointer to TADR
 			if (ptag == NULL) {					 //Is ptag empty?
 				psHu32(DMAC_STAT)|= DMAC_STAT_BEIS;		 //If yes, set BEIS (BUSERR) in DMAC_STAT register
@@ -311,16 +316,16 @@ void GIFdma()
 				GIF_LOG("dmaIrq Set");
 				gspath3done = 1;
 			}
-		}
+		//}
 	}
 
 	prevcycles = 0;
 	if (!(vif1Regs->mskpath3 || (psHu32(GIF_MODE) & 0x1)))	{
-		if (gspath3done == 0)
+		if (gspath3done == 0 || gif->qwc > 0)
 		{
-			if ((psHu32(GIF_MODE) & 0x4) && gif->qwc != 0)
+			if (gif->qwc != 0)
 			{
-				CPU_INT(2, min( 8, (int)gif->qwc )/** BIAS*/);
+				GIFdmaEnd();
 			} 
 			else
 			{
@@ -340,7 +345,7 @@ void GIFdma()
 void dmaGIF() {
 	 //We used to addd wait time for the buffer to fill here, fixing some timing problems in path 3 masking
 	//It takes the time of 24 QW for the BUS to become ready - The Punisher, And1 Streetball
-	
+	GIF_LOG("dmaGIFstart chcr = %lx, madr = %lx, qwc  = %lx\n tadr = %lx, asr0 = %lx, asr1 = %lx", gif->chcr, gif->madr, gif->qwc, gif->tadr, gif->asr0, gif->asr1);
 	if ((psHu32(DMAC_CTRL) & 0xC) == 0xC ) { // GIF MFIFO
 		Console::WriteLn("GIF MFIFO");
 		gifMFIFOInterrupt();
@@ -359,7 +364,8 @@ void dmaGIF() {
 		gif->qwc  = (u16)ptag[0];			    //QWC set to lower 16bits of the tag
 		gif->chcr = ( gif->chcr & 0xFFFF ) | ( (*ptag) & 0xFFFF0000 );  //Transfer upper part of tag to CHCR bits 31-15
 		
-		dmaGIFend();
+		//gspath3done = hwDmacSrcChainWithStack(gif, (ptag[0] >> 28) & 0x7);
+		GIFdmaEnd();
 		gif->qwc = 0;
 		return;
 	}
diff --git a/pcsx2/IPU/IPU.cpp b/pcsx2/IPU/IPU.cpp
index a1b7626df5..04152bfd10 100644
--- a/pcsx2/IPU/IPU.cpp
+++ b/pcsx2/IPU/IPU.cpp
@@ -1370,6 +1370,9 @@ int FIFOto_write(u32* pMem, int size)
 	}	\
 }
 
+#define gif ((DMACh*)&PS2MEM_HW[0xA000])
+extern void gsInterrupt();
+
 int IPU1dma()
 {
 	u32 *ptag, *pMem;
@@ -1383,6 +1386,13 @@ int IPU1dma()
 
 	assert(!(g_nDMATransfer & IPU_DMA_TIE1));
 
+	//We need to make sure GIF has flushed before sending IPU data, it seems to REALLY screw FFX videos
+	while(gif->chcr & 0x100) 
+	{
+		GIF_LOG("Flushing gif chcr %x tadr %x madr %x qwc %x", gif->chcr, gif->tadr, gif->madr, gif->qwc);
+		gsInterrupt();
+	}
+
 	// in kh, qwc == 0 when dma_actv1 is set
 	if ((g_nDMATransfer & IPU_DMA_ACTV1) && ipu1dma->qwc > 0)
 	{
diff --git a/pcsx2/VUops.cpp b/pcsx2/VUops.cpp
index 299803d2d6..bd8b81b0b7 100644
--- a/pcsx2/VUops.cpp
+++ b/pcsx2/VUops.cpp
@@ -1585,7 +1585,7 @@ void _vuLQ(VURegs * VU) {
 	if (_Ft_ == 0) return; 
  
 	imm = (VU->code & 0x400) ? (VU->code & 0x3ff) | 0xfc00 : (VU->code & 0x3ff); 
-	addr = (imm + VU->VI[_Fs_].SS[0]) * 16;
+	addr = ((imm + VU->VI[_Fs_].SS[0]) * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 
  	ptr = (u32*)GET_VU_MEM(VU, addr);
 	if (_X) VU->VF[_Ft_].UL[0] = ptr[0]; 
@@ -1601,7 +1601,7 @@ void _vuLQD( VURegs * VU ) {
 	if (_Fs_ != 0) VU->VI[_Fs_].US[0]--; 
 	if (_Ft_ == 0) return;
 
-	addr = VU->VI[_Fs_].US[0] * 16;
+	addr = (VU->VI[_Fs_].US[0] * 16) & (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u32*)GET_VU_MEM(VU, addr);
 	if (_X) VU->VF[_Ft_].UL[0] = ptr[0]; 
 	if (_Y) VU->VF[_Ft_].UL[1] = ptr[1]; 
@@ -1614,7 +1614,7 @@ void _vuLQI(VURegs * VU) {
 		u32 addr;
 		u32 *ptr;
 
-		addr = VU->VI[_Fs_].US[0] * 16;
+		addr = (VU->VI[_Fs_].US[0] * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 		ptr = (u32*)GET_VU_MEM(VU, addr);
 		if (_X) VU->VF[_Ft_].UL[0] = ptr[0]; 
 		if (_Y) VU->VF[_Ft_].UL[1] = ptr[1]; 
@@ -1631,7 +1631,7 @@ void _vuSQ(VURegs * VU) {
 	u32 *ptr; 
  
 	imm = (VU->code & 0x400) ? (VU->code & 0x3ff) | 0xfc00 : (VU->code & 0x3ff); 
-	addr = (imm + VU->VI[_Ft_].SS[0]) * 16;
+	addr = ((imm + VU->VI[_Ft_].SS[0]) * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u32*)GET_VU_MEM(VU, addr);
 	if (_X) ptr[0] = VU->VF[_Fs_].UL[0]; 
 	if (_Y) ptr[1] = VU->VF[_Fs_].UL[1]; 
@@ -1644,7 +1644,7 @@ void _vuSQD(VURegs * VU) {
 	u32 *ptr; 
  
 	if(_Ft_ != 0) VU->VI[_Ft_].US[0]--; 
-	addr = VU->VI[_Ft_].US[0] * 16;
+	addr = (VU->VI[_Ft_].US[0] * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u32*)GET_VU_MEM(VU, addr);
 	if (_X) ptr[0] = VU->VF[_Fs_].UL[0]; 
 	if (_Y) ptr[1] = VU->VF[_Fs_].UL[1]; 
@@ -1656,7 +1656,7 @@ void _vuSQI(VURegs * VU) {
 	u32 addr;
 	u32 *ptr; 
  
-	addr = VU->VI[_Ft_].US[0] * 16;
+	addr = (VU->VI[_Ft_].US[0] * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u32*)GET_VU_MEM(VU, addr);
 	if (_X) ptr[0] = VU->VF[_Fs_].UL[0];
 	if (_Y) ptr[1] = VU->VF[_Fs_].UL[1];
@@ -1673,7 +1673,7 @@ void _vuILW(VURegs * VU) {
 	if (_Ft_ == 0) return; 
  
  	imm = (VU->code & 0x400) ? (VU->code & 0x3ff) | 0xfc00 : (VU->code & 0x3ff); 
-	addr = (imm + VU->VI[_Fs_].SS[0]) * 16;
+	addr = ((imm + VU->VI[_Fs_].SS[0]) * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u16*)GET_VU_MEM(VU, addr);
 	if (_X) VU->VI[_Ft_].US[0] = ptr[0]; 
 	if (_Y) VU->VI[_Ft_].US[0] = ptr[2]; 
@@ -1687,7 +1687,7 @@ void _vuISW(VURegs * VU) {
 	u16 *ptr; 
  
  	imm = (VU->code & 0x400) ? (VU->code & 0x3ff) | 0xfc00 : (VU->code & 0x3ff); 
-	addr = (imm + VU->VI[_Fs_].SS[0]) * 16;
+	addr = ((imm + VU->VI[_Fs_].SS[0]) * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u16*)GET_VU_MEM(VU, addr);
 	if (_X) { ptr[0] = VU->VI[_Ft_].US[0]; ptr[1] = 0; }
 	if (_Y) { ptr[2] = VU->VI[_Ft_].US[0]; ptr[3] = 0; }
@@ -1700,7 +1700,7 @@ void _vuILWR(VURegs * VU) {
 	u16 *ptr; 
 	if (_Ft_ == 0) return; 
  
-	addr = VU->VI[_Fs_].US[0] * 16;
+	addr = (VU->VI[_Fs_].US[0] * 16)& (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u16*)GET_VU_MEM(VU, addr);
 	if (_X) VU->VI[_Ft_].US[0] = ptr[0]; 
 	if (_Y) VU->VI[_Ft_].US[0] = ptr[2]; 
@@ -1712,7 +1712,7 @@ void _vuISWR(VURegs * VU) {
 	u32 addr;
 	u16 *ptr; 
  
-	addr = VU->VI[_Fs_].US[0] * 16;
+	addr = (VU->VI[_Fs_].US[0] * 16) & (VU == &VU1 ? 0x3fff : 0xfff);
 	ptr = (u16*)GET_VU_MEM(VU, addr);
 	if (_X) { ptr[0] = VU->VI[_Ft_].US[0]; ptr[1] = 0; }
 	if (_Y) { ptr[2] = VU->VI[_Ft_].US[0]; ptr[3] = 0; }
@@ -2045,9 +2045,20 @@ void _vuXITOP(VURegs * VU) {
 
 void _vuXGKICK(VURegs * VU)
 {
+	u32* ptr = (u32*)GET_VU_MEM(VU, (VU->VI[_Fs_].US[0]*16) & (VU == &VU1 ? 0x3fff : 0xfff));
+	int temp = 0x4000 - ((VU->VI[_Fs_].US[0]*16) & 0x3fff);
+	u32 tempmem[0x8000];
+	
 	// flush all pipelines first (in the right order)
 	_vuFlushAll(VU);
-	GSGIFTRANSFER1((u32*)VU->Mem, (VU->VI[_Fs_].US[0]*16) & 0x3fff);
+
+	//Gonna be slow but reshuffles the memory so overflows wont occur
+		memset(tempmem, 0, sizeof(tempmem));
+		memcpy(tempmem, ptr, temp);
+		ptr = (u32*)GET_VU_MEM(VU, 0);
+		memcpy(&tempmem[temp], ptr, ((VU->VI[_Fs_].US[0]*16) & 0x3fff));
+		GSGIFTRANSFER1((u32*)&tempmem[0], 0);
+	//} else GSGIFTRANSFER1((u32*)VU->Mem, (VU->VI[_Fs_].US[0]*16) & 0x3fff);
 }
 
 void _vuXTOP(VURegs * VU) {
diff --git a/pcsx2/Vif.h b/pcsx2/Vif.h
index 64335c4dde..2f84369b5a 100644
--- a/pcsx2/Vif.h
+++ b/pcsx2/Vif.h
@@ -92,16 +92,16 @@ static __forceinline u32 setVifRowRegs(u32 reg, u32 data)
 	switch (reg)
 	{
 		case 0:
-			vifRegs->r0 = data;
+			vifRegs->r0 += data;
 			break;
 		case 1:
-			vifRegs->r1 = data;
+			vifRegs->r1 += data;
 			break;
 		case 2:
-			vifRegs->r2 = data;
+			vifRegs->r2 += data;
 			break;
 		case 3:
-			vifRegs->r3 = data;
+			vifRegs->r3 += data;
 			break;
 			jNO_DEFAULT;
 	}
diff --git a/pcsx2/VifDma.cpp b/pcsx2/VifDma.cpp
index 6e594881fe..5b57c430f7 100644
--- a/pcsx2/VifDma.cpp
+++ b/pcsx2/VifDma.cpp
@@ -80,10 +80,10 @@ struct VIFUnpackFuncTable
 	UNPACKFUNCTYPE       funcU;
 	UNPACKFUNCTYPE       funcS;
 
-	int bsize; // currently unused
-	int dsize; // byte size of one channel
-	int gsize; // size of data in bytes used for each write cycle
-	int qsize; // used for unpack parts, num of vectors that
+	u32 bsize; // currently unused
+	u32 dsize; // byte size of one channel
+	u32 gsize; // size of data in bytes used for each write cycle
+	u32 qsize; // used for unpack parts, num of vectors that
 	// will be decompressed from data for 1 cycle
 };
 
@@ -333,7 +333,7 @@ static void ProcessMemSkip(int size, unsigned int unpackType, const unsigned int
 	
 }
 
-static int VIFalign(u32 *data, vifCode *v, int size, const unsigned int VIFdmanum)
+static int VIFalign(u32 *data, vifCode *v, unsigned int size, const unsigned int VIFdmanum)
 {
 	u32 *dest;
 	u32 unpackType;
@@ -485,7 +485,7 @@ static int VIFalign(u32 *data, vifCode *v, int size, const unsigned int VIFdmanu
 }
 
 
-static void VIFunpack(u32 *data, vifCode *v, int size, const unsigned int VIFdmanum)
+static void VIFunpack(u32 *data, vifCode *v, unsigned int size, const unsigned int VIFdmanum)
 {
 	u32 *dest;
 	u32 unpackType;
@@ -493,6 +493,7 @@ static void VIFunpack(u32 *data, vifCode *v, int size, const unsigned int VIFdma
 	const VIFUnpackFuncTable *ft;
 	VURegs * VU;
 	u8 *cdata = (u8*)data;
+	u32 tempsize = 0;
 	
 #ifdef _DEBUG
 	u32 memsize = VIFdmanum ? 0x4000 : 0x1000;
@@ -554,6 +555,18 @@ static void VIFunpack(u32 *data, vifCode *v, int size, const unsigned int VIFdma
 	}
 #endif
 
+	tempsize = (vif->tag.addr + (size / (ft->gsize * vifRegs->cycle.wl)) * 
+		((vifRegs->cycle.cl - vifRegs->cycle.wl) * 16)) + ((size / ft->gsize) * 16);
+
+	//Sanity Check (memory overflow)
+	if(tempsize > (u32)(VIFdmanum ? 0x4000 : 0x1000)) 
+	{
+		
+	//	DevCon::Notice("VIF%x Unpack ending %x > %x", params VIFdmanum, tempsize, VIFdmanum ? 0x4000 : 0x1000);
+		tempsize = size;
+		size = 0;
+	} else tempsize = 0;
+
 	if (vifRegs->cycle.cl >= vifRegs->cycle.wl)   // skipping write
 	{
 
@@ -658,7 +671,48 @@ static void VIFunpack(u32 *data, vifCode *v, int size, const unsigned int VIFdma
 			}
 
 		} 
-		else if (size >= ft->dsize && vifRegs->num > 0) //Else write what we do have
+		else if(tempsize)
+		{
+			int incdest = ((vifRegs->cycle.cl - vifRegs->cycle.wl) << 2) + 4;
+			size = 0;
+			
+			
+			while ((tempsize >= ft->gsize) && (vifRegs->num > 0))
+			{
+				//VIFUNPACK_LOG("sorting tempsize :p, size %d, vifnum %d, addr %x", tempsize, vifRegs->num, vif->tag.addr);
+				func(dest, (u32*)cdata, ft->qsize);
+				cdata += ft->gsize;
+				tempsize -= ft->gsize;
+
+				vifRegs->num--;
+				++vif->cl;
+				if (vif->cl == vifRegs->cycle.wl)
+				{
+					dest += incdest;
+					v->addr = (v->addr + (incdest * 4)) & (VIFdmanum ? 0x3fff : 0xfff);
+					if(v->addr <= (u32)(VIFdmanum ? 0x3000 : 0x500)) dest = (u32*)(VU->Mem + v->addr);
+					vif->cl = 0;
+				}
+				else
+				{
+					dest += 4;
+					v->addr = (v->addr + 16) & (VIFdmanum ? 0x3fff : 0xfff);
+					if(v->addr <= (u32)(VIFdmanum ? 0x3000 : 0x500)) dest = (u32*)(VU->Mem + v->addr);
+				}
+			}
+
+			if(vifRegs->mode == 2)
+			{
+				//Update the reg rows for SSE
+				vifRow[0] = vifRegs->r0;
+				vifRow[1] = vifRegs->r1;
+				vifRow[2] = vifRegs->r2;
+				vifRow[3] = vifRegs->r3;
+			}
+			if(tempsize > 0) size = tempsize;
+
+		}
+		if (size >= ft->dsize && vifRegs->num > 0) //Else write what we do have
 		{
 			//VIF_LOG("warning, end with size = %d", size);