diff --git a/code/nel/include/nel/3d/matrix_3x4.h b/code/nel/include/nel/3d/matrix_3x4.h
index d7ed660fc..45901c20e 100644
--- a/code/nel/include/nel/3d/matrix_3x4.h
+++ b/code/nel/include/nel/3d/matrix_3x4.h
@@ -108,281 +108,6 @@ public:
 };
 
 
-// ***************************************************************************
-// ***************************************************************************
-// SSE Matrix
-// ***************************************************************************
-// ***************************************************************************
-
-
-// ***************************************************************************
-#if defined(NL_OS_WINDOWS) && !defined(NL_NO_ASM)
-
-
-/** For fast vector/point multiplication. Special usage for Skinning.
- *	NB: SSE is no more used (no speed gain, some memory problem), but keep it for possible future usage.
- */
-class	CMatrix3x4SSE
-{
-public:
-	// Order them in memory column first, for SSE column multiplication.
-	float	a11, a21, a31, a41;
-	float	a12, a22, a32, a42;
-	float	a13, a23, a33, a43;
-	float	a14, a24, a34, a44;
-
-	// Copy from a matrix.
-	void	set(const CMatrix &mat)
-	{
-		const float	*m =mat.get();
-		a11= m[0]; a12= m[4]; a13= m[8] ; a14= m[12];
-		a21= m[1]; a22= m[5]; a23= m[9] ; a24= m[13];
-		a31= m[2]; a32= m[6]; a33= m[10]; a34= m[14];
-		// not used.
-		a41= 0   ; a42= 0   ; a43= 0    ; a44= 1;
-	}
-
-
-	// mulSetvector. NB: in should be different as v!! (else don't work).
-	void	mulSetVector(const CVector &vin, CVector &vout)
-	{
-		__asm
-		{
-			mov		eax, vin
-			mov		ebx, this
-			mov		edi, vout
-			// Load in vector in op[0]
-			movss	xmm0, [eax]vin.x
-			movss	xmm1, [eax]vin.y
-			movss	xmm2, [eax]vin.z
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm0, xmm0, 0
-			shufps	xmm1, xmm1, 0
-			shufps	xmm2, xmm2, 0
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-
-			// write the result.
-			movss	[edi]vout.x, xmm0
-			shufps	xmm0, xmm0, 33
-			movss	[edi]vout.y, xmm0
-			movhlps	xmm0, xmm0
-			movss	[edi]vout.z, xmm0
-		}
-	}
-	// mulSetpoint. NB: in should be different as v!! (else don't work).
-	void	mulSetPoint(const CVector &vin, CVector &vout)
-	{
-		__asm
-		{
-			mov		eax, vin
-			mov		ebx, this
-			mov		edi, vout
-			// Load in vector in op[0]
-			movss	xmm0, [eax]vin.x
-			movss	xmm1, [eax]vin.y
-			movss	xmm2, [eax]vin.z
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm0, xmm0, 0
-			shufps	xmm1, xmm1, 0
-			shufps	xmm2, xmm2, 0
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-			// Add Matrix translate column vector
-			addps	xmm0, [ebx]this.a14
-
-			// write the result.
-			movss	[edi]vout.x, xmm0
-			shufps	xmm0, xmm0, 33
-			movss	[edi]vout.y, xmm0
-			movhlps	xmm0, xmm0
-			movss	[edi]vout.z, xmm0
-		}
-	}
-
-
-	// mulSetvector. NB: vin should be different as v!! (else don't work).
-	void	mulSetVector(const CVector &vin, float scale, CVector &vout)
-	{
-		__asm
-		{
-			mov		eax, vin
-			mov		ebx, this
-			mov		edi, vout
-			// Load in vector in op[0]
-			movss	xmm0, [eax]vin.x
-			movss	xmm1, [eax]vin.y
-			movss	xmm2, [eax]vin.z
-			// Load scale in op[0]
-			movss	xmm3, scale
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm0, xmm0, 0
-			shufps	xmm1, xmm1, 0
-			shufps	xmm2, xmm2, 0
-			shufps	xmm3, xmm3, 0
-			// Store vertex column in other regs.
-			movaps	xmm5, xmm0
-			movaps	xmm6, xmm1
-			movaps	xmm7, xmm2
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-
-			// mul final result with scale
-			mulps	xmm0, xmm3
-
-			// store it in xmm4 for future use.
-			movaps	xmm4, xmm0
-		}
-	}
-	// mulSetpoint. NB: vin should be different as v!! (else don't work).
-	void	mulSetPoint(const CVector &vin, float scale, CVector &vout)
-	{
-		__asm
-		{
-			mov		eax, vin
-			mov		ebx, this
-			mov		edi, vout
-			// Load in vector in op[0]
-			movss	xmm0, [eax]vin.x
-			movss	xmm1, [eax]vin.y
-			movss	xmm2, [eax]vin.z
-			// Load scale in op[0]
-			movss	xmm3, scale
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm0, xmm0, 0
-			shufps	xmm1, xmm1, 0
-			shufps	xmm2, xmm2, 0
-			shufps	xmm3, xmm3, 0
-			// Store vertex column in other regs.
-			movaps	xmm5, xmm0
-			movaps	xmm6, xmm1
-			movaps	xmm7, xmm2
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-			// Add Matrix translate column vector
-			addps	xmm0, [ebx]this.a14
-
-			// mul final result with scale
-			mulps	xmm0, xmm3
-
-			// store it in xmm4 for future use.
-			movaps	xmm4, xmm0
-		}
-	}
-
-
-	// mulAddvector. NB: vin should be different as v!! (else don't work).
-	void	mulAddVector(const CVector &/* vin */, float scale, CVector &vout)
-	{
-		__asm
-		{
-			mov		ebx, this
-			mov		edi, vout
-			// Load vin vector loaded in mulSetVector
-			movaps	xmm0, xmm5
-			movaps	xmm1, xmm6
-			movaps	xmm2, xmm7
-			// Load scale in op[0]
-			movss	xmm3, scale
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm3, xmm3, 0
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-
-			// mul final result with scale
-			mulps	xmm0, xmm3
-
-			// Add result, with prec sum.
-			addps	xmm0, xmm4
-
-			// store it in xmm4 for future use.
-			movaps	xmm4, xmm0
-
-			// write the result.
-			movss	[edi]vout.x, xmm0
-			shufps	xmm0, xmm0, 33
-			movss	[edi]vout.y, xmm0
-			movhlps	xmm0, xmm0
-			movss	[edi]vout.z, xmm0
-		}
-	}
-	// mulAddpoint. NB: vin should be different as v!! (else don't work).
-	void	mulAddPoint(const CVector &/* vin */, float scale, CVector &vout)
-	{
-		__asm
-		{
-			mov		ebx, this
-			mov		edi, vout
-			// Load vin vector loaded in mulSetPoint
-			movaps	xmm0, xmm5
-			movaps	xmm1, xmm6
-			movaps	xmm2, xmm7
-			// Load scale in op[0]
-			movss	xmm3, scale
-			// Expand op[0] to op[1], op[2], op[3]
-			shufps	xmm3, xmm3, 0
-			// Mul each vector with 3 Matrix column
-			mulps	xmm0, [ebx]this.a11
-			mulps	xmm1, [ebx]this.a12
-			mulps	xmm2, [ebx]this.a13
-			// Add each column vector.
-			addps	xmm0, xmm1
-			addps	xmm0, xmm2
-			// Add Matrix translate column vector
-			addps	xmm0, [ebx]this.a14
-
-			// mul final result with scale
-			mulps	xmm0, xmm3
-
-			// Add result, with prec sum.
-			addps	xmm0, xmm4
-
-			// store it in xmm4 for future use.
-			movaps	xmm4, xmm0
-
-			// write the result.
-			movss	[edi]vout.x, xmm0
-			shufps	xmm0, xmm0, 33
-			movss	[edi]vout.y, xmm0
-			movhlps	xmm0, xmm0
-			movss	[edi]vout.z, xmm0
-		}
-	}
-
-};
-
-#else // NL_OS_WINDOWS
-/// dummy CMatrix3x4SSE for non windows platform
-class CMatrix3x4SSE : public  CMatrix3x4 { };
-#endif
-
-
-
 } // NL3D
 
 
diff --git a/code/nel/src/3d/mesh_mrm_skin.cpp b/code/nel/src/3d/mesh_mrm_skin.cpp
index 13e8bdd21..a34eeb766 100644
--- a/code/nel/src/3d/mesh_mrm_skin.cpp
+++ b/code/nel/src/3d/mesh_mrm_skin.cpp
@@ -39,124 +39,6 @@ namespace NL3D
 {
 
 
-// ***************************************************************************
-// ***************************************************************************
-// CMatrix3x4SSE array correctly aligned
-// ***************************************************************************
-// ***************************************************************************
-
-
-
-// ***************************************************************************
-#define	NL3D_SSE_ALIGNEMENT		16
-/**
- *	A CMatrix3x4SSE array correctly aligned
- *	NB: SSE is no more used (no speed gain, some memory problem), but keep it for possible future usage.
- */
-class	CMatrix3x4SSEArray
-{
-private:
-	void	*_AllocData;
-	void	*_Data;
-	uint	_Size;
-	uint	_Capacity;
-
-public:
-	CMatrix3x4SSEArray()
-	{
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-	}
-	~CMatrix3x4SSEArray()
-	{
-		clear();
-	}
-	CMatrix3x4SSEArray(const CMatrix3x4SSEArray &other)
-	{
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-		*this= other;
-	}
-	CMatrix3x4SSEArray &operator=(const CMatrix3x4SSEArray &other)
-	{
-		if( this == &other)
-			return *this;
-		resize(other.size());
-		// copy data from aligned pointers to aligned pointers.
-		memcpy(_Data, other._Data, size() * sizeof(CMatrix3x4SSE) );
-
-		return *this;
-	}
-
-
-	CMatrix3x4SSE	*getPtr()
-	{
-		return (CMatrix3x4SSE*)_Data;
-	}
-
-	void	clear()
-	{
-		delete [] ((uint8 *)_AllocData);
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-	}
-
-	void	resize(uint n)
-	{
-		// reserve ??
-		if(n>_Capacity)
-			reserve( max(2*_Capacity, n));
-		_Size= n;
-	}
-
-	void	reserve(uint n)
-	{
-		if(n==0)
-			clear();
-		else if(n>_Capacity)
-		{
-			// Alloc new data.
-			void	*newAllocData;
-			void	*newData;
-
-			// Alloc for alignement.
-			newAllocData= new uint8 [n * sizeof(CMatrix3x4SSE) + NL3D_SSE_ALIGNEMENT-1];
-			if(newAllocData==NULL)
-				throw Exception("SSE Allocation Failed");
-
-			// Align ptr
-			newData= (void*) ( ((ptrdiff_t)newAllocData+NL3D_SSE_ALIGNEMENT-1) & (~(NL3D_SSE_ALIGNEMENT-1)) );
-
-			// copy valid data from old to new.
-			memcpy(newData, _Data, size() * sizeof(CMatrix3x4SSE) );
-
-			// release old.
-			if(_AllocData)
-				delete [] ((uint8*)_AllocData);
-
-			// change ptrs and capacity.
-			_Data= newData;
-			_AllocData= newAllocData;
-			_Capacity= n;
-
-			// TestYoyo
-			//nlwarning("YOYO Tst SSE P4: %X, %d", _Data, n);
-		}
-	}
-
-	uint	size() const {return _Size;}
-
-
-	CMatrix3x4SSE	&operator[](uint i) {return ((CMatrix3x4SSE*)_Data)[i];}
-};
-
-
 
 // ***************************************************************************
 // ***************************************************************************
diff --git a/code/nel/src/3d/mesh_mrm_skinned.cpp b/code/nel/src/3d/mesh_mrm_skinned.cpp
index 2b1c3beb6..6af29bf73 100644
--- a/code/nel/src/3d/mesh_mrm_skinned.cpp
+++ b/code/nel/src/3d/mesh_mrm_skinned.cpp
@@ -2247,123 +2247,6 @@ void CMeshMRMSkinnedGeom::getSkinWeights (std::vector<CMesh::CSkinWeight> &skinW
 	}
 }
 
-// ***************************************************************************
-// ***************************************************************************
-// CMatrix3x4SSE array correctly aligned
-// ***************************************************************************
-// ***************************************************************************
-
-
-
-// ***************************************************************************
-#define	NL3D_SSE_ALIGNEMENT		16
-/**
- *	A CMatrix3x4SSEArray array correctly aligned
- *	NB: SSE is no more used (no speed gain, some memory problem), but keep it for possible future usage.
- */
-class	CMatrix3x4SSEArray
-{
-private:
-	void	*_AllocData;
-	void	*_Data;
-	uint	_Size;
-	uint	_Capacity;
-
-public:
-	CMatrix3x4SSEArray()
-	{
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-	}
-	~CMatrix3x4SSEArray()
-	{
-		clear();
-	}
-	CMatrix3x4SSEArray(const CMatrix3x4SSEArray &other)
-	{
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-		*this= other;
-	}
-	CMatrix3x4SSEArray &operator=(const CMatrix3x4SSEArray &other)
-	{
-		if( this == &other)
-			return *this;
-		resize(other.size());
-		// copy data from aligned pointers to aligned pointers.
-		memcpy(_Data, other._Data, size() * sizeof(CMatrix3x4SSE) );
-
-		return *this;
-	}
-
-
-	CMatrix3x4SSE	*getPtr()
-	{
-		return (CMatrix3x4SSE*)_Data;
-	}
-
-	void	clear()
-	{
-		delete [] ((uint8 *) _AllocData);
-		_AllocData= NULL;
-		_Data= NULL;
-		_Size= 0;
-		_Capacity= 0;
-	}
-
-	void	resize(uint n)
-	{
-		// reserve ??
-		if(n>_Capacity)
-			reserve( max(2*_Capacity, n));
-		_Size= n;
-	}
-
-	void	reserve(uint n)
-	{
-		if(n==0)
-			clear();
-		else if(n>_Capacity)
-		{
-			// Alloc new data.
-			void	*newAllocData;
-			void	*newData;
-
-			// Alloc for alignement.
-			newAllocData= new uint8 [n * sizeof(CMatrix3x4SSE) + NL3D_SSE_ALIGNEMENT-1];
-			if(newAllocData==NULL)
-				throw Exception("SSE Allocation Failed");
-
-			// Align ptr
-			newData= (void*) ( ((ptrdiff_t)newAllocData+NL3D_SSE_ALIGNEMENT-1) & (~(NL3D_SSE_ALIGNEMENT-1)) );
-
-			// copy valid data from old to new.
-			memcpy(newData, _Data, size() * sizeof(CMatrix3x4SSE) );
-
-			// release old.
-			if(_AllocData)
-				delete [] ((uint8*)_AllocData);
-
-			// change ptrs and capacity.
-			_Data= newData;
-			_AllocData= newAllocData;
-			_Capacity= n;
-
-			// TestYoyo
-			//nlwarning("YOYO Tst SSE P4: %X, %d", _Data, n);
-		}
-	}
-
-	uint	size() const {return _Size;}
-
-
-	CMatrix3x4SSE	&operator[](uint i) {return ((CMatrix3x4SSE*)_Data)[i];}
-};
-
 
 // ***************************************************************************
 // ***************************************************************************