Fixed build fixed SSE fast det

glm/core/intrinsic_matrix.inl

@@ -421,23 +421,23 @@ inline __m128 sse_det_ps
 	//T SubFactor05 = m[2][0] * m[3][1] - m[3][0] * m[2][1];
 
 	// First 2 columns
-	__m128 Swp2A = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(0, 1, 1, 2));
-	__m128 Swp3A = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(3, 2, 3, 3));
+ 	__m128 Swp2A = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(0, 1, 1, 2));
+ 	__m128 Swp3A = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(3, 2, 3, 3));
 	__m128 MulA = _mm_mul_ps(Swp2A, Swp3A);
 
 	// Second 2 columns
-	__m128 Swp2B = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(0, 1, 1, 2));
-	__m128 Swp3B = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(3, 2, 3, 3));
-	__m128 MulB = _mm_mul_ps(Swp2A, Swp3A);
+	__m128 Swp2B = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(3, 2, 3, 3));
+	__m128 Swp3B = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(0, 1, 1, 2));
+	__m128 MulB = _mm_mul_ps(Swp2B, Swp3B);
 
 	// Columns subtraction
-	__m128 SubAB = _mm_sub_ps(MulA, MulB);
+	__m128 SubE = _mm_sub_ps(MulA, MulB);
 
 	// Last 2 rows
-	__m128 Swp2C = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(1, 2, 0, 0));
-	__m128 Swp3C = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(0, 0, 1, 2));
+	__m128 Swp2C = _mm_shuffle_ps(m[2], m[2], _MM_SHUFFLE(0, 0, 1, 2));
+	__m128 Swp3C = _mm_shuffle_ps(m[3], m[3], _MM_SHUFFLE(1, 2, 0, 0));
 	__m128 MulC = _mm_mul_ps(Swp2C, Swp3C);
-	__m128 SwpD = __mm_hl_ps(MulC);
+	__m128 SubF = _mm_sub_ps(_mm_movehl_ps(MulC, MulC), MulC);
 
 	//detail::tvec4<T> DetCof(
 	//	+ (m[1][1] * SubFactor00 - m[1][2] * SubFactor01 + m[1][3] * SubFactor02),
@@ -445,29 +445,31 @@ inline __m128 sse_det_ps
 	//	+ (m[1][0] * SubFactor01 - m[1][1] * SubFactor03 + m[1][3] * SubFactor05),
 	//	- (m[1][0] * SubFactor02 - m[1][1] * SubFactor04 + m[1][2] * SubFactor05));
 
-	__128 SubFacA = _mm_shuffle_ps(SubAB, SubAB, _MM_SHUFFLE(2, 1, 0, 0));
-	__128 SwpFacA = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(0, 0, 0, 1));
-	__128 MulFacA = __mm_mul_ps(SwpFacA, SubFacA);
+	__m128 SubFacA = _mm_shuffle_ps(SubE, SubE, _MM_SHUFFLE(2, 1, 0, 0));
+	__m128 SwpFacA = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(0, 0, 0, 1));
+	__m128 MulFacA = _mm_mul_ps(SwpFacA, SubFacA);
 
-	__128 SubFacB = ;
-	__128 SwpFacB = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(1, 1, 2, 2));
-	__128 MulFacB = __mm_mul_ps(SwpFacB, SubFacB);
+	__m128 SubTmpB = _mm_shuffle_ps(SubE, SubF, _MM_SHUFFLE(0, 0, 3, 1));
+	__m128 SubFacB = _mm_shuffle_ps(SubTmpB, SubTmpB, _MM_SHUFFLE(3, 1, 1, 0));//SubF[0], SubE[3], SubE[3], SubE[1];
+	__m128 SwpFacB = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(1, 1, 2, 2));
+	__m128 MulFacB = _mm_mul_ps(SwpFacB, SubFacB);
 
-	__128 SubRes = __mm_sub_ps(MulFacA, MulFacA);
+	__m128 SubRes = _mm_sub_ps(MulFacA, MulFacB);
 
-	__128 SubFacC = ;
-	__128 SwpFacC = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(1, 1, 2, 2));
-	__128 MulFacC = __mm_mul_ps(SwpFacC, SubFacC);
+	__m128 SubTmpC = _mm_shuffle_ps(SubE, SubF, _MM_SHUFFLE(1, 0, 2, 2));
+	__m128 SubFacC = _mm_shuffle_ps(SubTmpC, SubTmpC, _MM_SHUFFLE(3, 3, 2, 0));
+	__m128 SwpFacC = _mm_shuffle_ps(m[1], m[1], _MM_SHUFFLE(2, 3, 3, 3));
+	__m128 MulFacC = _mm_mul_ps(SwpFacC, SubFacC);
 
-	__m128 AddRes = __mm_add_ps(SubRes, MulFacC);
-	__m128 DetCof = __mm_mul_ps(AddRes, _mm_set_ps( 1.0f,-1.0f, 1.0f,-1.0f));
+	__m128 AddRes = _mm_add_ps(SubRes, MulFacC);
+	__m128 DetCof = _mm_mul_ps(AddRes, _mm_setr_ps( 1.0f,-1.0f, 1.0f,-1.0f));
 
 	//return m[0][0] * DetCof[0]
 	//	 + m[0][1] * DetCof[1]
 	//	 + m[0][2] * DetCof[2]
 	//	 + m[0][3] * DetCof[3];
 
-	return sse_dot_ps(m[0], Signed);
+	return sse_dot_ps(m[0], DetCof);
 }
 
 inline void sse_inverse_ps(__m128 const in[4], __m128 out[4])
@@ -692,7 +694,7 @@ inline void sse_inverse_ps(__m128 const in[4], __m128 out[4])
 	//						+ m[0][1] * Inverse[1][0] 
 	//						+ m[0][2] * Inverse[2][0] 
 	//						+ m[0][3] * Inverse[3][0];
-	__m128 Det0 = _mm_dot_ps(in[0], Row2);
+	__m128 Det0 = sse_dot_ps(in[0], Row2);
 	__m128 Rcp0 = _mm_div_ps(one, Det0);
 	//__m128 Rcp0 = _mm_rcp_ps(Det0);
 
@@ -1008,10 +1010,10 @@ void sse_rotate_ps(__m128 const in[4], float Angle, float const v[3], __m128 out
 
 void sse_outer_ps(__m128 const & c, __m128 const & r, __m128 out[4])
 {
-	out[0] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(0, 0, 0, 0));
-	out[1] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(1, 1, 1, 1));
-	out[2] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(2, 2, 2, 2));
-	out[3] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(3, 3, 3, 3));
+	out[0] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(0, 0, 0, 0)));
+	out[1] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(1, 1, 1, 1)));
+	out[2] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(2, 2, 2, 2)));
+	out[3] = _mm_mul_ps(c, _mm_shuffle_ps(r, r, _MM_SHUFFLE(3, 3, 3, 3)));
 }
 
 }//namespace detail

glm/gtx/simd_mat4.inl

@@ -132,7 +132,7 @@ namespace detail
 		fmat4x4SIMD const & m
 	)
     {
-		_mm_mul_ps(&this->Data[0].Data, &m.Data[0].Data, &this->Data[0].Data);
+		sse_mul_ps(&this->Data[0].Data, &m.Data[0].Data, &this->Data[0].Data);
         return *this;
     }
 
@@ -142,8 +142,8 @@ namespace detail
 	)
     {
 		__m128 Inv[4];
-		_mm_inverse_ps(&this->Data[0].Data, Inv);
-		_mm_mul_ps(&this->Data[0].Data, Inv, &this->Data[0].Data);
+		sse_inverse_ps(&this->Data[0].Data, Inv);
+		sse_mul_ps(&this->Data[0].Data, Inv, &this->Data[0].Data);
         return *this;
     }
 
@@ -234,7 +234,7 @@ namespace simd_mat4
 	inline detail::fmat4x4SIMD simd_transpose(detail::fmat4x4SIMD const & m)
 	{
 		detail::fmat4x4SIMD result;
-		sse_transpose_ps(&m[0].Data, &result[0].Data);
+		detail::sse_transpose_ps(&m[0].Data, &result[0].Data);
 		return result;
 	}
 
@@ -246,7 +246,7 @@ namespace simd_mat4
 	inline detail::fmat4x4SIMD simd_inverse(detail::fmat4x4SIMD const & m)
 	{
 		detail::fmat4x4SIMD result;
-		sse_inverse_ps(&m[0].Data, &result[0].Data);
+		detail::sse_inverse_ps(&m[0].Data, &result[0].Data);
 		return result;
 	}
 }//namespace simd_mat4

glm/gtx/simd_vec4.inl

@@ -273,7 +273,7 @@ namespace glm
 				detail::fvec4SIMD const & b
 			)
 			{
-				return _mm_xpd_ps(a.Data, b.Data);
+				return detail::sse_xpd_ps(a.Data, b.Data);
 			}
 #		else//(GLM_INSTRUCTION_SET & GLM_INSTRUCTION_SET_PURE)
 			inline detail::fvec4SIMD cross
@@ -282,7 +282,7 @@ namespace glm
 				detail::fvec4SIMD const & b
 			)
 			{
-				return glm::cross(a, b);
+				return detail::sse_xpd_ps(a.Data, b.Data);
 			}
 #		endif
 	}//namespace simd_vec4

test/gtx/gtx-simd-mat4.cpp

@@ -30,8 +30,8 @@ int main(int argc, void* argv[])
 		glm::simd_vec4(0.5f, 3.0f, 0.6f, 0.02f),
 		glm::simd_vec4(0.2f, 0.4f, 2.0f, 0.03f),
 		glm::simd_vec4(4.0f, 3.0f, 2.0f, 1.00f));
-	__m128 DetB = sse_slow_det_ps(&IdentityB.Data[0].Data); 
-	__m128 DetC = sse_det_ps(&IdentityB.Data[0].Data);
+	__m128 DetB = glm::detail::sse_slow_det_ps(&IdentityB.Data[0].Data); 
+	__m128 DetC = glm::detail::sse_det_ps(&IdentityB.Data[0].Data);
 
 	return 0;
 }