mewin/glm
1
0
Fork 0
Code Releases Activity

Fixed SIMD code path selection

Browse Source
This commit is contained in:
Christophe Riccio
2016-05-28 18:24:39 +02:00
parent fdb985a0eb
commit d33e3df02a
9 changed files with 41 additions and 31 deletions
Download Patch File Download Diff File Expand all files Collapse all files

2
glm/detail/func_matrix_simd.inl
View File

@@ -8,7 +8,7 @@
namespace glm{
namespace detail
{
# if GLM_ARCH & GLM_ARCH_SSE2
# if GLM_ARCH & GLM_ARCH_SSE2_FLAG
template <precision P>
struct compute_inverse<tmat4x4, float, P>
{

26
glm/detail/setup.hpp
View File

@@ -67,25 +67,29 @@
# define GLM_MESSAGE_ARCH_DISPLAYED
# if(GLM_ARCH == GLM_ARCH_PURE)
# pragma message("GLM: Platform independent code")
# elif(GLM_ARCH & GLM_ARCH_ARM)
# pragma message("GLM: ARM instruction set")
# elif(GLM_ARCH & GLM_ARCH_AVX2)
# elif(GLM_ARCH == GLM_ARCH_AVX2)
# pragma message("GLM: AVX2 instruction set")
# elif(GLM_ARCH & GLM_ARCH_AVX)
# elif(GLM_ARCH == GLM_ARCH_AVX)
# pragma message("GLM: AVX instruction set")
# elif(GLM_ARCH & GLM_ARCH_SSE3)
# elif(GLM_ARCH == GLM_ARCH_SSE42)
# pragma message("GLM: SSE4.2 instruction set")
# elif(GLM_ARCH == GLM_ARCH_SSE41)
# pragma message("GLM: SSE4.1 instruction set")
# elif(GLM_ARCH == GLM_ARCH_SSSE3)
# pragma message("GLM: SSSE3 instruction set")
# elif(GLM_ARCH == GLM_ARCH_SSE3)
# pragma message("GLM: SSE3 instruction set")
# elif(GLM_ARCH & GLM_ARCH_SSE2)
# elif(GLM_ARCH == GLM_ARCH_SSE2)
# pragma message("GLM: SSE2 instruction set")
# elif(GLM_ARCH & GLM_ARCH_X86)
# elif(GLM_ARCH == GLM_ARCH_X86)
# pragma message("GLM: x86 instruction set")
# elif(GLM_ARCH & GLM_ARCH_NEON)
# elif(GLM_ARCH == GLM_ARCH_NEON)
# pragma message("GLM: NEON instruction set")
# elif(GLM_ARCH & GLM_ARCH_ARM)
# elif(GLM_ARCH == GLM_ARCH_ARM)
# pragma message("GLM: ARM instruction set")
# elif(GLM_ARCH & GLM_ARCH_MIPS)
# elif(GLM_ARCH == GLM_ARCH_MIPS)
# pragma message("GLM: MIPS instruction set")
# elif(GLM_ARCH & GLM_ARCH_PPC)
# elif(GLM_ARCH == GLM_ARCH_PPC)
# pragma message("GLM: PowerPC architechture")
# endif//GLM_ARCH
#endif//GLM_MESSAGE

2
glm/gtx/simd_mat4.hpp
View File

@@ -17,7 +17,7 @@
#if(GLM_ARCH != GLM_ARCH_PURE)
#if(GLM_ARCH & GLM_ARCH_SSE2)
#if(GLM_ARCH & GLM_ARCH_SSE2_FLAG)
# include "../detail/intrinsic_matrix.hpp"
# include "../gtx/simd_vec4.hpp"
#else

2
glm/gtx/simd_quat.hpp
View File

@@ -19,7 +19,7 @@
#if(GLM_ARCH != GLM_ARCH_PURE)
#if(GLM_ARCH & GLM_ARCH_SSE2)
#if(GLM_ARCH & GLM_ARCH_SSE2_FLAG)
# include "../gtx/simd_mat4.hpp"
#else
# error "GLM: GLM_GTX_simd_quat requires compiler support of SSE2 through intrinsics"

2
glm/gtx/simd_quat.inl
View File

@@ -122,7 +122,7 @@ GLM_FUNC_QUALIFIER fquatSIMD operator* (fquatSIMD const & q1, fquatSIMD const &
__m128 mul2 = _mm_mul_ps(q1.Data, _mm_shuffle_ps(q2.Data, q2.Data, _MM_SHUFFLE(2, 3, 0, 1)));
__m128 mul3 = _mm_mul_ps(q1.Data, q2.Data);
# if((GLM_ARCH & GLM_ARCH_SSE4))
# if(GLM_ARCH & GLM_ARCH_SSE41_FLAG)
__m128 add0 = _mm_dp_ps(mul0, _mm_set_ps(1.0f, -1.0f, 1.0f, 1.0f), 0xff);
__m128 add1 = _mm_dp_ps(mul1, _mm_set_ps(1.0f, 1.0f, 1.0f, -1.0f), 0xff);
__m128 add2 = _mm_dp_ps(mul2, _mm_set_ps(1.0f, 1.0f, -1.0f, 1.0f), 0xff);

2
glm/gtx/simd_vec4.hpp
View File

@@ -17,7 +17,7 @@
#if(GLM_ARCH != GLM_ARCH_PURE)
#if(GLM_ARCH & GLM_ARCH_SSE2)
#if(GLM_ARCH & GLM_ARCH_SSE2_FLAG)
# include "../detail/intrinsic_common.hpp"
# include "../detail/intrinsic_geometric.hpp"
# include "../detail/intrinsic_integer.hpp"

2
glm/simd/geometric.h
View File

@@ -119,4 +119,4 @@ GLM_FUNC_QUALIFIER __m128 glm_f32v4_rfa(__m128 I, __m128 N, __m128 eta)
return sub2;
}
#endif//GLM_ARCH & GLM_ARCH_SSE2
#endif//GLM_ARCH & GLM_ARCH_SSE2_FLAG