AuroraMiddleware/glm
1
0
Fork 0
Code Issues Pull Requests Releases Wiki Activity

Added FMA SIMD optimization

Browse Source
This commit is contained in:
Christophe Riccio 2016-05-28 12:31:43 +02:00
parent 71e6b537cc
commit fdec412ff7
2 changed files with 21 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
glm/simd/common.h
View File

@ -162,6 +162,24 @@ GLM_FUNC_QUALIFIER __m128 glm_f32v4_inf(__m128 x)
return _mm_castsi128_ps(_mm_cmpeq_epi32(t2, _mm_set1_epi32(0xFF000000))); // exponent is all 1s, fraction is 0
}
GLM_FUNC_QUALIFIER __m128 glm_f32v1_fma(__m128 a, __m128 b, __m128 c)
{
# if GLM_ARCH & GLM_ARCH_AVX2
return _mm_fmadd_ss(a, b, c);
# else
return _mm_add_ss(_mm_mul_ss(a, b), c);
# endif
}
GLM_FUNC_QUALIFIER __m128 glm_f32v4_fma(__m128 a, __m128 b, __m128 c)
{
# if GLM_ARCH & GLM_ARCH_AVX2
return _mm_fmadd_ps(a, b, c);
# else
return _mm_add_ps(_mm_mul_ps(a, b), c);
# endif
}
// SSE scalar reciprocal sqrt using rsqrt op, plus one Newton-Rhaphson iteration
// By Elan Ruskin, http://assemblyrequired.crashworks.org/
GLM_FUNC_QUALIFIER __m128 glm_f32v1_sqrt_wip(__m128 x)
@ -188,3 +206,4 @@ GLM_FUNC_QUALIFIER __m128 glm_f32v4_sqrt_wip(__m128 x)
return Mul3;
}

2
test/core/core_func_common.cpp
View File

@ -1237,6 +1237,8 @@ int main()
{
int Error = 0;
__m128 const flr0 = glm_f32v4_flr(_mm_set_ps(1.1f, 1.9f, -1.1f, -1.9f));
glm::ivec4 const a(1);
glm::ivec4 const b = ~a;