diff options
Diffstat (limited to 'include/glm/simd/common.h')
| -rw-r--r-- | include/glm/simd/common.h | 169 |
1 files changed, 169 insertions, 0 deletions
diff --git a/include/glm/simd/common.h b/include/glm/simd/common.h new file mode 100644 index 0000000..14af85b --- /dev/null +++ b/include/glm/simd/common.h @@ -0,0 +1,169 @@ +#if(GLM_COMPILER & GLM_COMPILER_VC) +#pragma warning(push) +#pragma warning(disable : 4510 4512 4610) +#endif + + union ieee754_QNAN + { + const float f; + struct i + { + const unsigned int mantissa:23, exp:8, sign:1; + }; + + ieee754_QNAN() : f(0.0)/*, mantissa(0x7FFFFF), exp(0xFF), sign(0x0)*/ {} + }; + +#if(GLM_COMPILER & GLM_COMPILER_VC) +#pragma warning(pop) +#endif + +static const __m128 GLM_VAR_USED glm_zero = _mm_setzero_ps(); +static const __m128 GLM_VAR_USED glm_one = _mm_set_ps1(1.0f); +static const __m128 GLM_VAR_USED glm_half = _mm_set_ps1(0.5f); +static const __m128 GLM_VAR_USED glm_minus_one = _mm_set_ps1(-1.0f); +static const __m128 GLM_VAR_USED glm_two = _mm_set_ps1(2.0f); +static const __m128 GLM_VAR_USED glm_three = _mm_set_ps1(3.0f); + +static const ieee754_QNAN glm_abs_mask; +static const __m128 GLM_VAR_USED glm_abs4_mask = _mm_set_ps1(glm_abs_mask.f); +static const __m128 GLM_VAR_USED glm_epi32_sign_mask = _mm_castsi128_ps(_mm_set1_epi32(static_cast<int>(0x80000000))); +static const __m128 GLM_VAR_USED glm_ps_2pow23 = _mm_set_ps1(8388608.0f); +static const __m128 GLM_VAR_USED glm_ps_1 = _mm_set_ps1(1.0f); + +GLM_FUNC_QUALIFIER __m128 glm_abs_ps(__m128 x) +{ + return _mm_and_ps(glm_abs4_mask, x); +} + +//sign +GLM_FUNC_QUALIFIER __m128 glm_sgn_ps(__m128 x) +{ + __m128 const Cmp0 = _mm_cmplt_ps(x, glm_zero); + __m128 const Cmp1 = _mm_cmpgt_ps(x, glm_zero); + __m128 const And0 = _mm_and_ps(Cmp0, glm_minus_one); + __m128 const And1 = _mm_and_ps(Cmp1, glm_one); + return _mm_or_ps(And0, And1); +} + +//round +GLM_FUNC_QUALIFIER __m128 glm_rnd_ps(__m128 x) +{ + __m128 const and0 = _mm_and_ps(glm_epi32_sign_mask, x); + __m128 const or0 = _mm_or_ps(and0, glm_ps_2pow23); + __m128 const add0 = _mm_add_ps(x, or0); + __m128 const sub0 = _mm_sub_ps(add0, or0); + return sub0; +} + +//floor +GLM_FUNC_QUALIFIER __m128 glm_flr_ps(__m128 x) +{ + __m128 const rnd0 = glm_rnd_ps(x); + __m128 const cmp0 = _mm_cmplt_ps(x, rnd0); + __m128 const and0 = _mm_and_ps(cmp0, glm_ps_1); + __m128 const sub0 = _mm_sub_ps(rnd0, and0); + return sub0; +} + +//trunc +//GLM_FUNC_QUALIFIER __m128 _mm_trc_ps(__m128 v) +//{ +// return __m128(); +//} + +//roundEven +GLM_FUNC_QUALIFIER __m128 glm_rde_ps(__m128 x) +{ + __m128 const and0 = _mm_and_ps(glm_epi32_sign_mask, x); + __m128 const or0 = _mm_or_ps(and0, glm_ps_2pow23); + __m128 const add0 = _mm_add_ps(x, or0); + __m128 const sub0 = _mm_sub_ps(add0, or0); + return sub0; +} + +GLM_FUNC_QUALIFIER __m128 glm_ceil_ps(__m128 x) +{ + __m128 const rnd0 = glm_rnd_ps(x); + __m128 const cmp0 = _mm_cmpgt_ps(x, rnd0); + __m128 const and0 = _mm_and_ps(cmp0, glm_ps_1); + __m128 const add0 = _mm_add_ps(rnd0, and0); + return add0; +} + +GLM_FUNC_QUALIFIER __m128 glm_frc_ps(__m128 x) +{ + __m128 const flr0 = glm_flr_ps(x); + __m128 const sub0 = _mm_sub_ps(x, flr0); + return sub0; +} + +GLM_FUNC_QUALIFIER __m128 glm_mod_ps(__m128 x, __m128 y) +{ + __m128 const div0 = _mm_div_ps(x, y); + __m128 const flr0 = glm_flr_ps(div0); + __m128 const mul0 = _mm_mul_ps(y, flr0); + __m128 const sub0 = _mm_sub_ps(x, mul0); + return sub0; +} + +GLM_FUNC_QUALIFIER __m128 glm_clp_ps(__m128 v, __m128 minVal, __m128 maxVal) +{ + __m128 const min0 = _mm_min_ps(v, maxVal); + __m128 const max0 = _mm_max_ps(min0, minVal); + return max0; +} + +GLM_FUNC_QUALIFIER __m128 glm_mix_ps(__m128 v1, __m128 v2, __m128 a) +{ + __m128 const sub0 = _mm_sub_ps(glm_one, a); + __m128 const mul0 = _mm_mul_ps(v1, sub0); + __m128 const mul1 = _mm_mul_ps(v2, a); + __m128 const add0 = _mm_add_ps(mul0, mul1); + return add0; +} + +//step +GLM_FUNC_QUALIFIER __m128 glm_stp_ps(__m128 edge, __m128 x) +{ + __m128 const cmp = _mm_cmple_ps(x, edge); + return _mm_movemask_ps(cmp) == 0 ? glm_one : glm_zero; +} + +// smoothstep +GLM_FUNC_QUALIFIER __m128 glm_ssp_ps(__m128 edge0, __m128 edge1, __m128 x) +{ + __m128 const sub0 = _mm_sub_ps(x, edge0); + __m128 const sub1 = _mm_sub_ps(edge1, edge0); + __m128 const div0 = _mm_sub_ps(sub0, sub1); + __m128 const clp0 = glm_clp_ps(div0, glm_zero, glm_one); + __m128 const mul0 = _mm_mul_ps(glm_two, clp0); + __m128 const sub2 = _mm_sub_ps(glm_three, mul0); + __m128 const mul1 = _mm_mul_ps(clp0, clp0); + __m128 const mul2 = _mm_mul_ps(mul1, sub2); + return mul2; +} + +/// \todo +//GLM_FUNC_QUALIFIER __m128 glm_nan_ps(__m128 x) +//{ +// __m128 empty; +// return empty; +//} + +/// \todo +//GLM_FUNC_QUALIFIER __m128 glm_inf_ps(__m128 x) +//{ +// __m128 empty; +// return empty; +//} + +// SSE scalar reciprocal sqrt using rsqrt op, plus one Newton-Rhaphson iteration +// By Elan Ruskin, http://assemblyrequired.crashworks.org/ +GLM_FUNC_QUALIFIER __m128 glm_sqrt_wip_ss(__m128 x) +{ + __m128 const recip = _mm_rsqrt_ss(x); // "estimate" opcode + __m128 const halfrecip = _mm_mul_ss(glm_half, recip); + __m128 const threeminus_xrr = _mm_sub_ss(glm_three, _mm_mul_ss(x, _mm_mul_ss(recip, recip))); + return _mm_mul_ss(halfrecip, threeminus_xrr); +} |