External/cglm
1
0
Fork 0
mirror of https://github.com/recp/cglm.git synced 2025-12-24 20:34:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

neon: add neon support for most vec4 operations

Browse Source
This commit is contained in:
Recep Aslantas
2019-01-21 23:14:04 +03:00
parent 07e60bd098
commit b117f3bf80
2 changed files with 76 additions and 13 deletions
Download Patch File Download Diff File Expand all files Collapse all files

18
include/cglm/simd/intrin.h
View File

@@ -105,10 +105,24 @@ glmm_store3(__m128 vx, float v[3]) {
#endif
#if defined(__SSE3__)
# include <x86intrin.h>
#endif
#if defined(__SSE4_1__)
# include <smmintrin.h>
#endif
#if defined(__SSE4_2__)
# include <nmmintrin.h>
#endif
/* ARM Neon */
#if defined(__ARM_NEON) && defined(__ARM_NEON_FP)
#if defined(__ARM_NEON)
# include <arm_neon.h>
# define CGLM_NEON_FP 1
# if defined(__ARM_NEON_FP)
# define CGLM_NEON_FP 1
# endif
#else
# undef CGLM_NEON_FP
#endif

71
include/cglm/vec4.h
View File

@@ -122,6 +122,8 @@ void
glm_vec4_copy(vec4 v, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, glmm_load(v));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vld1q_f32(v));
#else
dest[0] = v[0];
dest[1] = v[1];
@@ -157,6 +159,8 @@ void
glm_vec4_zero(vec4 v) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(v, _mm_setzero_ps());
#elif defined(CGLM_NEON_FP)
vst1q_f32(v, vdupq_n_f32(0.0f));
#else
v[0] = 0.0f;
v[1] = 0.0f;
@@ -175,6 +179,8 @@ void
glm_vec4_one(vec4 v) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(v, _mm_set1_ps(1.0f));
#elif defined(CGLM_NEON_FP)
vst1q_f32(v, vdupq_n_f32(1.0f));
#else
v[0] = 1.0f;
v[1] = 1.0f;
@@ -194,11 +200,24 @@ glm_vec4_one(vec4 v) {
CGLM_INLINE
float
glm_vec4_dot(vec4 a, vec4 b) {
#if defined( __SSE__ ) || defined( __SSE2__ )
#if (defined(__SSE4_1__) || defined(__SSE4_2__)) && defined(CGLM_SSE4_DOT)
return _mm_cvtss_f32(_mm_dp_ps(glmm_load(a), glmm_load(b), 0xFF));
#elif defined(__SSE3__) && defined(CGLM_SSE3_DOT)
__m128 x0, x1;
x0 = _mm_mul_ps(glmm_load(a), glmm_load(b));
x1 = _mm_hadd_ps(x0, x0);
return _mm_cvtss_f32(_mm_hadd_ps(x1, x1));
#elif defined(__SSE__) || defined(__SSE2__)
__m128 x0;
x0 = _mm_mul_ps(glmm_load(a), glmm_load(b));
x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 3, 2));
return _mm_cvtss_f32(_mm_add_ss(x0, glmm_shuff1(x0, 0, 1, 0, 1)));
#elif defined(CGLM_NEON_FP)
float32x4_t v0, v1, v2;
v0 = vmulq_f32(vld1q_f32(a), vld1q_f32(b));
v1 = vaddq_f32(v0, vrev64q_f32(v0));
v2 = vaddq_f32(v1, vcombine_f32(vget_high_f32(v1), vget_low_f32(v1)));
return vgetq_lane_f32(v2, 0);
#else
return a[0] * b[0] + a[1] * b[1] + a[2] * b[2] + a[3] * b[3];
#endif
@@ -218,15 +237,7 @@ glm_vec4_dot(vec4 a, vec4 b) {
CGLM_INLINE
float
glm_vec4_norm2(vec4 v) {
#if defined( __SSE__ ) || defined( __SSE2__ )
__m128 x0;
x0 = glmm_load(v);
x0 = _mm_mul_ps(x0, x0);
x0 = _mm_add_ps(x0, glmm_shuff1(x0, 1, 0, 3, 2));
return _mm_cvtss_f32(_mm_add_ss(x0, glmm_shuff1(x0, 0, 1, 0, 1)));
#else
return v[0] * v[0] + v[1] * v[1] + v[2] * v[2] + v[3] * v[3];
#endif
return glm_vec4_dot(v, v);
}
/*!
@@ -244,7 +255,7 @@ glm_vec4_norm(vec4 v) {
x0 = glmm_load(v);
return _mm_cvtss_f32(_mm_sqrt_ss(glmm_dot(x0, x0)));
#else
return sqrtf(glm_vec4_norm2(v));
return sqrtf(glm_vec4_dot(v, v));
#endif
}
@@ -260,6 +271,8 @@ void
glm_vec4_add(vec4 a, vec4 b, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_add_ps(glmm_load(a), glmm_load(b)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(a), vld1q_f32(b)));
#else
dest[0] = a[0] + b[0];
dest[1] = a[1] + b[1];
@@ -280,6 +293,8 @@ void
glm_vec4_adds(vec4 v, float s, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_add_ps(glmm_load(v), _mm_set1_ps(s)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(v), vdupq_n_f32(s)));
#else
dest[0] = v[0] + s;
dest[1] = v[1] + s;
@@ -300,6 +315,8 @@ void
glm_vec4_sub(vec4 a, vec4 b, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_sub_ps(glmm_load(a), glmm_load(b)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vsubq_f32(vld1q_f32(a), vld1q_f32(b)));
#else
dest[0] = a[0] - b[0];
dest[1] = a[1] - b[1];
@@ -320,6 +337,8 @@ void
glm_vec4_subs(vec4 v, float s, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_sub_ps(glmm_load(v), _mm_set1_ps(s)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vsubq_f32(vld1q_f32(v), vdupq_n_f32(s)));
#else
dest[0] = v[0] - s;
dest[1] = v[1] - s;
@@ -340,6 +359,8 @@ void
glm_vec4_mul(vec4 a, vec4 b, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_mul_ps(glmm_load(a), glmm_load(b)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vmulq_f32(vld1q_f32(a), vld1q_f32(b)));
#else
dest[0] = a[0] * b[0];
dest[1] = a[1] * b[1];
@@ -360,6 +381,8 @@ void
glm_vec4_scale(vec4 v, float s, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_mul_ps(glmm_load(v), _mm_set1_ps(s)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vmulq_f32(vld1q_f32(v), vdupq_n_f32(s)));
#else
dest[0] = v[0] * s;
dest[1] = v[1] * s;
@@ -442,6 +465,10 @@ glm_vec4_addadd(vec4 a, vec4 b, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_add_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vaddq_f32(vld1q_f32(a),
vld1q_f32(b))));
#else
dest[0] += a[0] + b[0];
dest[1] += a[1] + b[1];
@@ -466,6 +493,10 @@ glm_vec4_subadd(vec4 a, vec4 b, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_sub_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vsubq_f32(vld1q_f32(a),
vld1q_f32(b))));
#else
dest[0] += a[0] - b[0];
dest[1] += a[1] - b[1];
@@ -490,6 +521,10 @@ glm_vec4_muladd(vec4 a, vec4 b, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_mul_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vmulq_f32(vld1q_f32(a),
vld1q_f32(b))));
#else
dest[0] += a[0] * b[0];
dest[1] += a[1] * b[1];
@@ -514,6 +549,10 @@ glm_vec4_muladds(vec4 a, float s, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_mul_ps(glmm_load(a),
_mm_set1_ps(s))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vsubq_f32(vld1q_f32(a),
vdupq_n_f32(s))));
#else
dest[0] += a[0] * s;
dest[1] += a[1] * s;
@@ -538,6 +577,10 @@ glm_vec4_maxadd(vec4 a, vec4 b, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_max_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vmaxq_f32(vld1q_f32(a),
vld1q_f32(b))));
#else
dest[0] += glm_max(a[0], b[0]);
dest[1] += glm_max(a[1], b[1]);
@@ -562,6 +605,10 @@ glm_vec4_minadd(vec4 a, vec4 b, vec4 dest) {
glmm_store(dest, _mm_add_ps(glmm_load(dest),
_mm_min_ps(glmm_load(a),
glmm_load(b))));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, vaddq_f32(vld1q_f32(dest),
vminq_f32(vld1q_f32(a),
vld1q_f32(b))));
#else
dest[0] += glm_min(a[0], b[0]);
dest[1] += glm_min(a[1], b[1]);
@@ -581,6 +628,8 @@ void
glm_vec4_negate_to(vec4 v, vec4 dest) {
#if defined( __SSE__ ) || defined( __SSE2__ )
glmm_store(dest, _mm_xor_ps(glmm_load(v), _mm_set1_ps(-0.0f)));
#elif defined(CGLM_NEON_FP)
vst1q_f32(dest, veorq_s32(vld1q_f32(v), vdupq_n_f32(-0.0f)));
#else
dest[0] = -v[0];
dest[1] = -v[1];