-
sparker
thiago
Manta Interactive Ray Tracer Development Mailing List
Text archives Help
- From: "James Bigler" <bigler@cs.utah.edu>
- To: manta@sci.utah.edu
- Subject: [Manta] r2060 - trunk/Model/Primitives
- Date: Wed, 13 Feb 2008 15:46:10 -0700 (MST)
Author: bigler
Date: Wed Feb 13 15:46:10 2008
New Revision: 2060
Added:
trunk/Model/Primitives/SphereRTSL.cc
Modified:
trunk/Model/Primitives/Sphere.cc
Log:
Model/Primitives/Sphere.cc
Model/Primitives/SphereRTSL.cc
Included some code to perform the RTSL compiled version of Thiago's
modification to intersect. This intersection method is more robust to a
smaller T_EPSILON. Initial benchmarks put it slightly slower than the hand
written Sphere code, but perhaps we could hand tweak some of the code.
Modified: trunk/Model/Primitives/Sphere.cc
==============================================================================
--- trunk/Model/Primitives/Sphere.cc (original)
+++ trunk/Model/Primitives/Sphere.cc Wed Feb 13 15:46:10 2008
@@ -14,6 +14,8 @@
using namespace Manta;
using namespace std;
+#define USE_RTSL_OUTPUT 1
+
Sphere::Sphere()
{
}
@@ -48,42 +50,12 @@
bbox.extendBySphere(center, radius);
}
+#if USE_RTSL_OUTPUT
+# include "SphereRTSL.cc"
+#else
+
void Sphere::intersect(const RenderContext&, RayPacket& rays) const
{
-#if 0
- //This version will find a t value close to the actual value
- //(actually finds the midpoint between the two intersections), and
- //then uses that initial t to cast a ray from there in order to get
- //a more exact t value. This is more robust and allows for a smaller
- //T_EPSILON.
- for(int i = rays.begin();i<rays.end();i++){
- Vector O(rays.getOrigin(i)-center);
- const Vector D(rays.getDirection(i));
- const Real A = Dot(D, D);
- const Real A_inv = static_cast<Real>(1)/A;
- Real B = Dot(O, D);
- Real C = Dot(O, O) - radius*radius;
- Real disc = B*B-A*C;
- if(disc >= 0){
- Real t = -B*A_inv; //average of two t intersections.
- O += D*t;
- Real B = Dot(O, D);
- Real C = Dot(O, O) - radius*radius;
- disc = B*B-A*C;
-
- if (disc >= 0) {
- const Real r = Sqrt(disc);
- const Real t1 = t + -(r+B)*A_inv;
- if (t1 <= T_EPSILON)
- t += (r-B)*A_inv;
- else
- t = t1;
-
- rays.hit(i, t, getMaterial(), this, getTexCoordMapper());
- }
- }
- }
-#else
switch(rays.getAllFlags() &
(RayPacket::ConstantOrigin|RayPacket::NormalizedDirections)){
case RayPacket::ConstantOrigin|RayPacket::NormalizedDirections:
{
@@ -349,7 +321,6 @@
}
break;
}
-#endif
}
void Sphere::computeNormal(const RenderContext&, RayPacket& rays) const
@@ -362,6 +333,7 @@
}
rays.setFlag(RayPacket::HaveUnitNormals);
}
+#endif // #if USE_RTSL_OUTPUT
void Sphere::computeTexCoords2(const RenderContext&,
RayPacket& rays) const
Added: trunk/Model/Primitives/SphereRTSL.cc
==============================================================================
--- (empty file)
+++ trunk/Model/Primitives/SphereRTSL.cc Wed Feb 13 15:46:10 2008
@@ -0,0 +1,2353 @@
+void Sphere::intersect(const RenderContext& context, RayPacket& rays) const
+{
+ if(rays.getAllFlags() & RayPacket::NormalizedDirections){
+ // RayPacket::NormalizedDirections
+ if(rays.getAllFlags() & RayPacket::ConstantOrigin){
+ // RayPacket::ConstantOrigin
+ // Special case: constant origin, unit directions
+
+ // Begin loop invariant setup code
+ Vector rtsl_O = (rays.getOrigin(0) - center);
+ float rtsl_A = 1.f;
+ float rtsl_A_inv = (1 / rtsl_A);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - (radius * radius));
+ float rtsl_invariant0 = (radius * radius);
+ // End loop invariant setup
+
+ RayPacketData* data = rays.data;
+ int debugFlag = rays.getAllFlags()&RayPacket::DebugPacket;
+ if (debugFlag) {
+ cerr << "Sphere::intersect called" << endl;
+ }
+
+#ifdef MANTA_SSE
+ if((rays.rayBegin ^ (rays.rayEnd-1)) & ~3){
+ int i = rays.rayBegin & ~3;
+ // Prologue
+ if(i != rays.rayBegin){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_set1_ps(rtsl_C));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray0;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray0;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray0;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ i+=4;
+ }
+ prologue_next_ray0:;
+ // Primary loop body
+ int e = rays.rayEnd - 3;
+ for(;i<e;i+=4){
+ {
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0xffffffff));
+ int intmask = 15;
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but
not read yet
+
+ // Section 0: END Declaring variables that were written but
not read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_set1_ps(rtsl_C));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ =
_mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined, rtsl_disc_),
_mm_and_ps(ifmask3_section_undefined, _mm_sub_ps(_mm_mul_ps(rtsl_B_,
rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end1;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end1;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end1;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ for_loop_end1:;
+ }
+ // Epilogue
+ if(i != rays.rayEnd){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_set1_ps(rtsl_C));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray2;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray2;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray2;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ epilogue_next_ray2:;
+ } else {
+ // Single SSE vector
+ int i = rays.rayBegin & ~3;
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ __m128 mask =
_mm_castsi128_ps(_mm_and_si128(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)), _mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i))));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 4
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 4, scope for O = -1
+ // Current scope = 4, scope for D = 4
+ // scope for B: 4
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for B = 4
+ // scope for disc: 4
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_set1_ps(rtsl_C));
+ // Current scope = 4, scope for disc = 4
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 5, scope for B = 4
+ // scope for t: 5
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 5, scope for D = 4
+ // Current scope = 5, scope for t = 5
+ // Current scope = 5, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 4
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for O = -1
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for C = 5
+ // Current scope = 5, scope for disc = 4
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 5, scope for disc = 4
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 6, scope for disc = 4
+ // scope for r: 6
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 6, scope for t = 5
+ // Current scope = 6, scope for r = 6
+ // Current scope = 6, scope for B = 5
+ // scope for t1: 6
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 6, scope for t1 = 6
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 7, scope for r = 6
+ // Current scope = 7, scope for B = 5
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ = _mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined,
rtsl_t_), _mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 7, scope for t1 = 6
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray3;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 6, scope for t = 5
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray3;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray3;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ single_vector_next_ray3:;
+#else
+ for(int i = rays.begin(); i < rays.end(); i++){
+ {
+ // Section 0: BEGIN Loading variables from previous sections
+
+ Vector rtsl_O(s_O_x[i], s_O_y[i], s_O_z[i]);
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // No declaration for O
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ Vector rtsl_D = rays.getDirection(i);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_disc = ((rtsl_B * rtsl_B) - rtsl_C);
+ if((rtsl_disc >= 0.f)){
+ float rtsl_t = ((-(rtsl_B)) * rtsl_A_inv);
+ rtsl_O += (rtsl_D * rtsl_t);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant0);
+ rtsl_disc = ((rtsl_B * rtsl_B) - rtsl_C);
+ if((rtsl_disc >= 0.f)){
+ float rtsl_r = Sqrt(rtsl_disc);
+ float rtsl_t1 = (rtsl_t + ((-((rtsl_r + rtsl_B))) *
rtsl_A_inv));
+ if((rtsl_t1 <= T_EPSILON)){
+ rtsl_t += ((rtsl_r - rtsl_B) * rtsl_A_inv);
+ } else {
+ rtsl_t = rtsl_t1;
+ }
+ rays.hit(i, rtsl_t, getMaterial(), this,
getTexCoordMapper());
+ }
+ }
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ }
+#endif
+ } else {
+ // !RayPacket::ConstantOrigin
+ // Special case: unit directions
+
+ // Begin loop invariant setup code
+ float rtsl_A = 1.f;
+ float rtsl_A_inv = (1 / rtsl_A);
+ float rtsl_invariant0 = (radius * radius);
+ float rtsl_invariant1 = (radius * radius);
+ // End loop invariant setup
+
+ RayPacketData* data = rays.data;
+ int debugFlag = rays.getAllFlags()&RayPacket::DebugPacket;
+ if (debugFlag) {
+ cerr << "Sphere::intersect called" << endl;
+ }
+
+#ifdef MANTA_SSE
+ if((rays.rayBegin ^ (rays.rayEnd-1)) & ~3){
+ int i = rays.rayBegin & ~3;
+ // Prologue
+ if(i != rays.rayBegin){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
rtsl_C_);
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray5;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray5;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray5;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ i+=4;
+ }
+ prologue_next_ray5:;
+ // Primary loop body
+ int e = rays.rayEnd - 3;
+ for(;i<e;i+=4){
+ {
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0xffffffff));
+ int intmask = 15;
+ // Section 0: BEGIN Declaring variables that were written but
not read yet
+
+ // Section 0: END Declaring variables that were written but
not read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
rtsl_C_);
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ =
_mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined, rtsl_disc_),
_mm_and_ps(ifmask3_section_undefined, _mm_sub_ps(_mm_mul_ps(rtsl_B_,
rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end6;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end6;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end6;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ for_loop_end6:;
+ }
+ // Epilogue
+ if(i != rays.rayEnd){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
rtsl_C_);
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray7;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray7;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray7;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ epilogue_next_ray7:;
+ } else {
+ // Single SSE vector
+ int i = rays.rayBegin & ~3;
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ __m128 mask =
_mm_castsi128_ps(_mm_and_si128(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)), _mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i))));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 4
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 4
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 4, scope for O = 4
+ // Current scope = 4, scope for D = 4
+ // scope for B: 4
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 4, scope for O = 4
+ // Current scope = 4, scope for O = 4
+ // scope for C: 4
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for C = 4
+ // scope for disc: 4
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
rtsl_C_);
+ // Current scope = 4, scope for disc = 4
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 5, scope for B = 4
+ // scope for t: 5
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv));
+ // Current scope = 5, scope for D = 4
+ // Current scope = 5, scope for t = 5
+ // Current scope = 5, scope for O = 4
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 5, scope for O = 4
+ // Current scope = 5, scope for D = 4
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 4
+ // Current scope = 5, scope for O = 4
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for C = 5
+ // Current scope = 5, scope for disc = 4
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), rtsl_C_)));
+ // Current scope = 5, scope for disc = 4
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 6, scope for disc = 4
+ // scope for r: 6
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 6, scope for t = 5
+ // Current scope = 6, scope for r = 6
+ // Current scope = 6, scope for B = 5
+ // scope for t1: 6
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), _mm_set1_ps(rtsl_A_inv)));
+ // Current scope = 6, scope for t1 = 6
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 7, scope for r = 6
+ // Current scope = 7, scope for B = 5
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ = _mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined,
rtsl_t_), _mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), _mm_set1_ps(rtsl_A_inv)))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 7, scope for t1 = 6
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray8;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 6, scope for t = 5
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray8;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray8;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ single_vector_next_ray8:;
+#else
+ for(int i = rays.begin(); i < rays.end(); i++){
+ {
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ Vector rtsl_O = (rays.getOrigin(i) - center);
+ Vector rtsl_D = rays.getDirection(i);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant0);
+ float rtsl_disc = ((rtsl_B * rtsl_B) - rtsl_C);
+ if((rtsl_disc >= 0.f)){
+ float rtsl_t = ((-(rtsl_B)) * rtsl_A_inv);
+ rtsl_O += (rtsl_D * rtsl_t);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant1);
+ rtsl_disc = ((rtsl_B * rtsl_B) - rtsl_C);
+ if((rtsl_disc >= 0.f)){
+ float rtsl_r = Sqrt(rtsl_disc);
+ float rtsl_t1 = (rtsl_t + ((-((rtsl_r + rtsl_B))) *
rtsl_A_inv));
+ if((rtsl_t1 <= T_EPSILON)){
+ rtsl_t += ((rtsl_r - rtsl_B) * rtsl_A_inv);
+ } else {
+ rtsl_t = rtsl_t1;
+ }
+ rays.hit(i, rtsl_t, getMaterial(), this,
getTexCoordMapper());
+ }
+ }
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ }
+#endif
+ }
+ } else {
+ // !RayPacket::NormalizedDirections
+ if(rays.getAllFlags() & RayPacket::ConstantOrigin){
+ // RayPacket::ConstantOrigin
+ // Special case: constant origin
+
+ // Begin loop invariant setup code
+ Vector rtsl_O = (rays.getOrigin(0) - center);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - (radius * radius));
+ float rtsl_invariant0 = (radius * radius);
+ // End loop invariant setup
+
+ RayPacketData* data = rays.data;
+ int debugFlag = rays.getAllFlags()&RayPacket::DebugPacket;
+ if (debugFlag) {
+ cerr << "Sphere::intersect called" << endl;
+ }
+
+#ifdef MANTA_SSE
+ if((rays.rayBegin ^ (rays.rayEnd-1)) & ~3){
+ int i = rays.rayBegin & ~3;
+ // Prologue
+ if(i != rays.rayBegin){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, _mm_set1_ps(rtsl_C)));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray10;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray10;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray10;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ i+=4;
+ }
+ prologue_next_ray10:;
+ // Primary loop body
+ int e = rays.rayEnd - 3;
+ for(;i<e;i+=4){
+ {
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0xffffffff));
+ int intmask = 15;
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but
not read yet
+
+ // Section 0: END Declaring variables that were written but
not read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, _mm_set1_ps(rtsl_C)));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ =
_mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined, rtsl_disc_),
_mm_and_ps(ifmask3_section_undefined, _mm_sub_ps(_mm_mul_ps(rtsl_B_,
rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end11;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end11;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end11;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ for_loop_end11:;
+ }
+ // Epilogue
+ if(i != rays.rayEnd){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, _mm_set1_ps(rtsl_C)));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = -1
+ // Current scope = 6, scope for O = -1
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray12;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray12;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray12;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ epilogue_next_ray12:;
+ } else {
+ // Single SSE vector
+ int i = rays.rayBegin & ~3;
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ __m128 mask =
_mm_castsi128_ps(_mm_and_si128(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)), _mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i))));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Loading variables from previous sections
+
+ __m128 rtsl_O_x = _mm_set1_ps(rtsl_O.x());
+ __m128 rtsl_O_y = _mm_set1_ps(rtsl_O.y());
+ __m128 rtsl_O_z = _mm_set1_ps(rtsl_O.z());
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for D: 4
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 4, scope for D = 4
+ // Current scope = 4, scope for D = 4
+ // scope for A: 4
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 4, scope for A = 4
+ // scope for A_inv: 4
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 4, scope for O = -1
+ // Current scope = 4, scope for D = 4
+ // scope for B: 4
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for A = 4
+ // scope for disc: 4
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, _mm_set1_ps(rtsl_C)));
+ // Current scope = 4, scope for disc = 4
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 5, scope for B = 4
+ // Current scope = 5, scope for A_inv = 4
+ // scope for t: 5
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 5, scope for D = 4
+ // Current scope = 5, scope for t = 5
+ // Current scope = 5, scope for O = -1
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for D = 4
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = -1
+ // Current scope = 5, scope for O = -1
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 4
+ // Current scope = 5, scope for C = 5
+ // Current scope = 5, scope for disc = 4
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 5, scope for disc = 4
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 6, scope for disc = 4
+ // scope for r: 6
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 6, scope for t = 5
+ // Current scope = 6, scope for r = 6
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 4
+ // scope for t1: 6
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 6, scope for t1 = 6
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 7, scope for r = 6
+ // Current scope = 7, scope for B = 5
+ // Current scope = 7, scope for A_inv = 4
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ = _mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined,
rtsl_t_), _mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 7, scope for t1 = 6
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray13;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 6, scope for t = 5
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray13;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray13;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ single_vector_next_ray13:;
+#else
+ for(int i = rays.begin(); i < rays.end(); i++){
+ {
+ // Section 0: BEGIN Loading variables from previous sections
+
+ Vector rtsl_O(s_O_x[i], s_O_y[i], s_O_z[i]);
+ // Section 0: END Loading variables from previous sections
+
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // No declaration for O
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ Vector rtsl_D = rays.getDirection(i);
+ float rtsl_A = Dot(rtsl_D, rtsl_D);
+ float rtsl_A_inv = (1 / rtsl_A);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_disc = ((rtsl_B * rtsl_B) - (rtsl_A * rtsl_C));
+ if((rtsl_disc >= 0.f)){
+ float rtsl_t = ((-(rtsl_B)) * rtsl_A_inv);
+ rtsl_O += (rtsl_D * rtsl_t);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant0);
+ rtsl_disc = ((rtsl_B * rtsl_B) - (rtsl_A * rtsl_C));
+ if((rtsl_disc >= 0.f)){
+ float rtsl_r = Sqrt(rtsl_disc);
+ float rtsl_t1 = (rtsl_t + ((-((rtsl_r + rtsl_B))) *
rtsl_A_inv));
+ if((rtsl_t1 <= T_EPSILON)){
+ rtsl_t += ((rtsl_r - rtsl_B) * rtsl_A_inv);
+ } else {
+ rtsl_t = rtsl_t1;
+ }
+ rays.hit(i, rtsl_t, getMaterial(), this,
getTexCoordMapper());
+ }
+ }
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ }
+#endif
+ } else {
+ // !RayPacket::ConstantOrigin
+ // General case:
+
+ // Begin loop invariant setup code
+ float rtsl_invariant0 = (radius * radius);
+ float rtsl_invariant1 = (radius * radius);
+ // End loop invariant setup
+
+ RayPacketData* data = rays.data;
+ int debugFlag = rays.getAllFlags()&RayPacket::DebugPacket;
+ if (debugFlag) {
+ cerr << "Sphere::intersect called" << endl;
+ }
+
+#ifdef MANTA_SSE
+ if((rays.rayBegin ^ (rays.rayEnd-1)) & ~3){
+ int i = rays.rayBegin & ~3;
+ // Prologue
+ if(i != rays.rayBegin){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, rtsl_C_));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray15;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray15;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto prologue_next_ray15;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ i+=4;
+ }
+ prologue_next_ray15:;
+ // Primary loop body
+ int e = rays.rayEnd - 3;
+ for(;i<e;i+=4){
+ {
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0xffffffff));
+ int intmask = 15;
+ // Section 0: BEGIN Declaring variables that were written but
not read yet
+
+ // Section 0: END Declaring variables that were written but
not read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, rtsl_C_));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ =
_mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined, rtsl_disc_),
_mm_and_ps(ifmask3_section_undefined, _mm_sub_ps(_mm_mul_ps(rtsl_B_,
rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end16;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end16;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto for_loop_end16;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ for_loop_end16:;
+ }
+ // Epilogue
+ if(i != rays.rayEnd){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i)));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 5
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 5
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 5, scope for D = 5
+ // Current scope = 5, scope for D = 5
+ // scope for A: 5
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 5, scope for A = 5
+ // scope for A_inv: 5
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for D = 5
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 5
+ // Current scope = 5, scope for O = 5
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 5
+ // Current scope = 5, scope for C = 5
+ // scope for disc: 5
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, rtsl_C_));
+ // Current scope = 5, scope for disc = 5
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 5
+ // scope for t: 6
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 6, scope for D = 5
+ // Current scope = 6, scope for t = 6
+ // Current scope = 6, scope for O = 5
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for D = 5
+ // scope for B: 6
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 6, scope for O = 5
+ // Current scope = 6, scope for O = 5
+ // scope for C: 6
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for B = 6
+ // Current scope = 6, scope for A = 5
+ // Current scope = 6, scope for C = 6
+ // Current scope = 6, scope for disc = 5
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 6, scope for disc = 5
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 7, scope for disc = 5
+ // scope for r: 7
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 7, scope for t = 6
+ // Current scope = 7, scope for r = 7
+ // Current scope = 7, scope for B = 6
+ // Current scope = 7, scope for A_inv = 5
+ // scope for t1: 7
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 7, scope for t1 = 7
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 8, scope for r = 7
+ // Current scope = 8, scope for B = 6
+ // Current scope = 8, scope for A_inv = 5
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined, rtsl_t_),
_mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 8, scope for t1 = 7
+ // Current scope = 8, scope for t = 6
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray17;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 7, scope for t = 6
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray17;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto epilogue_next_ray17;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ epilogue_next_ray17:;
+ } else {
+ // Single SSE vector
+ int i = rays.rayBegin & ~3;
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ __m128 mask =
_mm_castsi128_ps(_mm_and_si128(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)), _mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i))));
+ int intmask = _mm_movemask_ps(mask);
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ // scope for O: 4
+ __m128 rtsl_O_x =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[0][i])),
_mm_set1_ps(center.x()));
+ __m128 rtsl_O_y =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[1][i])),
_mm_set1_ps(center.y()));
+ __m128 rtsl_O_z =
_mm_sub_ps(_mm_load_ps((float*)&(data->origin[2][i])),
_mm_set1_ps(center.z()));
+ // scope for D: 4
+ __m128 rtsl_D_x = _mm_load_ps((float*)&(data->direction[0][i]));
+ __m128 rtsl_D_y = _mm_load_ps((float*)&(data->direction[1][i]));
+ __m128 rtsl_D_z = _mm_load_ps((float*)&(data->direction[2][i]));
+ // Current scope = 4, scope for D = 4
+ // Current scope = 4, scope for D = 4
+ // scope for A: 4
+ __m128 rtsl_A_ = _mm_add_ps(_mm_mul_ps(rtsl_D_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_D_y, rtsl_D_y), _mm_mul_ps(rtsl_D_z, rtsl_D_z)));
+ // Current scope = 4, scope for A = 4
+ // scope for A_inv: 4
+ __m128 rtsl_A_inv_ = _mm_div_ps(_mm_set1_ps(1), rtsl_A_);
+ // Current scope = 4, scope for O = 4
+ // Current scope = 4, scope for D = 4
+ // scope for B: 4
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 4, scope for O = 4
+ // Current scope = 4, scope for O = 4
+ // scope for C: 4
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant0));
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for B = 4
+ // Current scope = 4, scope for A = 4
+ // Current scope = 4, scope for C = 4
+ // scope for disc: 4
+ __m128 rtsl_disc_ = _mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_),
_mm_mul_ps(rtsl_A_, rtsl_C_));
+ // Current scope = 4, scope for disc = 4
+ __m128 condmask1_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask2_section_undefined =
_mm_movemask_ps(condmask1_section_undefined);
+ __m128 ifmask3_section_undefined = _mm_and_ps(mask,
condmask1_section_undefined);
+ int ifintmask4_section_undefined =
_mm_movemask_ps(ifmask3_section_undefined);
+ if(ifintmask4_section_undefined != 0){
+ // Current scope = 5, scope for B = 4
+ // Current scope = 5, scope for A_inv = 4
+ // scope for t: 5
+ __m128 rtsl_t_ = _mm_mul_ps(_mm_xor_ps(rtsl_B_,
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_);
+ // Current scope = 5, scope for D = 4
+ // Current scope = 5, scope for t = 5
+ // Current scope = 5, scope for O = 4
+ rtsl_O_x = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_x), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_x,
_mm_mul_ps(rtsl_D_x, rtsl_t_))));
+ rtsl_O_y = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_y), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_y,
_mm_mul_ps(rtsl_D_y, rtsl_t_))));
+ rtsl_O_z = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_O_z), _mm_and_ps(ifmask3_section_undefined, _mm_add_ps(rtsl_O_z,
_mm_mul_ps(rtsl_D_z, rtsl_t_))));
+ // Current scope = 5, scope for O = 4
+ // Current scope = 5, scope for D = 4
+ // scope for B: 5
+ __m128 rtsl_B_ = _mm_add_ps(_mm_mul_ps(rtsl_O_x, rtsl_D_x),
_mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_D_y), _mm_mul_ps(rtsl_O_z, rtsl_D_z)));
+ // Current scope = 5, scope for O = 4
+ // Current scope = 5, scope for O = 4
+ // scope for C: 5
+ __m128 rtsl_C_ = _mm_sub_ps(_mm_add_ps(_mm_mul_ps(rtsl_O_x,
rtsl_O_x), _mm_add_ps(_mm_mul_ps(rtsl_O_y, rtsl_O_y), _mm_mul_ps(rtsl_O_z,
rtsl_O_z))), _mm_set1_ps(rtsl_invariant1));
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for B = 5
+ // Current scope = 5, scope for A = 4
+ // Current scope = 5, scope for C = 5
+ // Current scope = 5, scope for disc = 4
+ rtsl_disc_ = _mm_or_ps(_mm_andnot_ps(ifmask3_section_undefined,
rtsl_disc_), _mm_and_ps(ifmask3_section_undefined,
_mm_sub_ps(_mm_mul_ps(rtsl_B_, rtsl_B_), _mm_mul_ps(rtsl_A_, rtsl_C_))));
+ // Current scope = 5, scope for disc = 4
+ __m128 condmask5_section_undefined = _mm_cmpge_ps(rtsl_disc_,
_mm_set1_ps(0));
+ //int condintmask6_section_undefined =
_mm_movemask_ps(condmask5_section_undefined);
+ __m128 ifmask7_section_undefined =
_mm_and_ps(ifmask3_section_undefined, condmask5_section_undefined);
+ int ifintmask8_section_undefined =
_mm_movemask_ps(ifmask7_section_undefined);
+ if(ifintmask8_section_undefined != 0){
+ // Current scope = 6, scope for disc = 4
+ // scope for r: 6
+ __m128 rtsl_r_ = _mm_sqrt_ps(rtsl_disc_);
+ // Current scope = 6, scope for t = 5
+ // Current scope = 6, scope for r = 6
+ // Current scope = 6, scope for B = 5
+ // Current scope = 6, scope for A_inv = 4
+ // scope for t1: 6
+ __m128 rtsl_t1_ = _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_xor_ps(_mm_add_ps(rtsl_r_, rtsl_B_),
_mm_castsi128_ps(_mm_set1_epi32(0x80000000))), rtsl_A_inv_));
+ // Current scope = 6, scope for t1 = 6
+ __m128 condmask9_section_undefined = _mm_cmple_ps(rtsl_t1_,
_mm_set1_ps(T_EPSILON));
+ int condintmask10_section_undefined =
_mm_movemask_ps(condmask9_section_undefined);
+ __m128 ifmask11_section_undefined =
_mm_and_ps(ifmask7_section_undefined, condmask9_section_undefined);
+ int ifintmask12_section_undefined =
_mm_movemask_ps(ifmask11_section_undefined);
+ if(ifintmask12_section_undefined != 0){
+ // Current scope = 7, scope for r = 6
+ // Current scope = 7, scope for B = 5
+ // Current scope = 7, scope for A_inv = 4
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ = _mm_or_ps(_mm_andnot_ps(ifmask11_section_undefined,
rtsl_t_), _mm_and_ps(ifmask11_section_undefined, _mm_add_ps(rtsl_t_,
_mm_mul_ps(_mm_sub_ps(rtsl_r_, rtsl_B_), rtsl_A_inv_))));
+ }
+ __m128 elsemask13_section_undefined =
_mm_andnot_ps(condmask9_section_undefined, ifmask7_section_undefined);
+ int elseintmask14_section_undefined =
~condintmask10_section_undefined&ifintmask8_section_undefined;
+ if(elseintmask14_section_undefined != 0){
+ // Current scope = 7, scope for t1 = 6
+ // Current scope = 7, scope for t = 5
+ rtsl_t_ =
_mm_or_ps(_mm_andnot_ps(elsemask13_section_undefined, rtsl_t_),
_mm_and_ps(elsemask13_section_undefined, rtsl_t1_));
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray18;
+ }
+ // Update the current mask
+ ifmask7_section_undefined = _mm_and_ps(mask,
ifmask7_section_undefined);
+ ifintmask8_section_undefined = intmask &
ifintmask8_section_undefined;
+ // Current scope = 6, scope for t = 5
+ rays.hitWithMask(i, ifmask7_section_undefined, rtsl_t_,
getMaterial(), this, getTexCoordMapper());
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray18;
+ }
+ // Update the current mask
+ ifmask3_section_undefined = _mm_and_ps(mask,
ifmask3_section_undefined);
+ ifintmask4_section_undefined = intmask &
ifintmask4_section_undefined;
+ }
+ // Early termination
+ if (intmask == 0) {
+ goto single_vector_next_ray18;
+ }
+ // Update the current mask
+ mask = _mm_and_ps(mask, mask);
+ intmask = intmask & intmask;
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ single_vector_next_ray18:;
+#else
+ for(int i = rays.begin(); i < rays.end(); i++){
+ {
+ // Section 0: BEGIN Declaring variables that were written but not
read yet
+
+ // Section 0: END Declaring variables that were written but not
read yet
+
+ Vector rtsl_O = (rays.getOrigin(i) - center);
+ Vector rtsl_D = rays.getDirection(i);
+ float rtsl_A = Dot(rtsl_D, rtsl_D);
+ float rtsl_A_inv = (1 / rtsl_A);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant0);
+ float rtsl_disc = ((rtsl_B * rtsl_B) - (rtsl_A * rtsl_C));
+ if((rtsl_disc >= 0.f)){
+ float rtsl_t = ((-(rtsl_B)) * rtsl_A_inv);
+ rtsl_O += (rtsl_D * rtsl_t);
+ float rtsl_B = Dot(rtsl_O, rtsl_D);
+ float rtsl_C = (Dot(rtsl_O, rtsl_O) - rtsl_invariant1);
+ rtsl_disc = ((rtsl_B * rtsl_B) - (rtsl_A * rtsl_C));
+ if((rtsl_disc >= 0.f)){
+ float rtsl_r = Sqrt(rtsl_disc);
+ float rtsl_t1 = (rtsl_t + ((-((rtsl_r + rtsl_B))) *
rtsl_A_inv));
+ if((rtsl_t1 <= T_EPSILON)){
+ rtsl_t += ((rtsl_r - rtsl_B) * rtsl_A_inv);
+ } else {
+ rtsl_t = rtsl_t1;
+ }
+ rays.hit(i, rtsl_t, getMaterial(), this,
getTexCoordMapper());
+ }
+ }
+ // Section 0: BEGIN Storing variables people will need later.
+
+ // Section 0: END Storing variables people will need later.
+
+ }
+ }
+#endif
+ }
+ }
+}
+
+void Sphere::computeNormal(const RenderContext& context, RayPacket& rays)
const
+{
+ rays.computeHitPositions();
+ // General case:
+ RayPacketData* data = rays.data;
+ int debugFlag = rays.getAllFlags()&RayPacket::DebugPacket;
+ if (debugFlag) {
+ cerr << "Sphere::computeNormal called" << endl;
+ }
+
+#ifdef MANTA_SSE
+ if((rays.rayBegin ^ (rays.rayEnd-1)) & ~3){
+ int i = rays.rayBegin & ~3;
+ // Prologue
+ if(i != rays.rayBegin){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)));
+ //int intmask = _mm_movemask_ps(mask);
+ _mm_store_ps((float*)&(data->normal[0][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[0][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[0][i])),
_mm_set1_ps(center.x())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[1][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[1][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[1][i])),
_mm_set1_ps(center.y())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[2][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[2][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[2][i])),
_mm_set1_ps(center.z())), _mm_set1_ps(inv_radius)))));
+ i+=4;
+ }
+ // Primary loop body
+ int e = rays.rayEnd - 3;
+ for(;i<e;i+=4){
+ {
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_set1_epi32(0xffffffff));
+ //int intmask = 15;
+ _mm_store_ps((float*)&(data->normal[0][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[0][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[0][i])),
_mm_set1_ps(center.x())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[1][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[1][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[1][i])),
_mm_set1_ps(center.y())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[2][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[2][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[2][i])),
_mm_set1_ps(center.z())), _mm_set1_ps(inv_radius)))));
+ }
+ }
+ // Epilogue
+ if(i != rays.rayEnd){
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ // mask is on for active rays
+ __m128 mask = _mm_castsi128_ps(_mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i)));
+ //int intmask = _mm_movemask_ps(mask);
+ _mm_store_ps((float*)&(data->normal[0][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[0][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[0][i])),
_mm_set1_ps(center.x())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[1][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[1][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[1][i])),
_mm_set1_ps(center.y())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[2][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[2][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[2][i])),
_mm_set1_ps(center.z())), _mm_set1_ps(inv_radius)))));
+ }
+ } else {
+ // Single SSE vector
+ int i = rays.rayBegin & ~3;
+ __m128i ray_idx = _mm_set_epi32(3, 2, 1, 0);
+ __m128 mask = _mm_castsi128_ps(_mm_and_si128(_mm_cmpgt_epi32(ray_idx,
_mm_set1_epi32(rays.rayBegin-i-1)), _mm_cmplt_epi32(ray_idx,
_mm_set1_epi32(rays.rayEnd-i))));
+ //int intmask = _mm_movemask_ps(mask);
+ _mm_store_ps((float*)&(data->normal[0][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[0][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[0][i])),
_mm_set1_ps(center.x())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[1][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[1][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[1][i])),
_mm_set1_ps(center.y())), _mm_set1_ps(inv_radius)))));
+ _mm_store_ps((float*)&(data->normal[2][i]),
_mm_or_ps(_mm_andnot_ps(mask, _mm_load_ps((float*)&(data->normal[2][i]))),
_mm_and_ps(mask,
_mm_mul_ps(_mm_sub_ps(_mm_load_ps((float*)&(data->hitPosition[2][i])),
_mm_set1_ps(center.z())), _mm_set1_ps(inv_radius)))));
+ }
+#else
+ for(int i = rays.begin(); i < rays.end(); i++){
+ {
+ rays.setNormal(i, ((rays.getHitPosition(i) - center) * inv_radius));
+ }
+ }
+#endif
+}
- [Manta] r2060 - trunk/Model/Primitives, James Bigler, 02/13/2008
Archive powered by MHonArc 2.6.16.
