Оптимизация под GNU C компилятор / Optimization for GNU C Compiller
This commit is contained in:
parent
e4d75824f3
commit
081f794eb1
10 changed files with 671 additions and 755 deletions
|
|
@ -7,70 +7,6 @@ const BgFP32Versor BG_FP32_IDLE_VERSOR = { 1.0f, 0.0f, 0.0f, 0.0f };
|
|||
|
||||
const BgFP64Versor BG_FP64_IDLE_VERSOR = { 1.0, 0.0, 0.0, 0.0 };
|
||||
|
||||
// ==================== Set ===================== //
|
||||
|
||||
void bg_fp32_versor_set_values(const float s0, const float x1, const float x2, const float x3, BgFP32Versor* versor)
|
||||
{
|
||||
__BgFP32DarkTwinVersor* twin = (__BgFP32DarkTwinVersor*)versor;
|
||||
|
||||
twin->s0 = s0;
|
||||
twin->x1 = x1;
|
||||
twin->x2 = x2;
|
||||
twin->x3 = x3;
|
||||
|
||||
const float square_modulus = (s0 * s0 + x1 * x1) + (x2 * x2 + x3 * x3);
|
||||
|
||||
if (1.0f - BG_FP32_TWO_EPSYLON <= square_modulus && square_modulus <= 1.0f + BG_FP32_TWO_EPSYLON) {
|
||||
return;
|
||||
}
|
||||
|
||||
if (square_modulus <= BG_FP32_SQUARE_EPSYLON) {
|
||||
twin->s0 = 1.0f;
|
||||
twin->x1 = 0.0f;
|
||||
twin->x2 = 0.0f;
|
||||
twin->x3 = 0.0f;
|
||||
return;
|
||||
}
|
||||
|
||||
const float multiplier = sqrtf(1.0f / square_modulus);
|
||||
|
||||
twin->s0 *= multiplier;
|
||||
twin->x1 *= multiplier;
|
||||
twin->x2 *= multiplier;
|
||||
twin->x3 *= multiplier;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_set_values(const double s0, const double x1, const double x2, const double x3, BgFP64Versor* versor)
|
||||
{
|
||||
__BgFP64DarkTwinVersor* twin = (__BgFP64DarkTwinVersor*)versor;
|
||||
|
||||
twin->s0 = s0;
|
||||
twin->x1 = x1;
|
||||
twin->x2 = x2;
|
||||
twin->x3 = x3;
|
||||
|
||||
const double square_modulus = (s0 * s0 + x1 * x1) + (x2 * x2 + x3 * x3);
|
||||
|
||||
if (1.0 - BG_FP64_TWO_EPSYLON <= square_modulus && square_modulus <= 1.0 + BG_FP64_TWO_EPSYLON) {
|
||||
return;
|
||||
}
|
||||
|
||||
if (square_modulus <= BG_FP64_SQUARE_EPSYLON) {
|
||||
twin->s0 = 1.0;
|
||||
twin->x1 = 0.0;
|
||||
twin->x2 = 0.0;
|
||||
twin->x3 = 0.0;
|
||||
return;
|
||||
}
|
||||
|
||||
const double multiplier = sqrt(1.0 / square_modulus);
|
||||
|
||||
twin->s0 *= multiplier;
|
||||
twin->x1 *= multiplier;
|
||||
twin->x2 *= multiplier;
|
||||
twin->x3 *= multiplier;
|
||||
}
|
||||
|
||||
// =============== Set Crude Turn =============== //
|
||||
|
||||
void bg_fp32_versor_set_crude_turn(const float x1, const float x2, const float x3, const float angle, const angle_unit_t unit, BgFP32Versor* result)
|
||||
|
|
@ -164,258 +100,3 @@ void bg_fp64_versor_get_rotation(const BgFP64Versor* versor, BgFP64Rotation3* re
|
|||
result->axis.x2 = versor->x2 * multiplier;
|
||||
result->axis.x3 = versor->x3 * multiplier;
|
||||
}
|
||||
|
||||
// ================ Combination ================= //
|
||||
|
||||
void bg_fp32_versor_combine(const BgFP32Versor* second, const BgFP32Versor* first, BgFP32Versor* result)
|
||||
{
|
||||
const float s0 = (second->s0 * first->s0 - second->x1 * first->x1) - (second->x2 * first->x2 + second->x3 * first->x3);
|
||||
const float x1 = (second->x1 * first->s0 + second->s0 * first->x1) - (second->x3 * first->x2 - second->x2 * first->x3);
|
||||
const float x2 = (second->x2 * first->s0 + second->s0 * first->x2) - (second->x1 * first->x3 - second->x3 * first->x1);
|
||||
const float x3 = (second->x3 * first->s0 + second->s0 * first->x3) - (second->x2 * first->x1 - second->x1 * first->x2);
|
||||
|
||||
const float square_modulus = (s0 * s0 + x1 * x1) + (x2 * x2 + x3 * x3);
|
||||
|
||||
__BgFP32DarkTwinVersor* twin = (__BgFP32DarkTwinVersor*)result;
|
||||
|
||||
twin->s0 = s0;
|
||||
twin->x1 = x1;
|
||||
twin->x2 = x2;
|
||||
twin->x3 = x3;
|
||||
|
||||
if (1.0f - BG_FP32_TWO_EPSYLON <= square_modulus && square_modulus <= 1.0f + BG_FP32_TWO_EPSYLON) {
|
||||
return;
|
||||
}
|
||||
|
||||
if (square_modulus <= BG_FP32_SQUARE_EPSYLON) {
|
||||
twin->s0 = 1.0f;
|
||||
twin->x1 = 0.0f;
|
||||
twin->x2 = 0.0f;
|
||||
twin->x3 = 0.0f;
|
||||
return;
|
||||
}
|
||||
|
||||
const float multiplier = sqrtf(1.0f / square_modulus);
|
||||
|
||||
twin->s0 *= multiplier;
|
||||
twin->x1 *= multiplier;
|
||||
twin->x2 *= multiplier;
|
||||
twin->x3 *= multiplier;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_combine(const BgFP64Versor* second, const BgFP64Versor* first, BgFP64Versor* result)
|
||||
{
|
||||
const double s0 = (second->s0 * first->s0 - second->x1 * first->x1) - (second->x2 * first->x2 + second->x3 * first->x3);
|
||||
const double x1 = (second->x1 * first->s0 + second->s0 * first->x1) - (second->x3 * first->x2 - second->x2 * first->x3);
|
||||
const double x2 = (second->x2 * first->s0 + second->s0 * first->x2) - (second->x1 * first->x3 - second->x3 * first->x1);
|
||||
const double x3 = (second->x3 * first->s0 + second->s0 * first->x3) - (second->x2 * first->x1 - second->x1 * first->x2);
|
||||
|
||||
const double square_modulus = (s0 * s0 + x1 * x1) + (x2 * x2 + x3 * x3);
|
||||
|
||||
__BgFP64DarkTwinVersor* twin = (__BgFP64DarkTwinVersor*)result;
|
||||
|
||||
twin->s0 = s0;
|
||||
twin->x1 = x1;
|
||||
twin->x2 = x2;
|
||||
twin->x3 = x3;
|
||||
|
||||
if (1.0 - BG_FP64_TWO_EPSYLON <= square_modulus && square_modulus <= 1.0 + BG_FP64_TWO_EPSYLON) {
|
||||
return;
|
||||
}
|
||||
|
||||
if (square_modulus <= BG_FP64_SQUARE_EPSYLON) {
|
||||
twin->s0 = 1.0;
|
||||
twin->x1 = 0.0;
|
||||
twin->x2 = 0.0;
|
||||
twin->x3 = 0.0;
|
||||
return;
|
||||
}
|
||||
|
||||
const double multiplier = sqrt(1.0 / square_modulus);
|
||||
|
||||
twin->s0 *= multiplier;
|
||||
twin->x1 *= multiplier;
|
||||
twin->x2 *= multiplier;
|
||||
twin->x3 *= multiplier;
|
||||
}
|
||||
|
||||
// =========== Make Rotation Matrix3x3 ========== //
|
||||
|
||||
void bg_fp32_versor_get_rotation_matrix(const BgFP32Versor* versor, BgFP32Matrix3x3* matrix)
|
||||
{
|
||||
const float s0s0 = versor->s0 * versor->s0;
|
||||
const float x1x1 = versor->x1 * versor->x1;
|
||||
const float x2x2 = versor->x2 * versor->x2;
|
||||
const float x3x3 = versor->x3 * versor->x3;
|
||||
|
||||
const float s0x1 = 2.0f * versor->s0 * versor->x1;
|
||||
const float s0x2 = 2.0f * versor->s0 * versor->x2;
|
||||
const float s0x3 = 2.0f * versor->s0 * versor->x3;
|
||||
|
||||
const float x1x2 = 2.0f * versor->x1 * versor->x2;
|
||||
const float x1x3 = 2.0f * versor->x1 * versor->x3;
|
||||
const float x2x3 = 2.0f * versor->x2 * versor->x3;
|
||||
|
||||
matrix->r1c1 = (s0s0 + x1x1) - (x2x2 + x3x3);
|
||||
matrix->r2c2 = (s0s0 + x2x2) - (x1x1 + x3x3);
|
||||
matrix->r3c3 = (s0s0 + x3x3) - (x1x1 + x2x2);
|
||||
|
||||
matrix->r1c2 = x1x2 - s0x3;
|
||||
matrix->r2c3 = x2x3 - s0x1;
|
||||
matrix->r3c1 = x1x3 - s0x2;
|
||||
|
||||
matrix->r2c1 = x1x2 + s0x3;
|
||||
matrix->r3c2 = x2x3 + s0x1;
|
||||
matrix->r1c3 = x1x3 + s0x2;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_get_rotation_matrix(const BgFP64Versor* versor, BgFP64Matrix3x3* matrix)
|
||||
{
|
||||
const double s0s0 = versor->s0 * versor->s0;
|
||||
const double x1x1 = versor->x1 * versor->x1;
|
||||
const double x2x2 = versor->x2 * versor->x2;
|
||||
const double x3x3 = versor->x3 * versor->x3;
|
||||
|
||||
const double s0x1 = 2.0 * versor->s0 * versor->x1;
|
||||
const double s0x2 = 2.0 * versor->s0 * versor->x2;
|
||||
const double s0x3 = 2.0 * versor->s0 * versor->x3;
|
||||
|
||||
const double x1x2 = 2.0 * versor->x1 * versor->x2;
|
||||
const double x1x3 = 2.0 * versor->x1 * versor->x3;
|
||||
const double x2x3 = 2.0 * versor->x2 * versor->x3;
|
||||
|
||||
matrix->r1c1 = (s0s0 + x1x1) - (x2x2 + x3x3);
|
||||
matrix->r2c2 = (s0s0 + x2x2) - (x1x1 + x3x3);
|
||||
matrix->r3c3 = (s0s0 + x3x3) - (x1x1 + x2x2);
|
||||
|
||||
matrix->r1c2 = x1x2 - s0x3;
|
||||
matrix->r2c3 = x2x3 - s0x1;
|
||||
matrix->r3c1 = x1x3 - s0x2;
|
||||
|
||||
matrix->r2c1 = x1x2 + s0x3;
|
||||
matrix->r3c2 = x2x3 + s0x1;
|
||||
matrix->r1c3 = x1x3 + s0x2;
|
||||
}
|
||||
|
||||
// =========== Make Reverse Matrix3x3 =========== //
|
||||
|
||||
void bg_fp32_versor_get_reverse_matrix(const BgFP32Versor* versor, BgFP32Matrix3x3* matrix)
|
||||
{
|
||||
const float s0s0 = versor->s0 * versor->s0;
|
||||
const float x1x1 = versor->x1 * versor->x1;
|
||||
const float x2x2 = versor->x2 * versor->x2;
|
||||
const float x3x3 = versor->x3 * versor->x3;
|
||||
|
||||
const float s0x1 = 2.0f * versor->s0 * versor->x1;
|
||||
const float s0x2 = 2.0f * versor->s0 * versor->x2;
|
||||
const float s0x3 = 2.0f * versor->s0 * versor->x3;
|
||||
|
||||
const float x1x2 = 2.0f * versor->x1 * versor->x2;
|
||||
const float x1x3 = 2.0f * versor->x1 * versor->x3;
|
||||
const float x2x3 = 2.0f * versor->x2 * versor->x3;
|
||||
|
||||
matrix->r1c1 = (s0s0 + x1x1) - (x2x2 + x3x3);
|
||||
matrix->r2c2 = (s0s0 + x2x2) - (x1x1 + x3x3);
|
||||
matrix->r3c3 = (s0s0 + x3x3) - (x1x1 + x2x2);
|
||||
|
||||
matrix->r1c2 = x1x2 + s0x3;
|
||||
matrix->r2c3 = x2x3 + s0x1;
|
||||
matrix->r3c1 = x1x3 + s0x2;
|
||||
|
||||
matrix->r2c1 = x1x2 - s0x3;
|
||||
matrix->r3c2 = x2x3 - s0x1;
|
||||
matrix->r1c3 = x1x3 - s0x2;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_get_reverse_matrix(const BgFP64Versor* versor, BgFP64Matrix3x3* matrix)
|
||||
{
|
||||
const double s0s0 = versor->s0 * versor->s0;
|
||||
const double x1x1 = versor->x1 * versor->x1;
|
||||
const double x2x2 = versor->x2 * versor->x2;
|
||||
const double x3x3 = versor->x3 * versor->x3;
|
||||
|
||||
const double s0x1 = 2.0 * versor->s0 * versor->x1;
|
||||
const double s0x2 = 2.0 * versor->s0 * versor->x2;
|
||||
const double s0x3 = 2.0 * versor->s0 * versor->x3;
|
||||
|
||||
const double x1x2 = 2.0 * versor->x1 * versor->x2;
|
||||
const double x1x3 = 2.0 * versor->x1 * versor->x3;
|
||||
const double x2x3 = 2.0 * versor->x2 * versor->x3;
|
||||
|
||||
matrix->r1c1 = (s0s0 + x1x1) - (x2x2 + x3x3);
|
||||
matrix->r2c2 = (s0s0 + x2x2) - (x1x1 + x3x3);
|
||||
matrix->r3c3 = (s0s0 + x3x3) - (x1x1 + x2x2);
|
||||
|
||||
matrix->r1c2 = x1x2 + s0x3;
|
||||
matrix->r2c3 = x2x3 + s0x1;
|
||||
matrix->r3c1 = x1x3 + s0x2;
|
||||
|
||||
matrix->r2c1 = x1x2 - s0x3;
|
||||
matrix->r3c2 = x2x3 - s0x1;
|
||||
matrix->r1c3 = x1x3 - s0x2;
|
||||
}
|
||||
|
||||
// ================ Turn Vector ================= //
|
||||
|
||||
void bg_fp32_versor_turn(const BgFP32Versor* versor, const BgFP32Vector3* vector, BgFP32Vector3* result)
|
||||
{
|
||||
const float tx1 = 2.0f * (versor->x2 * vector->x3 - versor->x3 * vector->x2);
|
||||
const float tx2 = 2.0f * (versor->x3 * vector->x1 - versor->x1 * vector->x3);
|
||||
const float tx3 = 2.0f * (versor->x1 * vector->x2 - versor->x2 * vector->x1);
|
||||
|
||||
const float x1 = (vector->x1 + tx1 * versor->s0) + (versor->x2 * tx3 - versor->x3 * tx2);
|
||||
const float x2 = (vector->x2 + tx2 * versor->s0) + (versor->x3 * tx1 - versor->x1 * tx3);
|
||||
const float x3 = (vector->x3 + tx3 * versor->s0) + (versor->x1 * tx2 - versor->x2 * tx1);
|
||||
|
||||
result->x1 = x1;
|
||||
result->x2 = x2;
|
||||
result->x3 = x3;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_turn(const BgFP64Versor* versor, const BgFP64Vector3* vector, BgFP64Vector3* result)
|
||||
{
|
||||
const double tx1 = 2.0 * (versor->x2 * vector->x3 - versor->x3 * vector->x2);
|
||||
const double tx2 = 2.0 * (versor->x3 * vector->x1 - versor->x1 * vector->x3);
|
||||
const double tx3 = 2.0 * (versor->x1 * vector->x2 - versor->x2 * vector->x1);
|
||||
|
||||
const double x1 = (vector->x1 + tx1 * versor->s0) + (versor->x2 * tx3 - versor->x3 * tx2);
|
||||
const double x2 = (vector->x2 + tx2 * versor->s0) + (versor->x3 * tx1 - versor->x1 * tx3);
|
||||
const double x3 = (vector->x3 + tx3 * versor->s0) + (versor->x1 * tx2 - versor->x2 * tx1);
|
||||
|
||||
result->x1 = x1;
|
||||
result->x2 = x2;
|
||||
result->x3 = x3;
|
||||
}
|
||||
|
||||
// ============== Turn Vector Back ============== //
|
||||
|
||||
void bg_fp32_versor_turn_back(const BgFP32Versor* versor, const BgFP32Vector3* vector, BgFP32Vector3* result)
|
||||
{
|
||||
const float tx1 = 2.0f * (versor->x2 * vector->x3 - versor->x3 * vector->x2);
|
||||
const float tx2 = 2.0f * (versor->x3 * vector->x1 - versor->x1 * vector->x3);
|
||||
const float tx3 = 2.0f * (versor->x1 * vector->x2 - versor->x2 * vector->x1);
|
||||
|
||||
const float x1 = (vector->x1 - tx1 * versor->s0) + (versor->x2 * tx3 - versor->x3 * tx2);
|
||||
const float x2 = (vector->x2 - tx2 * versor->s0) + (versor->x3 * tx1 - versor->x1 * tx3);
|
||||
const float x3 = (vector->x3 - tx3 * versor->s0) + (versor->x1 * tx2 - versor->x2 * tx1);
|
||||
|
||||
result->x1 = x1;
|
||||
result->x2 = x2;
|
||||
result->x3 = x3;
|
||||
}
|
||||
|
||||
void bg_fp64_versor_turn_back(const BgFP64Versor* versor, const BgFP64Vector3* vector, BgFP64Vector3* result)
|
||||
{
|
||||
const double tx1 = 2.0 * (versor->x2 * vector->x3 - versor->x3 * vector->x2);
|
||||
const double tx2 = 2.0 * (versor->x3 * vector->x1 - versor->x1 * vector->x3);
|
||||
const double tx3 = 2.0 * (versor->x1 * vector->x2 - versor->x2 * vector->x1);
|
||||
|
||||
const double x1 = (vector->x1 - tx1 * versor->s0) + (versor->x2 * tx3 - versor->x3 * tx2);
|
||||
const double x2 = (vector->x2 - tx2 * versor->s0) + (versor->x3 * tx1 - versor->x1 * tx3);
|
||||
const double x3 = (vector->x3 - tx3 * versor->s0) + (versor->x1 * tx2 - versor->x2 * tx1);
|
||||
|
||||
result->x1 = x1;
|
||||
result->x2 = x2;
|
||||
result->x3 = x3;
|
||||
}
|
||||
|
||||
|
|
|
|||
Loading…
Add table
Add a link
Reference in a new issue