mirror/glm
1
0
Fork 0
mirror of https://github.com/g-truc/glm.git synced 2024-11-10 12:41:54 +00:00
Code Issues Packages Projects Releases Wiki Activity

Generate SSE instructions for sub, mul and div including lowp div

Browse Source
This commit is contained in:
Christophe Riccio 2016-05-23 00:34:59 +02:00
parent 0e780a5efd
commit ff74b87b48
2 changed files with 121 additions and 233 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

310
glm/detail/type_vec4.inl
View File

@ -41,6 +41,33 @@ namespace detail
return tvec4<T, P>(a.x + b.x, a.y + b.y, a.z + b.z, a.w + b.w);
}
};
template <typename T, precision P>
struct compute_vec4_sub
{
static tvec4<T, P> call(tvec4<T, P> const & a, tvec4<T, P> const & b)
{
return tvec4<T, P>(a.x - b.x, a.y - b.y, a.z - b.z, a.w - b.w);
}
};
template <typename T, precision P>
struct compute_vec4_mul
{
static tvec4<T, P> call(tvec4<T, P> const & a, tvec4<T, P> const & b)
{
return tvec4<T, P>(a.x * b.x, a.y * b.y, a.z * b.z, a.w * b.w);
}
};
template <typename T, precision P>
struct compute_vec4_div
{
static tvec4<T, P> call(tvec4<T, P> const & a, tvec4<T, P> const & b)
{
return tvec4<T, P>(a.x / b.x, a.y / b.y, a.z / b.z, a.w / b.w);
}
};
}//namespace detail
// -- Implicit basic constructors --
@ -264,10 +291,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator+=(U scalar)
{
this->x += static_cast<T>(scalar);
this->y += static_cast<T>(scalar);
this->z += static_cast<T>(scalar);
this->w += static_cast<T>(scalar);
*this = detail::compute_vec4_add<T, P>::call(*this, tvec4<T, P>(scalar));
return *this;
}
@ -275,11 +299,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator+=(tvec1<U, P> const & v)
{
T const scalar = static_cast<T>(v.x);
this->x += scalar;
this->y += scalar;
this->z += scalar;
this->w += scalar;
*this = detail::compute_vec4_add<T, P>::call(*this, tvec4<T, P>(v.x));
return *this;
}
@ -295,10 +315,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator-=(U scalar)
{
this->x -= static_cast<T>(scalar);
this->y -= static_cast<T>(scalar);
this->z -= static_cast<T>(scalar);
this->w -= static_cast<T>(scalar);
*this = detail::compute_vec4_sub<T, P>::call(*this, tvec4<T, P>(scalar));
return *this;
}
@ -306,11 +323,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator-=(tvec1<U, P> const & v)
{
T const scalar = static_cast<T>(v.x);
this->x -= scalar;
this->y -= scalar;
this->z -= scalar;
this->w -= scalar;
*this = detail::compute_vec4_sub<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
@ -318,21 +331,15 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator-=(tvec4<U, P> const & v)
{
this->x -= static_cast<T>(v.x);
this->y -= static_cast<T>(v.y);
this->z -= static_cast<T>(v.z);
this->w -= static_cast<T>(v.w);
*this = detail::compute_vec4_sub<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
template <typename T, precision P>
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator*=(U v)
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator*=(U scalar)
{
this->x *= static_cast<T>(v);
this->y *= static_cast<T>(v);
this->z *= static_cast<T>(v);
this->w *= static_cast<T>(v);
*this = detail::compute_vec4_mul<T, P>::call(*this, tvec4<T, P>(scalar));
return *this;
}
@ -340,10 +347,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator*=(tvec1<U, P> const & v)
{
this->x *= static_cast<T>(v.x);
this->y *= static_cast<T>(v.x);
this->z *= static_cast<T>(v.x);
this->w *= static_cast<T>(v.x);
*this = detail::compute_vec4_mul<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
@ -351,21 +355,15 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator*=(tvec4<U, P> const & v)
{
this->x *= static_cast<T>(v.x);
this->y *= static_cast<T>(v.y);
this->z *= static_cast<T>(v.z);
this->w *= static_cast<T>(v.w);
*this = detail::compute_vec4_mul<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
template <typename T, precision P>
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator/=(U v)
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator/=(U scalar)
{
this->x /= static_cast<T>(v);
this->y /= static_cast<T>(v);
this->z /= static_cast<T>(v);
this->w /= static_cast<T>(v);
*this = detail::compute_vec4_div<T, P>::call(*this, tvec4<T, P>(scalar));
return *this;
}
@ -373,10 +371,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator/=(tvec1<U, P> const & v)
{
this->x /= static_cast<T>(v.x);
this->y /= static_cast<T>(v.x);
this->z /= static_cast<T>(v.x);
this->w /= static_cast<T>(v.x);
*this = detail::compute_vec4_div<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
@ -384,10 +379,7 @@ namespace detail
template <typename U>
GLM_FUNC_QUALIFIER tvec4<T, P> & tvec4<T, P>::operator/=(tvec4<U, P> const & v)
{
this->x /= static_cast<T>(v.x);
this->y /= static_cast<T>(v.y);
this->z /= static_cast<T>(v.z);
this->w /= static_cast<T>(v.w);
*this = detail::compute_vec4_div<T, P>::call(*this, tvec4<T, P>(v));
return *this;
}
@ -640,11 +632,7 @@ namespace detail
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator-(tvec4<T, P> const & v)
{
return tvec4<T, P>(
-v.x,
-v.y,
-v.z,
-v.w);
return tvec4<T, P>(0) -= v;
}
// -- Binary arithmetic operators --
@ -742,51 +730,31 @@ namespace detail
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator/(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x / scalar,
v.y / scalar,
v.z / scalar,
v.w / scalar);
return tvec4<T, P>(v) /= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator/(tvec4<T, P> const & v1, tvec1<T, P> const & v2)
{
return tvec4<T, P>(
v1.x / v2.x,
v1.y / v2.x,
v1.z / v2.x,
v1.w / v2.x);
return tvec4<T, P>(v1) /= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator/(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar / v.x,
scalar / v.y,
scalar / v.z,
scalar / v.w);
return tvec4<T, P>(scalar) /= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator/(tvec1<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x / v2.x,
v1.x / v2.y,
v1.x / v2.z,
v1.x / v2.w);
return tvec4<T, P>(v1) /= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator/(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x / v2.x,
v1.y / v2.y,
v1.z / v2.z,
v1.w / v2.w);
return tvec4<T, P>(v1) /= v2;
}
// -- Binary bit operators --
@ -794,311 +762,187 @@ namespace detail
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator%(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x % scalar,
v.y % scalar,
v.z % scalar,
v.w % scalar);
return tvec4<T, P>(v) %= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator%(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x % scalar.x,
v.y % scalar.x,
v.z % scalar.x,
v.w % scalar.x);
return tvec4<T, P>(v) %= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator%(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar % v.x,
scalar % v.y,
scalar % v.z,
scalar % v.w);
return tvec4<T, P>(scalar) %= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator%(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x % v.x,
scalar.x % v.y,
scalar.x % v.z,
scalar.x % v.w);
return tvec4<T, P>(scalar) %= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator%(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x % v2.x,
v1.y % v2.y,
v1.z % v2.z,
v1.w % v2.w);
return tvec4<T, P>(v1) %= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator&(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x & scalar,
v.y & scalar,
v.z & scalar,
v.w & scalar);
return tvec4<T, P>(v) %= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator&(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x & scalar.x,
v.y & scalar.x,
v.z & scalar.x,
v.w & scalar.x);
return tvec4<T, P>(v) %= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator&(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar & v.x,
scalar & v.y,
scalar & v.z,
scalar & v.w);
return tvec4<T, P>(scalar) &= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator&(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x & v.x,
scalar.x & v.y,
scalar.x & v.z,
scalar.x & v.w);
return tvec4<T, P>(scalar) &= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator&(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x & v2.x,
v1.y & v2.y,
v1.z & v2.z,
v1.w & v2.w);
return tvec4<T, P>(v1) &= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator|(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x | scalar,
v.y | scalar,
v.z | scalar,
v.w | scalar);
return tvec4<T, P>(v) |= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator|(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x | scalar.x,
v.y | scalar.x,
v.z | scalar.x,
v.w | scalar.x);
return tvec4<T, P>(v) |= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator|(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar | v.x,
scalar | v.y,
scalar | v.z,
scalar | v.w);
return tvec4<T, P>(v) |= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator|(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x | v.x,
scalar.x | v.y,
scalar.x | v.z,
scalar.x | v.w);
return tvec4<T, P>(v) |= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator|(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x | v2.x,
v1.y | v2.y,
v1.z | v2.z,
v1.w | v2.w);
return tvec4<T, P>(v1) |= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator^(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x ^ scalar,
v.y ^ scalar,
v.z ^ scalar,
v.w ^ scalar);
return tvec4<T, P>(v) ^= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator^(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x ^ scalar.x,
v.y ^ scalar.x,
v.z ^ scalar.x,
v.w ^ scalar.x);
return tvec4<T, P>(v) ^= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator^(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar ^ v.x,
scalar ^ v.y,
scalar ^ v.z,
scalar ^ v.w);
return tvec4<T, P>(v) ^= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator^(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x ^ v.x,
scalar.x ^ v.y,
scalar.x ^ v.z,
scalar.x ^ v.w);
return tvec4<T, P>(v) ^= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator^(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x ^ v2.x,
v1.y ^ v2.y,
v1.z ^ v2.z,
v1.w ^ v2.w);
return tvec4<T, P>(v1) ^= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator<<(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x << scalar,
v.y << scalar,
v.z << scalar,
v.w << scalar);
return tvec4<T, P>(v) <<= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator<<(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x << scalar.x,
v.y << scalar.x,
v.z << scalar.x,
v.w << scalar.x);
return tvec4<T, P>(v) <<= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator<<(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar << v.x,
scalar << v.y,
scalar << v.z,
scalar << v.w);
return tvec4<T, P>(scalar) <<= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator<<(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x << v.x,
scalar.x << v.y,
scalar.x << v.z,
scalar.x << v.w);
return tvec4<T, P>(scalar) <<= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator<<(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x << v2.x,
v1.y << v2.y,
v1.z << v2.z,
v1.w << v2.w);
return tvec4<T, P>(v1) <<= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator>>(tvec4<T, P> const & v, T scalar)
{
return tvec4<T, P>(
v.x >> scalar,
v.y >> scalar,
v.z >> scalar,
v.w >> scalar);
return tvec4<T, P>(v) >>= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator>>(tvec4<T, P> const & v, tvec1<T, P> const & scalar)
{
return tvec4<T, P>(
v.x >> scalar.x,
v.y >> scalar.x,
v.z >> scalar.x,
v.w >> scalar.x);
return tvec4<T, P>(v) >>= scalar;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator>>(T scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar >> v.x,
scalar >> v.y,
scalar >> v.z,
scalar >> v.w);
return tvec4<T, P>(scalar) >>= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator>>(tvec1<T, P> const & scalar, tvec4<T, P> const & v)
{
return tvec4<T, P>(
scalar.x >> v.x,
scalar.x >> v.y,
scalar.x >> v.z,
scalar.x >> v.w);
return tvec4<T, P>(scalar) >>= v;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator>>(tvec4<T, P> const & v1, tvec4<T, P> const & v2)
{
return tvec4<T, P>(
v1.x >> v2.x,
v1.y >> v2.y,
v1.z >> v2.z,
v1.w >> v2.w);
return tvec4<T, P>(v1) >>= v2;
}
template <typename T, precision P>
GLM_FUNC_QUALIFIER tvec4<T, P> operator~(tvec4<T, P> const & v)
{
return tvec4<T, P>(
~v.x,
~v.y,
~v.z,
~v.w);
return tvec4<T, P>(~v.x, ~v.y, ~v.z, ~v.w);
}
// -- Boolean operators --

44
glm/detail/type_vec4_simd.inl
View File

@ -16,6 +16,50 @@ namespace detail
return Result;
}
};
template <precision P>
struct compute_vec4_sub<float, P>
{
static tvec4<float, P> call(tvec4<float, P> const & a, tvec4<float, P> const & b)
{
tvec4<float, P> Result(uninitialize);
Result.data = _mm_sub_ps(a.data, b.data);
return Result;
}
};
template <precision P>
struct compute_vec4_mul<float, P>
{
static tvec4<float, P> call(tvec4<float, P> const & a, tvec4<float, P> const & b)
{
tvec4<float, P> Result(uninitialize);
Result.data = _mm_mul_ps(a.data, b.data);
return Result;
}
};
template <precision P>
struct compute_vec4_div<float, P>
{
static tvec4<float, P> call(tvec4<float, P> const & a, tvec4<float, P> const & b)
{
tvec4<float, P> Result(uninitialize);
Result.data = _mm_div_ps(a.data, b.data);
return Result;
}
};
template <>
struct compute_vec4_div<float, lowp>
{
static tvec4<float, lowp> call(tvec4<float, lowp> const & a, tvec4<float, lowp> const & b)
{
tvec4<float, lowp> Result(uninitialize);
Result.data = _mm_mul_ps(a.data, _mm_rcp_ps(b.data));
return Result;
}
};
}//namespace detail
# if !GLM_HAS_DEFAULTED_FUNCTIONS