From e9ec10a3823a3c734e70c99b4aa50ae2a7e7e81f Mon Sep 17 00:00:00 2001
From: Luke Street <luke.street@encounterpc.com>
Date: Wed, 3 Aug 2022 18:15:45 -0400
Subject: [PATCH] Add simd_none implementation

---
 include/zeus/simd/parallelism_v2_simd.hpp |   3 +
 include/zeus/simd/simd.hpp                |   5 +-
 include/zeus/simd/simd_none.hpp           | 224 ++++++++++++++++++++++
 src/CMatrix3f.cpp                         |  12 +-
 4 files changed, 236 insertions(+), 8 deletions(-)
 create mode 100644 include/zeus/simd/simd_none.hpp

diff --git a/include/zeus/simd/parallelism_v2_simd.hpp b/include/zeus/simd/parallelism_v2_simd.hpp
index d19d438..e946a29 100644
--- a/include/zeus/simd/parallelism_v2_simd.hpp
+++ b/include/zeus/simd/parallelism_v2_simd.hpp
@@ -1477,6 +1477,9 @@ private:
   friend class simd_mask;
 
 public:
+  constexpr __simd_storage(_Tp __rv) : __storage_{__rv, __rv, __rv, __rv} {}
+  constexpr __simd_storage(_Tp a, _Tp b, _Tp c, _Tp d) : __storage_{a, b, c, d} {}
+
   constexpr _Tp __get(size_t __index) const noexcept { return __storage_[__index]; };
   constexpr void __set(size_t __index, _Tp __val) noexcept { __storage_[__index] = __val; }
   constexpr std::enable_if_t<__num_element >= 4> __set4(float a, float b, float c, float d) noexcept {
diff --git a/include/zeus/simd/simd.hpp b/include/zeus/simd/simd.hpp
index a503b39..8570cc0 100644
--- a/include/zeus/simd/simd.hpp
+++ b/include/zeus/simd/simd.hpp
@@ -18,7 +18,7 @@ using namespace std;
 #elif __ARM_NEON
 #include "simd_neon.hpp"
 #else
-namespace simd_abi {
+namespace zeus::_simd::simd_abi {
 template <typename T>
 struct zeus_native {};
 template <>
@@ -29,7 +29,8 @@ template <>
 struct zeus_native<double> {
   using type = fixed_size<4>;
 };
-} // namespace simd_abi
+} // namespace zeus::_simd::simd_abi
+#include "simd_none.hpp"
 #endif
 #ifdef __GNUC__
 #pragma GCC diagnostic pop
diff --git a/include/zeus/simd/simd_none.hpp b/include/zeus/simd/simd_none.hpp
new file mode 100644
index 0000000..7b48a27
--- /dev/null
+++ b/include/zeus/simd/simd_none.hpp
@@ -0,0 +1,224 @@
+#pragma once
+#ifndef _ZEUS_SIMD_INCLUDED
+#error simd_none.hpp must not be included directly. Include simd.hpp instead.
+#endif
+namespace zeus::_simd {
+using m128_abi = __simd_abi<_StorageKind::_Array, 4>;
+using m128d_abi = __simd_abi<_StorageKind::_Array, 4>;
+
+// m128 ABI
+template <>
+inline simd<float, m128_abi> simd<float, m128_abi>::operator-() const {
+  return {-__s_.__storage_[0], -__s_.__storage_[1], -__s_.__storage_[2], -__s_.__storage_[3]};
+}
+
+inline simd<float, m128_abi> operator+(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  return {a[0] + b[0], a[1] + b[1], a[2] + b[2], a[3] + b[3]};
+}
+
+inline simd<float, m128_abi> operator-(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  return {a[0] - b[0], a[1] - b[1], a[2] - b[2], a[3] - b[3]};
+}
+
+inline simd<float, m128_abi> operator*(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  return {a[0] * b[0], a[1] * b[1], a[2] * b[2], a[3] * b[3]};
+}
+
+inline simd<float, m128_abi> operator/(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  return {a[0] / b[0], a[1] / b[1], a[2] / b[2], a[3] / b[3]};
+}
+
+inline simd<float, m128_abi>& operator+=(simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  a[0] += b[0];
+  a[1] += b[1];
+  a[2] += b[2];
+  a[3] += b[3];
+  return a;
+}
+
+inline simd<float, m128_abi>& operator-=(simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  a[0] -= b[0];
+  a[1] -= b[1];
+  a[2] -= b[2];
+  a[3] -= b[3];
+  return a;
+}
+
+inline simd<float, m128_abi>& operator*=(simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  a[0] *= b[0];
+  a[1] *= b[1];
+  a[2] *= b[2];
+  a[3] *= b[3];
+  return a;
+}
+
+inline simd<float, m128_abi>& operator/=(simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  a[0] /= b[0];
+  a[1] /= b[1];
+  a[2] /= b[2];
+  a[3] /= b[3];
+  return a;
+}
+
+inline simd<float, m128_abi>::mask_type operator==(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] == b[0];
+  ret[1] = a[1] == b[1];
+  ret[2] = a[2] == b[2];
+  ret[3] = a[3] == b[3];
+  return ret;
+}
+
+inline simd<float, m128_abi>::mask_type operator!=(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] != b[0];
+  ret[1] = a[1] != b[1];
+  ret[2] = a[2] != b[2];
+  ret[3] = a[3] != b[3];
+  return ret;
+}
+
+inline simd<float, m128_abi>::mask_type operator>=(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] >= b[0];
+  ret[1] = a[1] >= b[1];
+  ret[2] = a[2] >= b[2];
+  ret[3] = a[3] >= b[3];
+  return ret;
+}
+
+inline simd<float, m128_abi>::mask_type operator<=(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] <= b[0];
+  ret[1] = a[1] <= b[1];
+  ret[2] = a[2] <= b[2];
+  ret[3] = a[3] <= b[3];
+  return ret;
+}
+
+inline simd<float, m128_abi>::mask_type operator>(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] > b[0];
+  ret[1] = a[1] > b[1];
+  ret[2] = a[2] > b[2];
+  ret[3] = a[3] > b[3];
+  return ret;
+}
+
+inline simd<float, m128_abi>::mask_type operator<(const simd<float, m128_abi>& a, const simd<float, m128_abi>& b) {
+  simd<float, m128_abi>::mask_type ret;
+  ret[0] = a[0] < b[0];
+  ret[1] = a[1] < b[1];
+  ret[2] = a[2] < b[2];
+  ret[3] = a[3] < b[3];
+  return ret;
+}
+
+// m128d ABI
+template <>
+inline simd<double, m128d_abi> simd<double, m128d_abi>::operator-() const {
+  return {-__s_.__storage_[0], -__s_.__storage_[1], -__s_.__storage_[2], -__s_.__storage_[3]};
+}
+
+inline simd<double, m128d_abi> operator+(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  return {a[0] + b[0], a[1] + b[1], a[2] + b[2], a[3] + b[3]};
+}
+
+inline simd<double, m128d_abi> operator-(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  return {a[0] - b[0], a[1] - b[1], a[2] - b[2], a[3] - b[3]};
+}
+
+inline simd<double, m128d_abi> operator*(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  return {a[0] * b[0], a[1] * b[1], a[2] * b[2], a[3] * b[3]};
+}
+
+inline simd<double, m128d_abi> operator/(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  return {a[0] / b[0], a[1] / b[1], a[2] / b[2], a[3] / b[3]};
+}
+
+inline simd<double, m128d_abi>& operator+=(simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  a[0] += b[0];
+  a[1] += b[1];
+  a[2] += b[2];
+  a[3] += b[3];
+  return a;
+}
+
+inline simd<double, m128d_abi>& operator-=(simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  a[0] -= b[0];
+  a[1] -= b[1];
+  a[2] -= b[2];
+  a[3] -= b[3];
+  return a;
+}
+
+inline simd<double, m128d_abi>& operator*=(simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  a[0] *= b[0];
+  a[1] *= b[1];
+  a[2] *= b[2];
+  a[3] *= b[3];
+  return a;
+}
+
+inline simd<double, m128d_abi>& operator/=(simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  a[0] /= b[0];
+  a[1] /= b[1];
+  a[2] /= b[2];
+  a[3] /= b[3];
+  return a;
+}
+
+inline simd<double, m128d_abi>::mask_type operator==(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] == b[0];
+  ret[1] = a[1] == b[1];
+  ret[2] = a[2] == b[2];
+  ret[3] = a[3] == b[3];
+  return ret;
+}
+
+inline simd<double, m128d_abi>::mask_type operator!=(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] != b[0];
+  ret[1] = a[1] != b[1];
+  ret[2] = a[2] != b[2];
+  ret[3] = a[3] != b[3];
+  return ret;
+}
+
+inline simd<double, m128d_abi>::mask_type operator>=(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] >= b[0];
+  ret[1] = a[1] >= b[1];
+  ret[2] = a[2] >= b[2];
+  ret[3] = a[3] >= b[3];
+  return ret;
+}
+
+inline simd<double, m128d_abi>::mask_type operator<=(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] <= b[0];
+  ret[1] = a[1] <= b[1];
+  ret[2] = a[2] <= b[2];
+  ret[3] = a[3] <= b[3];
+  return ret;
+}
+
+inline simd<double, m128d_abi>::mask_type operator>(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] > b[0];
+  ret[1] = a[1] > b[1];
+  ret[2] = a[2] > b[2];
+  ret[3] = a[3] > b[3];
+  return ret;
+}
+
+inline simd<double, m128d_abi>::mask_type operator<(const simd<double, m128d_abi>& a, const simd<double, m128d_abi>& b) {
+  simd<double, m128d_abi>::mask_type ret;
+  ret[0] = a[0] < b[0];
+  ret[1] = a[1] < b[1];
+  ret[2] = a[2] < b[2];
+  ret[3] = a[3] < b[3];
+  return ret;
+}
+} // namespace zeus::_simd
diff --git a/src/CMatrix3f.cpp b/src/CMatrix3f.cpp
index 823cd30..441d84c 100644
--- a/src/CMatrix3f.cpp
+++ b/src/CMatrix3f.cpp
@@ -46,15 +46,15 @@ void CMatrix3f::transpose() {
   float tmp;
 
   tmp = m[0][1];
-  m[0][1] = m[1][0];
+  m[0][1] = m[1][0].operator float();
   m[1][0] = tmp;
 
   tmp = m[0][2];
-  m[0][2] = m[2][0];
+  m[0][2] = m[2][0].operator float();
   m[2][0] = tmp;
 
   tmp = m[1][2];
-  m[1][2] = m[2][1];
+  m[1][2] = m[2][1].operator float();
   m[2][1] = tmp;
 #endif
 }
@@ -80,15 +80,15 @@ CMatrix3f CMatrix3f::transposed() const {
   float tmp;
 
   tmp = ret.m[0][1];
-  ret.m[0][1] = ret.m[1][0];
+  ret.m[0][1] = ret.m[1][0].operator float();
   ret.m[1][0] = tmp;
 
   tmp = m[0][2];
-  ret.m[0][2] = ret.m[2][0];
+  ret.m[0][2] = ret.m[2][0].operator float();
   ret.m[2][0] = tmp;
 
   tmp = m[1][2];
-  ret.m[1][2] = ret.m[2][1];
+  ret.m[1][2] = ret.m[2][1].operator float();
   ret.m[2][1] = tmp;
 
   return ret;