From 178ee52e7d0378b73a0738faa198fc723efed3db Mon Sep 17 00:00:00 2001
From: Ahmet Inan <inan@aicodix.de>
Date: Tue, 5 Mar 2024 09:38:34 +0100
Subject: [PATCH] added sorting for SIMD types

---
 sort.hh | 112 ++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 112 insertions(+)
 create mode 100644 sort.hh

diff --git a/sort.hh b/sort.hh
new file mode 100644
index 0000000..c36623d
--- /dev/null
+++ b/sort.hh
@@ -0,0 +1,112 @@
+/*
+Sorting for SIMD types
+
+Copyright 2024 Ahmet Inan <inan@aicodix.de>
+*/
+
+#pragma once
+
+#include <algorithm>
+
+template <int WIDTH>
+static inline SIMD<uint32_t, WIDTH> vorder(SIMD<float, WIDTH> a)
+{
+	SIMD<uint32_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<uint64_t, WIDTH> vorder(SIMD<double, WIDTH> a)
+{
+	SIMD<uint64_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<uint8_t, WIDTH> vorder(SIMD<int8_t, WIDTH> a)
+{
+	SIMD<uint8_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<uint16_t, WIDTH> vorder(SIMD<int16_t, WIDTH> a)
+{
+	SIMD<uint16_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<uint32_t, WIDTH> vorder(SIMD<int32_t, WIDTH> a)
+{
+	SIMD<uint32_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<uint64_t, WIDTH> vorder(SIMD<int64_t, WIDTH> a)
+{
+	SIMD<uint64_t, WIDTH> tmp;
+	for (int i = 0; i < WIDTH; ++i)
+		tmp.v[i] = i;
+	std::stable_sort(tmp.v, tmp.v+WIDTH, [a](int i, int j){ return a.v[i] < a.v[j]; });
+	return tmp;
+}
+
+template <int WIDTH>
+static inline SIMD<float, WIDTH> vsort(SIMD<float, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+
+template <int WIDTH>
+static inline SIMD<double, WIDTH> vsort(SIMD<double, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+
+template <int WIDTH>
+static inline SIMD<int8_t, WIDTH> vsort(SIMD<int8_t, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+
+template <int WIDTH>
+static inline SIMD<int16_t, WIDTH> vsort(SIMD<int16_t, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+
+template <int WIDTH>
+static inline SIMD<int32_t, WIDTH> vsort(SIMD<int32_t, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+
+template <int WIDTH>
+static inline SIMD<int64_t, WIDTH> vsort(SIMD<int64_t, WIDTH> a)
+{
+	std::sort(a.v, a.v+WIDTH);
+	return a;
+}
+