vectorize8.h

#ifndef OPENMM_VECTORIZE8_H_
#define OPENMM_VECTORIZE8_H_

/* -------------------------------------------------------------------------- *
 *                                   OpenMM                                   *
 * -------------------------------------------------------------------------- *
 * This is part of the OpenMM molecular simulation toolkit originating from   *
 * Simbios, the NIH National Center for Physics-Based Simulation of           *
 * Biological Structures at Stanford, funded under the NIH Roadmap for        *
 * Medical Research, grant U54 GM072970. See https://simtk.org.               *
 *                                                                            *
 * Portions copyright (c) 2013-2014 Stanford University and the Authors.      *
 * Authors: Peter Eastman                                                     *
 * Contributors:                                                              *
 *                                                                            *
 * Permission is hereby granted, free of charge, to any person obtaining a    *
 * copy of this software and associated documentation files (the "Software"), *
 * to deal in the Software without restriction, including without limitation  *
 * the rights to use, copy, modify, merge, publish, distribute, sublicense,   *
 * and/or sell copies of the Software, and to permit persons to whom the      *
 * Software is furnished to do so, subject to the following conditions:       *
 *                                                                            *
 * The above copyright notice and this permission notice shall be included in *
 * all copies or substantial portions of the Software.                        *
 *                                                                            *
 * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR *
 * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,   *
 * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL    *
 * THE AUTHORS, CONTRIBUTORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,    *
 * DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR      *
 * OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE  *
 * USE OR OTHER DEALINGS IN THE SOFTWARE.                                     *
 * -------------------------------------------------------------------------- */

#include "vectorize.h"
#include <immintrin.h>

// This file defines classes and functions to simplify vectorizing code with AVX.

class ivec8;

class fvec8 {
public:
    __m256 val;
    
    fvec8() {}
    fvec8(float v) : val(_mm256_set1_ps(v)) {}
    fvec8(float v1, float v2, float v3, float v4, float v5, float v6, float v7, float v8) : val(_mm256_set_ps(v8, v7, v6, v5, v4, v3, v2, v1)) {}
    fvec8(__m256 v) : val(v) {}
    fvec8(const float* v) : val(_mm256_loadu_ps(v)) {}
    operator __m256() const {
        return val;
    }
    fvec4 lowerVec() const {
        return _mm256_castps256_ps128(val);
    }
    fvec4 upperVec() const {
        return _mm256_extractf128_ps(val, 1);
    }
    void store(float* v) const {
        _mm256_storeu_ps(v, val);
    }
    fvec8 operator+(const fvec8& other) const {
        return _mm256_add_ps(val, other);
    }
    fvec8 operator-(const fvec8& other) const {
        return _mm256_sub_ps(val, other);
    }
    fvec8 operator*(const fvec8& other) const {
        return _mm256_mul_ps(val, other);
    }
    fvec8 operator/(const fvec8& other) const {
        return _mm256_div_ps(val, other);
    }
    void operator+=(const fvec8& other) {
        val = _mm256_add_ps(val, other);
    }
    void operator-=(const fvec8& other) {
        val = _mm256_sub_ps(val, other);
    }
    void operator*=(const fvec8& other) {
        val = _mm256_mul_ps(val, other);
    }
    void operator/=(const fvec8& other) {
        val = _mm256_div_ps(val, other);
    }
    fvec8 operator-() const {
        return _mm256_sub_ps(_mm256_set1_ps(0.0f), val);
    }
    fvec8 operator&(const fvec8& other) const {
        return _mm256_and_ps(val, other);
    }
    fvec8 operator|(const fvec8& other) const {
        return _mm256_or_ps(val, other);
    }
    fvec8 operator==(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_EQ_OQ);
    }
    fvec8 operator!=(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_NEQ_OQ);
    }
    fvec8 operator>(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_GT_OQ);
    }
    fvec8 operator<(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_LT_OQ);
    }
    fvec8 operator>=(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_GE_OQ);
    }
    fvec8 operator<=(const fvec8& other) const {
        return _mm256_cmp_ps(val, other, _CMP_LE_OQ);
    }
    operator ivec8() const;
};

class ivec8 {
public:
    __m256i val;
    
    ivec8() {}
    ivec8(int v) : val(_mm256_set1_epi32(v)) {}
    ivec8(int v1, int v2, int v3, int v4, int v5, int v6, int v7, int v8) : val(_mm256_set_epi32(v8, v7, v6, v5, v4, v3, v2, v1)) {}
    ivec8(__m256i v) : val(v) {}
    ivec8(const int* v) : val(_mm256_loadu_si256((const __m256i*) v)) {}
    operator __m256i() const {
        return val;
    }
    ivec4 lowerVec() const {
        return _mm256_castsi256_si128(val);
    }
    ivec4 upperVec() const {
        return _mm256_extractf128_si256(val, 1);
    }
    void store(int* v) const {
        _mm256_storeu_si256((__m256i*) v, val);
    }
    ivec8 operator&(const ivec8& other) const {
        return _mm256_castps_si256(_mm256_and_ps(_mm256_castsi256_ps(val), _mm256_castsi256_ps(other.val)));
    }
    ivec8 operator|(const ivec8& other) const {
        return _mm256_castps_si256(_mm256_or_ps(_mm256_castsi256_ps(val), _mm256_castsi256_ps(other.val)));
    }
    operator fvec8() const;
};

// Conversion operators.

inline fvec8::operator ivec8() const {
    return _mm256_cvttps_epi32(val);
}

inline ivec8::operator fvec8() const {
    return _mm256_cvtepi32_ps(val);
}

// Functions that operate on fvec8s.

static inline fvec8 floor(const fvec8& v) {
    return fvec8(_mm256_round_ps(v.val, 0x09));
}

static inline fvec8 ceil(const fvec8& v) {
    return fvec8(_mm256_round_ps(v.val, 0x0A));
}

static inline fvec8 round(const fvec8& v) {
    return fvec8(_mm256_round_ps(v.val, _MM_FROUND_TO_NEAREST_INT));
}

static inline fvec8 min(const fvec8& v1, const fvec8& v2) {
    return fvec8(_mm256_min_ps(v1.val, v2.val));
}

static inline fvec8 max(const fvec8& v1, const fvec8& v2) {
    return fvec8(_mm256_max_ps(v1.val, v2.val));
}

static inline fvec8 abs(const fvec8& v) {
    static const __m256 mask = _mm256_castsi256_ps(_mm256_set1_epi32(0x7FFFFFFF));
    return fvec8(_mm256_and_ps(v.val, mask));
}

static inline fvec8 sqrt(const fvec8& v) {
    return fvec8(_mm256_sqrt_ps(v.val));
}

static inline float dot8(const fvec8& v1, const fvec8& v2) {
    fvec8 result = _mm256_dp_ps(v1, v2, 0xF1);
    return _mm_cvtss_f32(result.lowerVec())+_mm_cvtss_f32(result.upperVec());
}

static inline void transpose(const fvec4& in1, const fvec4& in2, const fvec4& in3, const fvec4& in4, const fvec4& in5, const fvec4& in6, const fvec4& in7, const fvec4& in8, fvec8& out1, fvec8& out2, fvec8& out3, fvec8& out4) {
    fvec4 i1 = in1, i2 = in2, i3 = in3, i4 = in4;
    fvec4 i5 = in5, i6 = in6, i7 = in7, i8 = in8;
    _MM_TRANSPOSE4_PS(i1, i2, i3, i4);
    _MM_TRANSPOSE4_PS(i5, i6, i7, i8);
#ifdef _MSC_VER
    // Visual Studio has a bug in _mm256_castps128_ps256, so we have to use the more expensive _mm256_insertf128_ps.
    out1 = _mm256_insertf128_ps(out1, i1, 0);
    out2 = _mm256_insertf128_ps(out2, i2, 0);
    out3 = _mm256_insertf128_ps(out3, i3, 0);
    out4 = _mm256_insertf128_ps(out4, i4, 0);
#else
    out1 = _mm256_castps128_ps256(i1);
    out2 = _mm256_castps128_ps256(i2);
    out3 = _mm256_castps128_ps256(i3);
    out4 = _mm256_castps128_ps256(i4);
#endif
    out1 = _mm256_insertf128_ps(out1, i5, 1);
    out2 = _mm256_insertf128_ps(out2, i6, 1);
    out3 = _mm256_insertf128_ps(out3, i7, 1);
    out4 = _mm256_insertf128_ps(out4, i8, 1);
}

static inline void transpose(const fvec8& in1, const fvec8& in2, const fvec8& in3, const fvec8& in4, fvec4& out1, fvec4& out2, fvec4& out3, fvec4& out4, fvec4& out5, fvec4& out6, fvec4& out7, fvec4& out8) {
    out1 = in1.lowerVec();
    out2 = in2.lowerVec();
    out3 = in3.lowerVec();
    out4 = in4.lowerVec();
    _MM_TRANSPOSE4_PS(out1, out2, out3, out4);
    out5 = in1.upperVec();
    out6 = in2.upperVec();
    out7 = in3.upperVec();
    out8 = in4.upperVec();
    _MM_TRANSPOSE4_PS(out5, out6, out7, out8);
}

// Functions that operate on ivec8s.

static inline bool any(const ivec8& v) {
    return !_mm256_testz_si256(v, _mm256_set1_epi32(0xFFFFFFFF));
}

// Mathematical operators involving a scalar and a vector.

static inline fvec8 operator+(float v1, const fvec8& v2) {
    return fvec8(v1)+v2;
}

static inline fvec8 operator-(float v1, const fvec8& v2) {
    return fvec8(v1)-v2;
}

static inline fvec8 operator*(float v1, const fvec8& v2) {
    return fvec8(v1)*v2;
}

static inline fvec8 operator/(float v1, const fvec8& v2) {
    return fvec8(v1)/v2;
}

// Operations for blending fvec8s based on an ivec8.

static inline fvec8 blend(const fvec8& v1, const fvec8& v2, const ivec8& mask) {
    return fvec8(_mm256_blendv_ps(v1.val, v2.val, _mm256_castsi256_ps(mask.val)));
}

#endif /*OPENMM_VECTORIZE8_H_*/