lib/Runtime/Language/SimdInt32x4OperationX86X64.cpp

//-------------------------------------------------------------------------------------------------------
// Copyright (C) Microsoft Corporation and contributors. All rights reserved.
// Licensed under the MIT license. See LICENSE.txt file in the project root for full license information.
//-------------------------------------------------------------------------------------------------------

#include "RuntimeLanguagePch.h"

#if _M_IX86 || _M_AMD64

namespace Js
{
    // SIMD.Int32x4 operation wrappers that cover intrinsics for x86/x64 system
    SIMDValue SIMDInt32x4Operation::OpInt32x4(int x, int y, int z, int w)
    {
        X86SIMDValue x86Result;
        x86Result.m128i_value = _mm_set_epi32(w, z, y, x);
        // Sets the 4 signed 32-bit integer values, note in revised order: starts with W below

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpSplat(int x)
    {
        X86SIMDValue x86Result;
        // set 4 signed 32-bit integers values to input value x
        x86Result.m128i_value = _mm_set1_epi32(x);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    // Conversions
    SIMDValue SIMDInt32x4Operation::OpFromFloat32x4(const SIMDValue& value, bool &throws)
    {
        X86SIMDValue x86Result = { 0 };
        X86SIMDValue v = X86SIMDValue::ToX86SIMDValue(value);
        X86SIMDValue temp;
        int mask = 0;

        // Converts the 4 single-precision, floating-point values to signed 32-bit integer values
        // using truncate, using truncate one instead of _mm_cvtps_epi32
        x86Result.m128i_value = _mm_cvttps_epi32(v.m128_value);

        // check if any value is potentially out of range (0x80000000 in output)
        temp.m128i_value = _mm_cmpeq_epi32(x86Result.m128i_value, X86_NEG_MASK_F4.m128i_value);
        mask = _mm_movemask_ps(temp.m128_value);

        if (mask)
        {
            // potential overflow. Do bound checks.
            temp.m128_value = _mm_cmpge_ps(v.m128_value, X86_TWO_31_F4.m128_value);
            mask = _mm_movemask_ps(temp.m128_value);
            if (mask)
            {
                throws = true;
                return X86SIMDValue::ToSIMDValue(x86Result);
            }
            temp.m128_value = _mm_cmplt_ps(v.m128_value, X86_NEG_TWO_31_F4.m128_value);
            mask = _mm_movemask_ps(temp.m128_value);
            if (mask)
            {
                throws = true;
                return X86SIMDValue::ToSIMDValue(x86Result);
            }
        }

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpFromFloat64x2(const SIMDValue& value)
    {
        X86SIMDValue x86Result;
        X86SIMDValue v = X86SIMDValue::ToX86SIMDValue(value);

        // Converts the 2 double-precision, floating-point values to 32-bit signed integers
        // using truncate. using truncate one instead of _mm_cvtpd_epi32
        x86Result.m128i_value = _mm_cvttpd_epi32(v.m128d_value);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    // Unary Ops
    SIMDValue SIMDInt32x4Operation::OpAbs(const SIMDValue& value)
    {
        SIMDValue result;

        X86SIMDValue x86Result;
        X86SIMDValue v = X86SIMDValue::ToX86SIMDValue(value);
        if (AutoSystemInfo::Data.SSE3Available())
        {
            x86Result.m128i_value = _mm_abs_epi32(v.m128i_value); // only available after SSE3
            result = X86SIMDValue::ToSIMDValue(x86Result);
        }
        else
        {
            X86SIMDValue  temp, SIGNMASK;
            SIGNMASK.m128i_value = _mm_srai_epi32(v.m128i_value, 31);                // mask = value >> 31
            temp.m128i_value = _mm_xor_si128(v.m128i_value, SIGNMASK.m128i_value);   // temp = value ^ mask
            x86Result.m128i_value = _mm_sub_epi32(temp.m128i_value, SIGNMASK.m128i_value);  // temp - mask
            result = X86SIMDValue::ToSIMDValue(x86Result);
        }
        return result;
    }

    SIMDValue SIMDInt32x4Operation::OpNeg(const SIMDValue& value)
    {
        X86SIMDValue x86Result;

        X86SIMDValue SIGNMASK, temp;
        X86SIMDValue negativeOnes = { { -1, -1, -1, -1 } };
        X86SIMDValue v = X86SIMDValue::ToX86SIMDValue(value);

        temp.m128i_value = _mm_andnot_si128(v.m128i_value, negativeOnes.m128i_value); // (~value) & (negative ones)
        SIGNMASK.m128i_value = _mm_set1_epi32(0x00000001);                            // set SIGNMASK to 1
        x86Result.m128i_value = _mm_add_epi32(SIGNMASK.m128i_value, temp.m128i_value);// add 4 integers respectively

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpNot(const SIMDValue& value)
    {
        X86SIMDValue x86Result;

        X86SIMDValue negativeOnes = { { -1, -1, -1, -1 } };
        X86SIMDValue temp = X86SIMDValue::ToX86SIMDValue(value);
        x86Result.m128i_value = _mm_andnot_si128(temp.m128i_value, negativeOnes.m128i_value);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpAdd(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        x86Result.m128i_value = _mm_add_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // a + b

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpSub(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        x86Result.m128i_value = _mm_sub_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // a - b

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpMul(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        SIMDValue result;
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        // a * b, only available in SSE4
        // x86Result.m128i_value = _mm_mullo_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value);
        // result = X86SIMDValue::ToSIMDValue(x86Result);

        // SSE2
        // mul 2,0: r0 = a0*b0; r1 = a2*b2
        __m128i tmp1 = _mm_mul_epu32(tmpaValue.m128i_value, tmpbValue.m128i_value);
        // mul 3,1: r0=a1*b1; r1=a3*b3
        __m128i tmp2 = _mm_mul_epu32(_mm_srli_si128(tmpaValue.m128i_value, 4), _mm_srli_si128(tmpbValue.m128i_value, 4));
        // shuffle x86Results to [63..0] and pack
        x86Result.m128i_value = _mm_unpacklo_epi32(_mm_shuffle_epi32(tmp1, _MM_SHUFFLE(0, 0, 2, 0)), _mm_shuffle_epi32(tmp2, _MM_SHUFFLE(0, 0, 2, 0)));
        result = X86SIMDValue::ToSIMDValue(x86Result);

        return result;
    }

    SIMDValue SIMDInt32x4Operation::OpAnd(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        x86Result.m128i_value = _mm_and_si128(tmpaValue.m128i_value, tmpbValue.m128i_value); // a & b

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpOr(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        x86Result.m128i_value = _mm_or_si128(tmpaValue.m128i_value, tmpbValue.m128i_value); // a | b

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpXor(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);

        x86Result.m128i_value = _mm_xor_si128(tmpaValue.m128i_value, tmpbValue.m128i_value); // a ^ b

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpMin(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        // choose the smaller value of the two parameters, only available after SSE4
        // x86Result.m128i_value = _mm_min_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value);
        // result = X86SIMDValue::ToSIMDValue(x86Result);

        // SSE2
        SIMDValue selector = SIMDInt32x4Operation::OpLessThan(aValue, bValue);
        return SIMDInt32x4Operation::OpSelect(selector, aValue, bValue);
    }

    SIMDValue SIMDInt32x4Operation::OpMax(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        // choose the larger value of the two parameters, only available after SSE4
        // x86Result.m128i_value = _mm_max_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // a ^ b
        // result = X86SIMDValue::ToSIMDValue(x86Result);

        // SSE2
        SIMDValue selector = SIMDInt32x4Operation::OpLessThan(bValue, aValue);
        return SIMDInt32x4Operation::OpSelect(selector, aValue, bValue);
    }

    SIMDValue SIMDInt32x4Operation::OpLessThan(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);
        x86Result.m128i_value = _mm_cmplt_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // compare a < b?

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpLessThanOrEqual(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        SIMDValue tmpResult = SIMDInt32x4Operation::OpGreaterThan(aValue, bValue);
        X86SIMDValue x86Result = X86SIMDValue::ToX86SIMDValue(tmpResult);

        X86SIMDValue negativeOnes = X86_ALL_NEG_ONES; // { { -1, -1, -1, -1 } };
        x86Result.m128i_value = _mm_andnot_si128(x86Result.m128i_value, negativeOnes.m128i_value);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpEqual(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);
        x86Result.m128i_value = _mm_cmpeq_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // compare a == b?

        return X86SIMDValue::ToSIMDValue(x86Result);
    }


    SIMDValue SIMDInt32x4Operation::OpNotEqual(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        SIMDValue tmpResult = SIMDInt32x4Operation::OpEqual(aValue, bValue);
        X86SIMDValue x86Result = X86SIMDValue::ToX86SIMDValue(tmpResult);

        X86SIMDValue negativeOnes = X86_ALL_NEG_ONES; // { { -1, -1, -1, -1 } };
        x86Result.m128i_value = _mm_andnot_si128(x86Result.m128i_value, negativeOnes.m128i_value);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpGreaterThan(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpaValue = X86SIMDValue::ToX86SIMDValue(aValue);
        X86SIMDValue tmpbValue = X86SIMDValue::ToX86SIMDValue(bValue);
        x86Result.m128i_value = _mm_cmpgt_epi32(tmpaValue.m128i_value, tmpbValue.m128i_value); // compare a > b?

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpGreaterThanOrEqual(const SIMDValue& aValue, const SIMDValue& bValue)
    {
        SIMDValue tmpResult = SIMDInt32x4Operation::OpLessThan(aValue, bValue);
        X86SIMDValue x86Result = X86SIMDValue::ToX86SIMDValue(tmpResult);

        X86SIMDValue negativeOnes = X86_ALL_NEG_ONES; // { { -1, -1, -1, -1 } };
        x86Result.m128i_value = _mm_andnot_si128(x86Result.m128i_value, negativeOnes.m128i_value);

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpShiftLeftByScalar(const SIMDValue& value, int count)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpValue = X86SIMDValue::ToX86SIMDValue(value);
        // Shifts the 4 signed 32-bit integers in a left by count bits while shifting in zeros
        x86Result.m128i_value = _mm_slli_epi32(tmpValue.m128i_value, count & SIMDUtils::SIMDGetShiftAmountMask(4));

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpShiftRightByScalar(const SIMDValue& value, int count)
    {
        X86SIMDValue x86Result;
        X86SIMDValue tmpValue = X86SIMDValue::ToX86SIMDValue(value);
        // Shifts the 4 signed 32-bit integers right by count bits while shifting in the sign bit
        x86Result.m128i_value = _mm_srai_epi32(tmpValue.m128i_value, count & SIMDUtils::SIMDGetShiftAmountMask(4));

        return X86SIMDValue::ToSIMDValue(x86Result);
    }

    SIMDValue SIMDInt32x4Operation::OpSelect(const SIMDValue& mV, const SIMDValue& tV, const SIMDValue& fV)
    {
        X86SIMDValue x86Result;
        X86SIMDValue maskValue  = X86SIMDValue::ToX86SIMDValue(mV);
        X86SIMDValue trueValue  = X86SIMDValue::ToX86SIMDValue(tV);
        X86SIMDValue falseValue = X86SIMDValue::ToX86SIMDValue(fV);

        X86SIMDValue tempTrue, tempFalse;
        tempTrue.m128i_value = _mm_and_si128(maskValue.m128i_value, trueValue.m128i_value); // mask & T
        tempFalse.m128i_value = _mm_andnot_si128(maskValue.m128i_value, falseValue.m128i_value); //!mask & F
        x86Result.m128i_value = _mm_or_si128(tempTrue.m128i_value, tempFalse.m128i_value);  // tempT | temp F

        return X86SIMDValue::ToSIMDValue(x86Result);
    }
}
#endif