0

我正在尝试将数组的元素与 SIMD 并行求和。为了避免锁定,我使用了可组合线程本地,它并不总是在 16 个字节上对齐,因为 _mm_add_epi32 正在抛出异常

concurrency::combinable<__m128i> sum_combine;

int length = 40; // multiple of 8
concurrency::parallel_for(0, length , 8, [&](int it)
{

    __m128i v1 = _mm_load_si128(reinterpret_cast<__m128i*>(input_arr + it));
    __m128i v2 = _mm_load_si128(reinterpret_cast<__m128i*>(input_arr + it + sizeof(uint32_t)));

    auto temp = _mm_add_epi32(v1, v2);

    auto &sum = sum_combine.local();   // here is the problem 


    TRACE(L"%d\n", it);
    TRACE(L"add %x\n", &sum);

    ASSERT(((unsigned long)&sum & 15) == 0);

    sum = _mm_add_epi32(temp, sum);
}
);

这是 ppl.h 中可组合的定义

template<typename _Ty>
class combinable
{
private:

// Disable warning C4324: structure was padded due to __declspec(align())
// This padding is expected and necessary.
#pragma warning(push)
#pragma warning(disable: 4324)
    __declspec(align(64))
    struct _Node
    {
        unsigned long _M_key;
        _Ty _M_value;                   // this might not be aligned on 16 bytes
        _Node* _M_chain;

        _Node(unsigned long _Key, _Ty _InitialValue)
            : _M_key(_Key), _M_value(_InitialValue), _M_chain(NULL)
        {
        }
    };

有时对齐没问题,代码工作正常,但大多数时候它不起作用

我尝试使用以下内容,但这无法编译

union combine 
{
        unsigned short x[sizeof(__m128i) / sizeof(unsigned int)];
        __m128i y;
};

concurrency::combinable<combine> sum_combine;
then auto &sum = sum_combine.local().y; 

任何纠正对齐问题的建议,仍然使用可组合。

在 x64 上,由于默认的 16 字节对齐,它可以正常工作。在 x86 上,有时存在对齐问题。

4

2 回答 2

1

刚刚使用未对齐负载加载总和

auto &sum = sum_combine.local();


#if !defined(_M_X64) 

if (((unsigned long)&sum & 15) != 0)
{
    // just for breakpoint means, sum  is unaligned.
    int a = 5;
}
auto sum_temp = _mm_loadu_si128(&sum);
sum = _mm_add_epi32(temp, sum_temp);

#else

sum = _mm_add_epi32(temp, sum);

#endif
于 2015-07-12T15:48:14.307 回答
0

由于使用的sum变量_mm_add_epi32未对齐,因此您需要sum使用未对齐的加载/存储(_mm_loadu_si128/ _mm_storeu_si128)显式加载/存储。改变:

sum = _mm_add_epi32(temp, sum);

到:

__m128i v2 = _mm_loadu_si128((__m128i *)&sum);
v2 = _mm_add_epi32(v2, temp);
_mm_storeu_si128((__m128i *)&sum, v2);
于 2015-07-12T15:16:03.403 回答