SIMD alignment issue with PPL Combinable

Question

I am trying to sum the elements of an array in parallel with SIMD. To avoid blocking, I use combinable thread local, which is not always 16 byte aligned because of this _mm_add_epi32 throws an exception

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); } ); 

here is the defination combinable of ppl.h

 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) { } }; 

sometimes alignment is ok and the code works fine but most of the time it doesn't work

I tried using the following, but this will not compile

 union combine { unsigned short x[sizeof(__m128i) / sizeof(unsigned int)]; __m128i y; }; concurrency::combinable<combine> sum_combine; then auto &sum = sum_combine.local().y; 

Any suggestions to fix the alignment problem still using combinable.

On x64, it works fine when aligning by default to 16 bytes by default. On x86, alignment issues sometimes occur.

Answer 1

Just loaded amount using unloaded load

 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 

Answer 2

Since the sum variable used with _mm_add_epi32 is not aligned, you need to explicitly load / save sum using non-standard loads / storages ( _mm_loadu_si128 / _mm_storeu_si128 ). Change:

 sum = _mm_add_epi32(temp, sum); 

in

 __m128i v2 = _mm_loadu_si128((__m128i *)&sum); v2 = _mm_add_epi32(v2, temp); _mm_storeu_si128((__m128i *)&sum, v2);