c++ - _mm_store_si128 throws exception -

so i've been tryna learn see optimization on own , i'm not quite getting it, thought simple function zeroes memory easy implement, went on , tried implement myself.

here 0 memory function loops buffer start buffer end , uses _mm_store_si128 0 out.

bool zeromem( byte * _dest, uint _sz ) {     if ( _dest == nullptr )         return false;     __m128i 0 = _mm_setzero_si128( );      ( auto = rcast<__m128i*>( _dest ),           end = rcast<__m128i*>( _dest + _sz );           < end; ++i )     {         _mm_store_si128( i, 0 );     }     return true; } 

exception thrown: access violation (0x00000) though pointer not 0x00000.

the test did allocating 1024 bytes of memory , calling zeromem.

the exception thrown on first iteration.

_mm_store_si128 translates movdqa , requires operands aligned on 16-byte boundary cause exception. iirc, example windows doesn't implement explicit alignment exception, causes access violation. concerning memset implementation might interested in this post comparing different approaches filling memory block bytes.