代码

#include "immintrin.h"
#pragma GCC optimize("Ofast")
#pragma GCC target("avx2")
void my_memset(char *a, char c, unsigned n) {
	__m256i s;
	for(unsigned i=0;i<32;i++)((char*)&s)[i]=c;
	while(reinterpret_cast<unsigned long long>(a)&31)*a++=c,n--;
	while(n&255)a[--n]=c;
	__m256i*b=(__m256i*)a;
	n>>=5;
	for(unsigned i = 0; i != n; i+=8,b+=8) 
	{
		*b=s;
		*(b+1)=s;
		*(b+2)=s;
		*(b+3)=s;
		*(b+4)=s;
		*(b+5)=s;
		*(b+6)=s;
		*(b+7)=s;
	}
	for(unsigned i = 0; i < n; i++)
		a[i] = c;
}

评测结果