负数变成0 A) int x = -12; if (x<0) { x = 0; } B) int x = -12; x = max(0, x) C） int x = -12; 位操作 请高手提供一下用bitwise （位操作）来实现这一功能。 同时，请问，以上三种方法，哪种速度最快？ 且还有更快的方法吗？
负数变成0 A) int x = -12; if (x<0) { x = 0; } B) int x = -12; x = max(0, x) C） int x = -12; 位操作 请高手提供一下用bitwise （位操作）来实现这一功能。 同时，请问，以上三种方法，哪种速度最快？ 且还有更快的方法吗？
microsat 发表于 2024-03-20 12:12
int32_t x = -12 x= (~(x >> 31)) & x Note that here we expect that x>>31 is arithmetic right shift. It is compiler implementation-defined. But for most of compilers, it is true. But being frankly, I think that bit operation may not be faster than highly optimized C++ code with either A or B, unless you want to do vector operations like SIMD.
B) int x = -12; x = max(0, x)
C） int x = -12; 位操作
请高手提供一下用bitwise （位操作）来实现这一功能。 同时，请问，以上三种方法，哪种速度最快？ 且还有更快的方法吗？
int x = -12; x = (x >> 31) & x; // 如果x是正数，x >> 31为0，结果是x；如果是负数，x >> 31为-1（即所有位都是1），与x按位与的结果是0
等高手答案
int32_t x = -12 x= (~(x >> 31)) & x
Note that here we expect that x>>31 is arithmetic right shift. It is compiler implementation-defined. But for most of compilers, it is true.
But being frankly, I think that bit operation may not be faster than highly optimized C++ code with either A or B, unless you want to do vector operations like SIMD.
你异或搞反了吧？ 哈哈
x = x^x
纯是脱裤子放屁！