编译时间优化Math.pow
我读过，通过生成可以巧妙地使用位移和编译器生成的魔法常量的代码，可以在编译时通过已知常量优化乘法。
I've read that it's possible to optimize multiplication by a known constant at compile time by generating code which makes clever use of bit shifting and compiler-generated magic constants.
我对以类似的hacky方式优化乘方的可能性感兴趣。我知道平方求幂，因此我想您可以大幅优化
I'm interested in possibilities for optimizing exponentiation in a similar hacky manner. I know about exponentiation by squaring, so I guess you could aggressively optimize
pow(CONSTANT, n)
通过将预先计算的连续正方形CONSTANT嵌入到可执行文件中。我不知道这是否是一个好主意。
by embedding precomputed successive squares of CONSTANT into the executable. I'm not sure whether this is actually a good idea.
但是，当涉及到
pow(n, CONSTANT)
有没有一种已知的方法来有效地做到这一点？
I can't think of anything. Is there a known way to do this efficiently? Do the minds of StackOverflow have ideas, on either problem?
假设 pow（a，b） $ c>被实现为 exp（b * log（a））（它可能是），如果 a 是一个常数，你可以预先计算它的 log 。如果 b 是常数，它只有在它也是一个整数时才有用。
Assuming pow(a,b) is implemented as exp(b * log(a)) (which it probably is), if a is a constant then you can precompute its log. If b is a constant, it only helps if it is also an integer.