我曾经和一个人一起工作过，他说“如果编译器笨到不能弄清楚你要做什么，并且不能优化它，那么你的编译器就坏了，是时候换一个新的了”。我确信在某些情况下汇编程序会打败你的C代码，但是如果你发现自己经常使用汇编程序来“赢得”编译器，那么你的编译器就完蛋了。

对于编写试图强制查询计划器执行操作的“优化”SQL也是如此。如果您发现自己重新安排查询以让计划器执行您想要的操作，那么您的查询计划器就完蛋了——请更换一个新的计划器。

我不能给出具体的例子，因为那是很多年前的事情了，但是在很多情况下，手工编写的汇编程序可以胜过任何编译器。原因:

您可以偏离调用约定，在寄存器中传递参数。 您可以仔细考虑如何使用寄存器，避免将变量存储在内存中。 对于跳转表之类的东西，可以避免检查索引的边界。

基本上，编译器在优化方面做得很好，这几乎总是“足够好”，但在某些情况下(如图形渲染)，你要为每一个周期付出高昂的代价，你可以走捷径，因为你知道代码，而编译器不能，因为它必须在安全的方面。

事实上，我听说过一些图形渲染代码，其中一个例程，如直线绘制或多边形填充例程，实际上在堆栈上生成了一小块机器代码并在那里执行，以避免关于线条样式、宽度、模式等的连续决策。

也就是说，我想让编译器为我生成好的汇编代码，但又不太聪明，它们通常都是这样做的。事实上，我讨厌Fortran的一个原因是它为了“优化”而打乱代码，通常没有什么重要的目的。

通常，当应用程序出现性能问题时，都是由于浪费的设计造成的。这些天，我永远不会推荐汇编程序的性能，除非整个应用程序已经在它的生命周期内进行了调优，仍然不够快，并且把所有的时间都花在了紧凑的内部循环中。

补充:我见过很多用汇编语言编写的应用程序，与C、Pascal、Fortran等语言相比，汇编语言的主要速度优势是因为程序员在用汇编语言编码时要谨慎得多。他或她每天要写大约100行代码，不管哪种语言，在编译器语言中，这将等于3或400条指令。

几乎任何时候编译器看到浮点代码，如果你使用的是旧的糟糕的编译器，手写的版本会更快。(2019年更新:对于现代编译器来说，这并不普遍。特别是在编译x87以外的东西时;编译器更容易使用SSE2或AVX进行标量数学运算，或任何具有平面FP寄存器集的非x86，不像x87的寄存器堆栈。)

主要原因是编译器不能执行任何健壮的优化。关于这个主题的讨论，请参阅来自MSDN的这篇文章。下面是一个例子，其中汇编版本的速度是C版本的两倍(用VS2K5编译):

#include "stdafx.h"
#include <windows.h>

float KahanSum(const float *data, int n)
{
   float sum = 0.0f, C = 0.0f, Y, T;

   for (int i = 0 ; i < n ; ++i) {
      Y = *data++ - C;
      T = sum + Y;
      C = T - sum - Y;
      sum = T;
   }

   return sum;
}

float AsmSum(const float *data, int n)
{
  float result = 0.0f;

  _asm
  {
    mov esi,data
    mov ecx,n
    fldz
    fldz
l1:
    fsubr [esi]
    add esi,4
    fld st(0)
    fadd st(0),st(2)
    fld st(0)
    fsub st(0),st(3)
    fsub st(0),st(2)
    fstp st(2)
    fstp st(2)
    loop l1
    fstp result
    fstp result
  }

  return result;
}

int main (int, char **)
{
  int count = 1000000;

  float *source = new float [count];

  for (int i = 0 ; i < count ; ++i) {
    source [i] = static_cast <float> (rand ()) / static_cast <float> (RAND_MAX);
  }

  LARGE_INTEGER start, mid, end;

  float sum1 = 0.0f, sum2 = 0.0f;

  QueryPerformanceCounter (&start);

  sum1 = KahanSum (source, count);

  QueryPerformanceCounter (&mid);

  sum2 = AsmSum (source, count);

  QueryPerformanceCounter (&end);

  cout << "  C code: " << sum1 << " in " << (mid.QuadPart - start.QuadPart) << endl;
  cout << "asm code: " << sum2 << " in " << (end.QuadPart - mid.QuadPart) << endl;

  return 0;
}

和一些数字从我的PC运行默认版本*:

  C code: 500137 in 103884668
asm code: 500137 in 52129147

出于兴趣，我用dec/jnz交换了循环，它对计时没有影响——有时更快，有时更慢。我想内存有限的方面使其他优化相形见绌。(编者注:更可能的情况是，FP延迟瓶颈足以隐藏循环的额外成本。对奇数/偶数元素并行进行两个Kahan求和，并在最后添加它们，可能会加快2倍的速度。)

哎呀，我正在运行一个稍微不同的代码版本，它输出的数字是错误的(即C更快!)修正并更新了结果。

C语言常常需要做一些从汇编编码员的角度看来不必要的事情，这只是因为C标准这么说。

例如，整数提升。如果你想在C语言中移动一个char变量，人们通常会期望代码实际上只做一个比特的移动。

然而，标准强制编译器在移位之前将符号扩展为int，然后将结果截断为char，这可能会使代码复杂化，这取决于目标处理器的架构。

使用SIMD指令的矩阵操作可能比编译器生成的代码更快。

以下是我个人经历中的几个例子:

Access to instructions that are not accessible from C. For instance, many architectures (like x86-64, IA-64, DEC Alpha, and 64-bit MIPS or PowerPC) support a 64 bit by 64 bit multiplication producing a 128 bit result. GCC recently added an extension providing access to such instructions, but before that assembly was required. And access to this instruction can make a huge difference on 64-bit CPUs when implementing something like RSA - sometimes as much as a factor of 4 improvement in performance. Access to CPU-specific flags. The one that has bitten me a lot is the carry flag; when doing a multiple-precision addition, if you don't have access to the CPU carry bit one must instead compare the result to see if it overflowed, which takes 3-5 more instructions per limb; and worse, which are quite serial in terms of data accesses, which kills performance on modern superscalar processors. When processing thousands of such integers in a row, being able to use addc is a huge win (there are superscalar issues with contention on the carry bit as well, but modern CPUs deal pretty well with it). SIMD. Even autovectorizing compilers can only do relatively simple cases, so if you want good SIMD performance it's unfortunately often necessary to write the code directly. Of course you can use intrinsics instead of assembly but once you're at the intrinsics level you're basically writing assembly anyway, just using the compiler as a register allocator and (nominally) instruction scheduler. (I tend to use intrinsics for SIMD simply because the compiler can generate the function prologues and whatnot for me so I can use the same code on Linux, OS X, and Windows without having to deal with ABI issues like function calling conventions, but other than that the SSE intrinsics really aren't very nice - the Altivec ones seem better though I don't have much experience with them). As examples of things a (current day) vectorizing compiler can't figure out, read about bitslicing AES or SIMD error correction - one could imagine a compiler that could analyze algorithms and generate such code, but it feels to me like such a smart compiler is at least 30 years away from existing (at best).

On the other hand, multicore machines and distributed systems have shifted many of the biggest performance wins in the other direction - get an extra 20% speedup writing your inner loops in assembly, or 300% by running them across multiple cores, or 10000% by running them across a cluster of machines. And of course high level optimizations (things like futures, memoization, etc) are often much easier to do in a higher level language like ML or Scala than C or asm, and often can provide a much bigger performance win. So, as always, there are tradeoffs to be made.