英文:
Measuring the execution time of a function call in C using a function-like macro
问题
以下是您要翻译的内容：
使用标准库函数 clock() 是在 C 中测量函数调用执行时间的几种方法之一。例如，可以如下比较两个函数（具有不同原型）的执行时间：
#include <stdio.h>#include <time.h>int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1, b, a+b); }int main(void) {double start, elapsed1, elapsed2;printf("n f g\n");for(int n = 30; n <= 40; n++) {start = (double) clock();f(n);elapsed1 = (clock() - start) / CLOCKS_PER_SEC;start = (double) clock();g(n, 1, 1);elapsed2 = ((double) clock() - start) / CLOCKS_PER_SEC;printf("%d %.1f %.1f\n", n, elapsed1, elapsed2);}return 0;}
为了使这段代码更加简洁，我们可以定义一个类似函数的宏（因为在 C 中没有“通用函数”指针）：
#include <stdio.h>#include <time.h>double START, ELAPSED;#define CPUTIME(FCALL) (START = (double) clock(), FCALL, ELAPSED = ((double) clock() - START) / CLOCKS_PER_SEC)int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1,b,a+b); }int main(void) {printf("n f g\n");for(int n = 30; n <= 40; n++)printf("%d %.1f %.1f\n", n, CPUTIME(f(n)), CPUTIME(g(n, 1, 1)) );return 0;}
问题： 是否有办法使变量 START 和 ELAPSED 局部于宏定义，以便宏仍然可以作为非 void “函数”调用？是否有更好的方法使代码更简洁，而不使用宏定义？
英文:
Using the standard library function clock() is one of several methods to measure the execution time of a function call in C. So, for example, the comparison of the execution times of two functions (with different prototypes) can be done as follows:
#include <stdio.h>#include <time.h>int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1, b, a+b); }int main(void) {double start, elapsed1, elapsed2;printf("n f g\n");for(int n = 30; n <= 40; n++) {start = (double) clock();f(n);elapsed1 = (clock() - start) / CLOCKS_PER_SEC;start = (double) clock();g(n, 1, 1);elapsed2 = ((double) clock() - start) / CLOCKS_PER_SEC;printf("%d %.1f %.1f\n", n, elapsed1, elapsed2);}return 0;}
To make this code more concise, we could define a function-like macro (since there is no "generic function" pointer in C):
#include <stdio.h>#include <time.h>double START, ELAPSED;#define CPUTIME(FCALL) (START = (double) clock(), FCALL, ELAPSED = ((double) clock() - START) / CLOCKS_PER_SEC)int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1,b,a+b); }int main(void) {printf("n f g\n");for(int n = 30; n <= 40; n++)printf("%d %.1f %.1f\n", n, CPUTIME(f(n)), CPUTIME(g(n, 1, 1)) );return 0;}
Questions: Is there a way to make variables START and ELAPSED local to the macro definition so that the macro can still be called as a non-void "function"? Is there a better way to make the code more concise, without using a macro definition?
答案1
得分: 2
你可以使用非标准的 statement-expression C 扩展（至少在 clang、gcc 和 tinycc 中可用）。
#include <stdio.h>#include <time.h>#define CPUTIME(FCALL) ({ double START = clock(); FCALL; ((double) clock() - START) / CLOCKS_PER_SEC; })int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1, b, a+b); }int main(void) {printf("n f g\n");for(int n = 30; n <= 40; n++)printf("%d %.1f %.1f\n", n, CPUTIME(f(n)), CPUTIME(g(n, 1, 1)) );return 0;}
这是一个括号包围的块，它被评估为其中的最后一个语句所评估的结果。
我认为在标准 C 中没有通用的解决方案能够满足您的 API 约束。如果您追求标准 C 的线程安全性和/或稍微更好的代码生成，您可以使用您的原始解决方案，并在调用宏的块内（重新）声明变量。
英文:
You can use the nonstandard statement-expression C extension
(available in at least clang, gcc, and tinycc).
#include <stdio.h>#include <time.h>#define CPUTIME(FCALL) ({ double START = clock(); FCALL; ((double) clock() - START) / CLOCKS_PER_SEC; })int f(int n) { return n < 3 ? 1 : f(n-1) + f(n-2); }int g(int n, int a, int b) { return n < 3 ? b : g(n-1,b,a+b); }int main(void) {printf("n f g\n");for(int n = 30; n <= 40; n++)printf("%d %.1f %.1f\n", n, CPUTIME(f(n)), CPUTIME(g(n, 1, 1)) );return 0;}
It's a parenthesized block that's evaluated to what the last statement in it evaluates to.
I don't think there's a generic solution within standard C that would satisfy your API constraints. If you're after standard-c-only threadsafety and/or slightly better codegen, you could use your original solution and (re)declare the variable(s) locally within the block that calls the macro.
答案2
得分: 1
以下是翻译好的内容：
"在不使用宏定义的情况下，有没有更简洁的方法来编写代码呢？
虽然宏定义很巧妙，但普通的函数可以减少重复，并且更容易理解：
double seconds_since(clock_t start){return (double)(clock() - start) / CLOCKS_PER_SEC;}int main(void) {clock_t start;double elapsed1, elapsed2;printf("n f g\n");for(int n = 30; n <= 40; n++) {start = clock();f(n);elapsed1 = seconds_since(start);start = clock();g(n, 1, 1);elapsed2 = seconds_since(start);printf("%d %.1f %.1f\n", n, elapsed1, elapsed2);}return 0;}```"<details><summary>英文:</summary>> Is there a better way to make the code more concise, without using a macro definition?While the macro definition is clever, just a plain function can reduce the repetition and is simple to understand:double seconds_since(clock_t start){return (double)(clock() - start) / CLOCKS_PER_SEC;}int main(void) {clock_t start;double elapsed1, elapsed2;printf("n f g\n");for(int n = 30; n <= 40; n++) {start = clock();f(n);elapsed1 = seconds_since(start);start = clock();g(n, 1, 1);elapsed2 = seconds_since(start);printf("%d %.1f %.1f\n", n, elapsed1, elapsed2);}return 0;}</details>
通过集体智慧和协作来改善编程学习和解决问题的方式。致力于成为全球开发者共同参与的知识库，让每个人都能够通过互相帮助和分享经验来进步。
评论