Fast I/O Code Optimizer

№	Пользователь	Рейтинг
1	tourist	4009
2	jiangly	3823
3	Benq	3738
4	Radewoosh	3633
5	jqdai0815	3620
6	orzdevinwang	3529
7	ecnerwala	3446
8	Um_nik	3396
9	ksun48	3390
10	gamegame	3386

№	Пользователь	Вклад
1	cry	167
2	Um_nik	163
3	maomao90	162
3	atcoder_official	162
5	adamant	159
6	-is-this-fft-	158
7	awoo	157
8	TheScrasse	154
9	Dominater069	153
9	nor	153

I think it's very bad when we try to submit a problem fully implemented in C++ language, in the expected complexity, and we get a time limit exceeded verdict. Then, we realize we have used default readings and writing commands from C++ (cin/cout), and we try to change every occurrence to faster ones. I really don't like these kind of problems with a very tight time limit. Because of this, I've created a new tool called Fast I/O Code Optimizer. This tool is very simple and it was built to prevent the users from replacing all cin/cout commands to new ones.

To use this optimizer you only need to access the website fastio.pythonanywhere.com and follow a few steps. After accessing the website, it is needed to insert the code to be optimized in the text field from the left, then click the button Optimize it and it's done, the optimized code will be pasted in the text field from the right.

In tests, after using this tool, the execution time performance improved up to 50%. This program can also be very important in an online competition to prevent wasting time trying to optimize the code.

Github Project: github.com/bfs07/Fast-IO-Code-Optimizer

#include <iostream> #include <bitset> #include <cstring> static struct FASTIO { #define tplType template <typename T> #define tplSize template <size_t N> char READKEY; bool REMAINING = false; char bufWrite[20]; inline void ignore() {REMAINING?REMAINING=false:READKEY=getchar();} inline void flush() {fflush(stdout);} inline void IGNORE_CHAR() {if(REMAINING^=1)READKEY=getchar();} inline void IGNORE_CHAR_SPECIAL() {for(IGNORE_CHAR();READKEY=='\n'||READKEY=='\t'||READKEY==' ';READKEY=getchar());} tplType inline bool READ_INT(T &x) {bool sign=0;for(IGNORE_CHAR();!(isdigit(READKEY)||READKEY==EOF);READKEY=getchar())sign^=(READKEY=='-');if(READKEY==EOF) return REMAINING=false;for(x=0;isdigit(READKEY);READKEY=getchar())x=x*10+READKEY-'0';return sign?-x:x,REMAINING=true;} tplType inline bool READ_STRING(T &x) {IGNORE_CHAR_SPECIAL();if(READKEY==EOF) return REMAINING=false;for(x="";!(READKEY=='\n'||READKEY=='\t'||READKEY==' '||READKEY== EOF);READKEY=getchar())x+=READKEY;return REMAINING=true;} inline bool READ_GETLINE(std::string &x) {IGNORE_CHAR();if(READKEY==EOF)return REMAINING=false;for(x="";!(READKEY=='\n'||READKEY==EOF);READKEY=getchar())x+=READKEY;return REMAINING=false,true;} tplType inline bool READ_CHAR(T &x) {IGNORE_CHAR();if(READKEY==EOF) return REMAINING=false;for(;READKEY=='\n'||READKEY=='\t'||READKEY==' ';READKEY=getchar());return x=READKEY,REMAINING=false,true;} tplSize inline bool READ_CHAR_ARRAY(char(&x)[N]) {IGNORE_CHAR_SPECIAL();if(READKEY==EOF)return REMAINING=false;char*ptr=&x[0];for(;!(READKEY=='\n'||READKEY=='\t'||READKEY==' '||READKEY==EOF);READKEY=getchar())*ptr++=READKEY;return *ptr='\0',REMAINING = true;} inline bool READ_CHAR_ARRAY(char*&x) {std::string y;if(!READ_STRING(y))return false;return x=new char[int(y.size())+1],strcpy(x, y.c_str()),true;} tplType inline bool READ_FLOAT(T &x) {return (scanf("%f",&x)!=EOF);} tplType inline bool READ_DOUBLE(T &x) {double y;if(scanf("%lf",&y)==EOF)return false;else return x=y,true;} tplSize inline bool READ_BITSET(std::bitset<N> &x) {IGNORE_CHAR_SPECIAL();if(READKEY==EOF)return REMAINING=false;for(int i=0;READKEY=='0'||READKEY=='1';READKEY=getchar())x[i++]=READKEY-'0';return REMAINING=true;} inline bool READ_VAR(short int &x) {return READ_INT(x);} inline bool READ_VAR(int &x) {return READ_INT(x);} inline bool READ_VAR(long int &x) {return READ_INT(x);} inline bool READ_VAR(long long int &x) {return READ_INT(x);} inline bool READ_VAR(unsigned short int &x) {return READ_INT(x);} inline bool READ_VAR(unsigned int &x) {return READ_INT(x);} inline bool READ_VAR(unsigned long &x) {return READ_INT(x);} inline bool READ_VAR(unsigned long long &x) {return READ_INT(x);} inline bool READ_VAR(std::string &x) {return READ_STRING(x);} inline bool READ_VAR(char &x) {return READ_CHAR(x);} tplSize inline bool READ_VAR(char (&x)[N]) {return READ_CHAR_ARRAY(x);} inline bool READ_VAR(char*& x) {return READ_CHAR_ARRAY(x);} inline bool READ_VAR(float &x) {return READ_FLOAT(x);} inline bool READ_VAR(double &x) {return READ_DOUBLE(x);} inline bool READ_VAR(long double &x) {return READ_DOUBLE(x);} tplSize inline bool READ_VAR(std::bitset<N> &x) {return READ_BITSET(x);} tplType inline void WRITE_INT(T x) {if(x<0)putchar('-'),x=-x;char*bufPos=bufWrite;do{*bufPos++=x%10+'0';}while(x/=10);do{putchar(*--bufPos);}while(bufPos>bufWrite);} inline void WRITE_CHAR(char x) {putchar(x);} inline void WRITE_CHAR_ARRAY(const char *x) {while(*x!='\0')putchar(*x++);} inline void WRITE_STRING(std::string &x) {for(char c:x)putchar(c);} inline void WRITE_FLOAT(float x) {printf("%f", x);} tplType inline void WRITE_DOUBLE(T x) {printf("%lf",(double)x);} tplSize inline void WRITE_BITSET(std::bitset<N> &x) {for(int i=int(x.size())-1;i>=0;--i)putchar(x[i]+48);} inline void WRITE_VAR(bool x) {WRITE_INT(x);} inline void WRITE_VAR(short int x) {WRITE_INT(x);} inline void WRITE_VAR(int x) {WRITE_INT(x);} inline void WRITE_VAR(long int x) {WRITE_INT(x);} inline void WRITE_VAR(long long int x) {WRITE_INT(x);} inline void WRITE_VAR(unsigned short int x) {WRITE_INT(x);} inline void WRITE_VAR(unsigned int x) {WRITE_INT(x);} inline void WRITE_VAR(unsigned long x) {WRITE_INT(x);} inline void WRITE_VAR(unsigned long long x) {WRITE_INT(x);} inline void WRITE_VAR(std::string &x) {WRITE_STRING(x);} inline void WRITE_VAR(char x) {WRITE_CHAR(x);} inline void WRITE_VAR(const char *x) {WRITE_CHAR_ARRAY(x);} inline void WRITE_VAR(float x) {WRITE_FLOAT(x);} inline void WRITE_VAR(double x) {WRITE_DOUBLE(x);} inline void WRITE_VAR(long double x) {WRITE_DOUBLE(x);} tplSize inline void WRITE_VAR(std::bitset<N> &x) {WRITE_BITSET(x);} } __FIO__; /// ============================================================================================================== /// using namespace std; int main() { }

Комментарии (12)

Написать комментарий?

dmkz

6 лет назад, # |

Thanks, very helpful! 2^1000 without any error, checked by wolfram.

Do you have I / O speed test results? Maybe like this

→ Ответить

dorijanlendvaj

6 лет назад, # ^ |

He has some here but all of the bad performances seem to be caused by using endl.

Yes, I see, thanks

bfs.07, please, do not use endl so often, because every time what you use it output buffer flushes, you can try '\n' instead.

bfs.07

← Rev. 2 →

Thanks for helping me. I will update the test results replacing endl to '\n'. However, the tool still have an improvement of peformance, mainly, on C++11 submissions.

bfs.07, I got compilation error, when I tried to apply script to next code. Can you fix your script for overloads of operator>>?

Sorry, but as stated on README.md this tool doesn't work with overloads of operator>>. I think I would waste too much time to fix it so I won't. However, it's a open source project and I would be very greatful if you or anyone could help me.

ILoveBitches

+13

And what is the guarantee that you are not using the code of those who are using your website? ;) just a doubt. Thanks

Unfortunately, I can't think an easy way to do this the way the tool was implemented. However, I guarantee your data hasn't been used. As an alternative, you can download the project on GitHub and use it offline.

SPyofgame

5 лет назад, # |

← Rev. 8 →

I compress and edit a bit to minimize the work space

5 лет назад, # ^ |

Suggest: Why dont you use putchar_unlocked() or _putchar_nolock() or fwrite() instead for faster writing

Thanks for the suggestion, SPyofgame! Unfortunately, I won't be giving support to this tool. If possible, please create a pull request on the repository on Github that I will probably approve this change. putchar_unlocked() doesn't work on Codeforces, that's why I don't use that. I don't remember if I have considered using the other two methods.

Here is my blog recently about Fast Output Implementation

Блог пользователя bfs.07