We evaluated the performance of various C++ input methods by processing 1,000,000 randomly generated integers (within INT32 range) on an Intel Core i5-12400 system running Windows 11.

Tested Input Methods

• Standard scanf
• Standard cin
• Custom fast read
• Bitwise optimized fast read
• fread + bitwise optimized read
• cin with sync disabled
• cin with sync disabled and tie broken

Implementation Code

Data Generator

#include <cstdio>
#include <cstdlib>
#include <ctime>

const int SIZE = 1e6;

int generate_random() {
    bool sign = rand() & 1;
    long long value = rand() + (rand() << 15) + (rand() << 30);
    return sign ? -(value % (1LL << 32)) : value % (1LL << 32);
}

int main() {
    freopen("input.txt", "w", stdout);
    srand(time(0));
    for(int i = 0; i < SIZE; i++)
        printf("%d ", generate_random());
    return 0;
}

Performance Measurement

#include <windows.h>

void measure_performance(const char* method_name, function<void()> test_func) {
    LARGE_INTEGER start, end, freq;
    QueryPerformanceFrequency(&freq);
    QueryPerformanceCounter(&start);
    
    test_func();
    
    QueryPerformanceCounter(&end);
    printf("%s: %.4f ms\n", method_name, 
        1000.0 * (end.QuadPart - start.QuadPart) / freq.QuadPart);
}

Input Method Implementations

Standard scanf

int val;
for(int i = 0; i < SIZE; i++)
    scanf("%d", &val);

Fast Read with Bit Optimization

inline int fast_read() {
    int num = 0, sign = 1;
    char c = getchar();
    while(c < '0' || c > '9') {
        if(c == '-') sign = -1;
        c = getchar();
    }
    while(c >= '0' && c <= '9') {
        num = (num << 3) + (num << 1) + (c ^ 48);
        c = getchar();
    }
    return num * sign;
}

Buffered Read

const int BUFFER_SIZE = 1 << 20;
char buffer[BUFFER_SIZE];
char* buffer_ptr = buffer;

inline char get_char() {
    if(buffer_ptr == buffer + BUFFER_SIZE) {
        fread(buffer, 1, BUFFER_SIZE, stdin);
        buffer_ptr = buffer;
    }
    return *buffer_ptr++;
}

int buffered_read() {
    int num = 0, sign = 1;
    char c = get_char();
    while(!isdigit(c)) {
        if(c == '-') sign = -1;
        c = get_char();
    }
    while(isdigit(c)) {
        num = (num << 3) + (num << 1) + (c ^ 48);
        c = get_char();
    }
    return num * sign;
}

Performance Results

The buffered read implementation demonstrated the best performance, being approximately 10x faster than standard cin and 3x faster than optimized cin with disabled synchronization.