Check Which Loops Were Vectorized using MSVC Compiler

Modern compilers often apply auto-vectorization to loops to improve performance by utilizing SIMD instructions. This enables the CPU to execute multiple operations simultaneously using vector registers, significantly enhancing performance in computation-heavy code. This tutorial shows how to check which loops were vectorized using MSVC compiler.

Here's a simple function that divides an array of floats by a constant divisor:

#include <stdio.h>

void divide(float *data, const size_t n, const float divisor) {
    for (size_t i = 0; i < n; ++i) {
        data[i] /= divisor;
    }
}

int main() {
    const int n = 18;
    float a[] = {
        0, 14, 28, 42, 56, 70, 84, 98, 112, 126, 140, 154, 168, 182, 196, 210, 224, 255
    };

    divide(a, n, 255);
    for (size_t i = 0; i < n; ++i) {
        printf("%f ", a[i]);
    }

    return 0;
}

This loop is well-suited for vectorization, as it performs a straightforward operation on a contiguous block of data.

The /Qvec-report:2 option tells to output the report which loops were vectorized. To compile the code and produce the vectorization report, use the following command:

cl /nologo /O2 /Qvec-report:2 main.c > vec.log

• /nologo - hides MSVC startup banner.
• /O2 - enables full optimizations (includes vectorization).
• /Qvec-report:2 - shows detailed vectorization reports (both success and failure).
• > vec.log - redirects output to a file.

After compiling, open vec.log. You might see a message like:

main.c

--- Analyzing function: divide
D:\main.c(4) : info C5001: loop vectorized

--- Analyzing function: main
D:\main.c(4) : info C5001: loop vectorized
D:\main.c(16) : info C5002: loop not vectorized due to reason '1200'

This means:

• The loop in divide was successfully vectorized.
• The loop that prints the array was not vectorized. The reason code 1200 means that the loop has data dependencies between iterations that block vectorization.