Updating the Linear Frame Buffer much slower on real hardware than QEMU

Question

I am currently booting my system into a 640x480x32b VESA mode with a linear frame buffer.

Using the following code, I flash red and green on the screen:

void clear(Color color) {
    u8* cursor = frame_buffer;
    int L = VIDEO_HEIGHT * VIDEO_WIDTH;

    for (int i = 0; i < L; i++) {
        *(cursor) = color.b;
        *(cursor + 1) = color.g;
        *(cursor + 2) = color.r;
        cursor = cursor + 4;
    }
}

void update_screen() {
    int L = VIDEO_HEIGHT * VIDEO_WIDTH * 4;
    
    for (int i = 0; i < L; i++) {
        *(LFB + i) = *(frame_buffer + i);
    }
}

Where LFB is the linear frame buffer used for rendering and frame_buffer is a secondary buffer.

void main() {
    init_video();

    while (1) {
        clear((Color) {255, 0, 0});
        update_screen();
        clear((Color) {0, 255, 0});
        update_screen();
    }
}

This works very well in QEMU, updating the entire screen nigh-instantly. However, when ran on real hardware, it takes nearly half a second, i.e. at least 10x slower than QEMU (probably much more than 10x, although I don't have an accurate way to measure this).

Any reason for why update_screen is so slow on the real hardware?