Rgb to grayscale conversion with arm neon

I´m trying to convert from rgb to grayscale efficiently, so I got a function from here where it explains how to convert from rgba to grayscale. Now I´m trying to do the same but with just rgb. I changed some things but it doesn´t seem to work well. I don´t know why, does anyone see my mistake?

void neon_asm_convert(uint8_t * __restrict dest, uint8_t * __restrict src, int numPixels)
{
    __asm__ volatile(
     "lsr %2, %2, #3 \n"
     "# build the three constants:  \n"
     "mov r4, #28                   \n" // Blue channel multiplier
     "mov r5, #151                  \n" // Green channel multiplier
     "mov r6, #77                   \n" // Red channel multiplier
     "vdup.8 d4, r4                 \n"
     "vdup.8 d5, r5                 \n"
     "vdup.8 d6, r6                 \n"
     "0: \n"
     "# load 8 pixels: \n"  //RGBR
     "vld4.8 {d0-d3}, [%1]! \n"
     "# do the weight average: \n"
     "vmull.u8 q7, d0, d4 \n"
     "vmlal.u8 q7, d1, d5 \n"
     "vmlal.u8 q7, d2, d6 \n"
     "# shift and store: \n"
     "vshrn.u16 d7, q7, #8 \n" // Divide q3 by 256 and store in the d7
     "vst1.8 {d7}, [%0]! \n"
     "subs %2, %2, #1 \n" // Decrement iteration count

     "# load 8 pixels: \n"
     "vld4.8 {d8-d11}, [%1]! \n" //Other GBRG
     "# do the weight average: \n"
     "vmull.u8 q7, d3, d4 \n"
     "vmlal.u8 q7, d8, d5 \n"
     "vmlal.u8 q7, d9, d6 \n"
     "# shift and store: \n"
     "vshrn.u16 d7, q7, #8 \n" // Divide q3 by 256 and store in the d7
     "vst1.8 {d7}, [%0]! \n"
     "subs %2, %2, #1 \n" // Decrement iteration count

     "# load 8 pixels: \n"
     "vld4.8 {d0-d3}, [%1]! \n"
     "# do the weight average: \n"
     "vmull.u8 q7, d10, d4 \n"
     "vmlal.u8 q7, d11, d5 \n"
     "vmlal.u8 q7, d0, d6 \n"
     "# shift and store: \n"
     "vshrn.u16 d7, q7, #8 \n" // Divide q3 by 256 and store in the d7
     "vst1.8 {d7}, [%0]! \n"
     "subs %2, %2, #1 \n" // Decrement iteration count


     "# do the weight average: \n"
     "vmull.u8 q7, d1, d4 \n"
     "vmlal.u8 q7, d2, d5 \n"
     "vmlal.u8 q7, d3, d6 \n"
     "# shift and store: \n"
     "vshrn.u16 d7, q7, #8 \n" // Divide q3 by 256 and store in the d7
     "vst1.8 {d7}, [%0]! \n"

     "subs %2, %2, #1 \n" // Decrement iteration count



     "bne 0b \n" // Repeat unil iteration count is not zero
     :
     : "r"(dest), "r"(src), "r"(numPixels)
     : "r4", "r5", "r6"
    );
}

Solution

You load four values (RGBA) instead of 3 (RGB).

you have RGB RGB RGB in your image, but you load RGBR GBRG B... etc in consecutive steps.

"vld4.8 {d0-d3}, [%1]! \n"

Instead you should

"vld3.8 {d0-d2}, [%1]! \n"

Note that I have no idea if my asm is correct, but here is the mistake. Also check for the same mistake when moving the pixels back to memory