• State Accepted Answer
  • Locked Locked
  • Replies 8 replies
  • Subscribers 136 subscribers
  • Views 8295 views
  • Users 0 members are here
Options
Related
How was your experience today?
This discussion has been locked.
You can no longer post new replies to this discussion. If you have a question you can start a new discussion

NEON intrinsics mixed Native C

BANGYONG SONG

Hello

I tested my source code used NEON intrinsics.

I wanted to compare the performance between using NEON and using Native C.

the code doesn't have meaningful and simple.

get an image from storage and each pixels shift to right by 2.

and get an output image using 256 x 256 LUT.

========================

for loop

uint8x8x3_t rgb = vld3_u8(img_data[i * width + j]);

// shift to right by 2

rgb.val[0] = vshr_n_u8(rgb.val[0], 2);

rgb.val[1] = vshr_n_u8(rgb.val[1], 2);

rgb.val[2] = vshr_n_u8(rgb.val[2], 2);

// get the data from LUT

int index = 0;

int p = i * width + j;

uint8_t r, g;

// lane 0

r = vget_lane_u8(rgb.val[0], 0);

g = vget_lane_u8(rgb.val[1], 0);

b = vget_lane_u8(rgb.val[2], 0);

img_data[p + 0] = LUT[r][g];

img_data[p + 1] = LUT[g][b];

img_data[p + 2] = LUT[b][r];

// lane 1

r = vget_lane_u8(rgb.val[0], 1);

g = vget_lane_u8(rgb.val[1], 1);

b = vget_lane_u8(rgb.val[2], 1);

img_data[p + 3] = LUT[r][g];

img_data[p + 4] = LUT[g][b];

img_data[p + 5] = LUT[b][r];

// lane 2

...

// lane 8

========================

and I compared the time between using Native C and using NEON intrinsics.

Native C is faster than NEON about 15%

I want to know why it was.

Parents
  • 0 in reply to BANGYONG SONG

    Hello,


    I think the optimization should be left to the compiler.
    The table look ahead would be compiled as the following.

    #  r0  r
#  r1  g
#  r2  b
#  r5  p
#
# for lane 0
  movw r3,#:lower16:LUT
  movt r3,#:upper16:LUT
  add r4,r1,r0, lsl #8  @ index [r][g]
  add r1,r2,r1, lsl #8  @ index [g][b]
  add r0,r0,r2, lsl #8  @ index [b][r]
  ldrb r4,[r3,r4]      @ LUT[r][g]
  ldrb r1,[r3,r1]      @ LUT[g][b]
  ldrb r0,[r3,r2]      @ LUT[b][r]
  movw r2,#:lower16:img_data
  movt r2,#:upper16:img_data
  add  r2,r2,r5    @ addr img_data[p]
  strb r4,[r2,#0]      @ img_data[p+0]
  strb r1,[r2,#1]      @ img_data[p+1]
  strb r0,[r2,#2]      @ img_data[p+2]

    That is, the reference would be done via pointers.
    Those codes are generated by my hand and they are not faster.
    To improve them, the order of instructions should be re-arranged to resolve register dependencies.


    Best regards,
    Yasuhiko Koumoto.

Reply
  • 0 in reply to BANGYONG SONG

    Hello,


    I think the optimization should be left to the compiler.
    The table look ahead would be compiled as the following.

    #  r0  r
#  r1  g
#  r2  b
#  r5  p
#
# for lane 0
  movw r3,#:lower16:LUT
  movt r3,#:upper16:LUT
  add r4,r1,r0, lsl #8  @ index [r][g]
  add r1,r2,r1, lsl #8  @ index [g][b]
  add r0,r0,r2, lsl #8  @ index [b][r]
  ldrb r4,[r3,r4]      @ LUT[r][g]
  ldrb r1,[r3,r1]      @ LUT[g][b]
  ldrb r0,[r3,r2]      @ LUT[b][r]
  movw r2,#:lower16:img_data
  movt r2,#:upper16:img_data
  add  r2,r2,r5    @ addr img_data[p]
  strb r4,[r2,#0]      @ img_data[p+0]
  strb r1,[r2,#1]      @ img_data[p+1]
  strb r0,[r2,#2]      @ img_data[p+2]

    That is, the reference would be done via pointers.
    Those codes are generated by my hand and they are not faster.
    To improve them, the order of instructions should be re-arranged to resolve register dependencies.


    Best regards,
    Yasuhiko Koumoto.

Children