• Locked Locked
  • Replies 2 replies
  • Subscribers 119 subscribers
  • Views 3646 views
  • Users 0 members are here
2025 survey

We are running a survey to help us improve the experience for all of our members. If you see the survey appear, please take the time to tell us about your experience if you can.

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

Optimising with NEON

Roumen Popov
Note: This was originally posted on 11th February 2010 at http://forums.arm.com

Hello guys, I am a bit new with ARM and NEON and was wondering if anyone could help me with advice on how to optimise a small piece of code by NEON.
The problem is that whatever I do I can't get the NEON code to execute faster than the standard C code.
I am running the code on Beagle board with the latest Angstrom distribution and building with the Code Sourcery Lite gcc.
My gcc command line options are  -O3 -mcpu=cortex-a8 -mfpu=neon -mfloat-abi=softfp -ffast-math
the piece of code in question is the following:

int32_t weightsH[4];
int16_t weightsV[4];
int16_t *lut1, *lut2;

// start of loop

quantLevel1 = ...
quantLevel2 = ...
frac = ...
frac_1 = ...

pxlOut      = 0;

lut1 = luts[quantLevel1];
lut2 = luts[quantLevel2];

w1 = lut1[0];
w2 = lut2[0];
weight = ((weightsH[0]>>16)*weightsV[0])>>13;
pxlOut += (weight*((w1*frac + w2*frac_1)>>14))>>14;

w1 = lut1[1];
w2 = lut2[1];
weight = ((weightsH[1]>>16)*weightsV[0])>>13;
pxlOut += (weight*((w1*frac + w2*frac_1)>>14))>>14;

w1 = lut1[2];
w2 = lut2[2];
weight = ((weightsH[0]>>16)*weightsV[1])>>13;
pxlOut += (weight*((w1*frac + w2*frac_1)>>14))>>14;

w1 = lut1[3];
w2 = lut2[3];
weight = ((weightsH[1]>>16)*weightsV[1])>>13;
pxlOut += (weight*((w1*frac + w2*frac_1)>>14))>>14;

weightsH[0] -= slope_x;
weightsH[1] += slope_x;

// end of loop

the NEON code with the best results that I managed to obtain is the following, yet it's still slower than the above, it's just killing me:

int16x4_t wVs, lt1, lt2, fr1, w;
int32x4_t wHs,s;
int32x4_t buf;
int32x2_t buf2;

slopes[0] = -slope_x;
slopes[1] = slope_x;
slopes[2] = -slope_x;
slopes[3] = slope_x;

wHs = vld1q_s32(weightsH);
s = vld1q_s32(slopes);
wVs = vld1_s16(weightsV);

// start of loop

quantLevel1 = ...
quantLevel2 = ...
frac = ...
frac_1 = ...

lut1 = luts[quantLevel1];
lut2 = luts[quantLevel2];

buf = vdupq_n_s32(0);
lt1 = vld1_s16(lut1);
lt2 = vld1_s16(lut2);
buf = vmlal_n_s16(buf, lt1, (int16_t)frac);
buf = vmlal_n_s16(buf, lt2, (int16_t)frac_1);
fr1 = vshrn_n_s32(buf, 14);
w = vshrn_n_s32(wHs, 16);
buf = vmull_s16(w, wVs);
w = vshrn_n_s32(buf, 13);
buf = vmull_s16(w, fr1);
fr1 = vshrn_n_s32(buf, 14);
buf2 = vpaddl_s16(fr1);
pxlOut = vget_lane_s32(buf2, 0) + vget_lane_s32(buf2, 1);

// end of loop

Thank you in advance,
Roumen
Parents
  • Note: This was originally posted on 15th February 2010 at http://forums.arm.com

    hi sim, and thanks for your prompt reply,
    i was thinking the same, that gcc might be already optimising through NEON the standard C code,
    however the assembly listing produced by the compiler shows that it doesn't, it produces the standard assembly instructions,
    which may mean that that piece of code can not be optimised through NEON to start with.
    I am beginning to get the impression that to get improvement using NEON, the code should map very well
    to NEON's instruction set, e.g. the combined instructions such as the multiply accumulate, and limit movement of data between
    NEON's registers and standard registers/memory as much as possible, but i am not sure because i don't have much experience with NEON.
    What is your opinion about this?

    thanks again and best regards,
    roumen
Reply
  • Note: This was originally posted on 15th February 2010 at http://forums.arm.com

    hi sim, and thanks for your prompt reply,
    i was thinking the same, that gcc might be already optimising through NEON the standard C code,
    however the assembly listing produced by the compiler shows that it doesn't, it produces the standard assembly instructions,
    which may mean that that piece of code can not be optimised through NEON to start with.
    I am beginning to get the impression that to get improvement using NEON, the code should map very well
    to NEON's instruction set, e.g. the combined instructions such as the multiply accumulate, and limit movement of data between
    NEON's registers and standard registers/memory as much as possible, but i am not sure because i don't have much experience with NEON.
    What is your opinion about this?

    thanks again and best regards,
    roumen
Children
No data