Hi @sangameshnr,
If I am correct, you get out of the allocated memory for the image in the lines provided below (for the case of tmp[12:4] =data[(i+3)*ldi+j:4]. This goes for the images that have odd dimensions only. So the images need to have even dimensions or be padded with zeros.
|
for(i = 0; i < irows-2; i += 2){ |
|
#pragma unroll(4) |
|
for(j = 0; j < (irows-2); j += 2){ |
|
tmp[0 :4] =data[(i+0)*ldi+j:4]; |
|
tmp[4 :4] =data[(i+1)*ldi+j:4]; |
|
tmp[8 :4] =data[(i+2)*ldi+j:4]; |
|
tmp[12:4] =data[(i+3)*ldi+j:4]; |
Hi @sangameshnr,
If I am correct, you get out of the allocated memory for the image in the lines provided below (for the case of tmp[12:4] =data[(i+3)*ldi+j:4]. This goes for the images that have odd dimensions only. So the images need to have even dimensions or be padded with zeros.
FALCON/src/falcon.c
Lines 58 to 64 in d2552c8