__local sumT* sum_p;
src_step = src_step >> 2;
gid = gid << 1;
- for (int i = 0; i < rows; i =i + LSIZE_1)
+ int lid_prim = ((lid & 127) << 1) + 1;
+ for (int i = 0; i < rows; i += LSIZE_1)
{
- src_t[0] = (i + lid < rows ? convertToSum4(src[mad24((lid+i), src_step, src_offset + gid)]) : (vecSumT)0);
- src_t[1] = (i + lid < rows ? convertToSum4(src[mad24((lid+i), src_step, src_offset + gid + 1)]) : (vecSumT)0);
+ if (i + lid < rows)
+ {
+ int src_index = mad24((lid+i), src_step, gid + src_offset);
+ src_t[0] = convertToSum4(src[src_index]);
+ src_t[1] = convertToSum4(src[src_index + 1]);
+ }
+ else
+ {
+ src_t[0] = (vecSumT)0;
+ src_t[1] = (vecSumT)0;
+ }
- sum_t[0] = (i == 0 ? (vecSumT)0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
- sum_t[1] = (i == 0 ? (vecSumT)0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
+ if (i == 0)
+ {
+ sum_t[0] = (vecSumT)0;
+ sum_t[1] = (vecSumT)0;
+ }
+ else
+ {
+ sum_t[0] = lm_sum[0][LSIZE_2 + LOG_LSIZE];
+ sum_t[1] = lm_sum[1][LSIZE_2 + LOG_LSIZE];
+ }
barrier(CLK_LOCAL_MEM_FENCE);
int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
- lm_sum[0][bf_loc] = src_t[0];
+ lm_sum[0][bf_loc] = src_t[0];
lm_sum[1][bf_loc] = src_t[1];
int offset = 1;
for (int d = LSIZE >> 1 ; d > 0; d>>=1)
{
barrier(CLK_LOCAL_MEM_FENCE);
- int ai = offset * (((lid & 127)<<1) +1) - 1,bi = ai + offset;
+ int ai = offset * lid_prim - 1,bi = ai + offset;
ai += GET_CONFLICT_OFFSET(ai);
bi += GET_CONFLICT_OFFSET(bi);
{
barrier(CLK_LOCAL_MEM_FENCE);
offset >>= 1;
- int ai = offset * (((lid & 127)<<1) +1) - 1,bi = ai + offset;
+ int ai = offset * lid_prim - 1,bi = ai + offset;
ai += GET_CONFLICT_OFFSET(ai);
bi += GET_CONFLICT_OFFSET(bi);
sum_p = (__local sumT*)(&(lm_sum[0][bf_loc]));
for (int k = 0; k < 4; k++)
{
- if(gid * 4 + k >= cols) continue;
+ if (gid * 4 + k >= cols)
+ break;
sum[loc_s0 + k * dst_step / 4] = sum_p[k];
}
sum_p = (__local sumT*)(&(lm_sum[1][bf_loc]));
for (int k = 0; k < 4; k++)
{
- if(gid * 4 + k + 4 >= cols) break;
+ if (gid * 4 + k + 4 >= cols)
+ break;
sum[loc_s1 + k * dst_step / 4] = sum_p[k];
}
}
}
}
+
kernel void integral_sum_rows(__global uchar *srcsum_ptr, __global uchar *sum_ptr,
int rows, int cols, int src_step, int sum_step, int sum_offset)
{
__local vecSumT lm_sum[2][LSIZE + LOG_LSIZE];
__local sumT *sum_p;
src_step = src_step >> 4;
- for (int i = 0; i < rows; i =i + LSIZE_1)
+ int lid_prim = ((lid & 127) << 1) + 1;
+ for (int i = 0; i < rows; i += LSIZE_1)
{
- src_t[0] = i + lid < rows ? srcsum[mad24((lid+i), src_step, gid * 2)] : 0;
- src_t[1] = i + lid < rows ? srcsum[mad24((lid+i), src_step, gid * 2 + 1)] : 0;
-
- sum_t[0] = (i == 0 ? 0 : lm_sum[0][LSIZE_2 + LOG_LSIZE]);
- sum_t[1] = (i == 0 ? 0 : lm_sum[1][LSIZE_2 + LOG_LSIZE]);
+ if (i + lid < rows)
+ {
+ int sum_idx = mad24(lid + i, src_step, gid * 2);
+ src_t[0] = srcsum[sum_idx];
+ src_t[1] = srcsum[sum_idx + 1];
+ }
+ else
+ {
+ src_t[0] = 0;
+ src_t[1] = 0;
+ }
+ if (i == 0)
+ {
+ sum_t[0] = 0;
+ sum_t[1] = 0;
+ }
+ else
+ {
+ sum_t[0] = lm_sum[0][LSIZE_2 + LOG_LSIZE];
+ sum_t[1] = lm_sum[1][LSIZE_2 + LOG_LSIZE];
+ }
barrier(CLK_LOCAL_MEM_FENCE);
int bf_loc = lid + GET_CONFLICT_OFFSET(lid);
- lm_sum[0][bf_loc] = src_t[0];
+ lm_sum[0][bf_loc] = src_t[0];
lm_sum[1][bf_loc] = src_t[1];
int offset = 1;
for (int d = LSIZE >> 1 ; d > 0; d>>=1)
{
barrier(CLK_LOCAL_MEM_FENCE);
- int ai = offset * (((lid & 127)<<1) +1) - 1,bi = ai + offset;
+ int ai = offset * lid_prim - 1, bi = ai + offset;
ai += GET_CONFLICT_OFFSET(ai);
bi += GET_CONFLICT_OFFSET(bi);
{
barrier(CLK_LOCAL_MEM_FENCE);
offset >>= 1;
- int ai = offset * (((lid & 127)<<1) +1) - 1,bi = ai + offset;
+ int ai = offset * lid_prim - 1,bi = ai + offset;
ai += GET_CONFLICT_OFFSET(ai);
bi += GET_CONFLICT_OFFSET(bi);
- if((lid & 127) < d)
+ if ((lid & 127) < d)
{
lm_sum[lid >> 7][bi] += lm_sum[lid >> 7][ai];
lm_sum[lid >> 7][ai] = lm_sum[lid >> 7][bi] - lm_sum[lid >> 7][ai];
int loc0 = gid * 2 * sum_step;
for(int k = 1; k <= 8; k++)
{
- if(gid * 8 + k > cols) break;
+ if (gid * 8 + k > cols)
+ break;
sum[sum_offset + loc0 + k * sum_step / 4] = 0;
}
}
sum_p = (__local sumT*)(&(lm_sum[0][bf_loc]));
for(int k = 0; k < 4; k++)
{
- if(gid * 8 + k >= cols) break;
+ if (gid * 8 + k >= cols)
+ break;
sum[loc_s0 + k * sum_step / 4] = sum_p[k];
}
sum_p = (__local sumT*)(&(lm_sum[1][bf_loc]));
for(int k = 0; k < 4; k++)
{
- if(gid * 8 + 4 + k >= cols) break;
+ if (gid * 8 + 4 + k >= cols)
+ break;
sum[loc_s1 + k * sum_step / 4] = sum_p[k];
}
}