https://gcc.gnu.org/bugzilla/show_bug.cgi?id=119209
Bug ID: 119209
Summary: SLP failed to recognize dot_prod pattern(it's taked as
a normal reduction)
Product: gcc
Version: 15.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: liuhongt at gcc dot gnu.org
Target Milestone: ---
int
foo (unsigned char* a, char* b, int n, int stride, int* __restrict dst)
{
int sum = 0;
sum += a[0] * b[0];
sum += a[1] * b[1];
sum += a[2] * b[2];
sum += a[3] * b[3];
sum += a[4] * b[4];
sum += a[5] * b[5];
sum += a[6] * b[6];
sum += a[7] * b[7];
return sum;
}
vect__36.5_116 = MEM <vector(8) unsigned char> [(unsigned char *)a_42(D)];
vect_patt_107.6_117 = (vector(8) unsigned short) vect__36.5_116;
# DEBUG D#40 => *a_42(D)
# DEBUG D#39 => (int) D#40
vect__38.9_119 = MEM <vector(8) char> [(char *)b_43(D)];
vect_patt_109.10_120 = (vector(8) signed short) vect__38.9_119;
vect_patt_111.11_121 = VIEW_CONVERT_EXPR<vector(8) unsigned
short>(vect_patt_109.10_120);
vect_patt_112.12_122 = vect_patt_107.6_117 * vect_patt_111.11_121;
vect_patt_113.13_123 = VIEW_CONVERT_EXPR<vector(8) signed
short>(vect_patt_112.12_122);
vect_patt_114.14_124 = (vector(8) int) vect_patt_113.13_123;
...
_125 = VIEW_CONVERT_EXPR<vector(8) unsigned int>(vect_patt_114.14_124);
_126 = .REDUC_PLUS (_125); [tail call]
_127 = (int) _126;
For O3, inner loop with simple dot_prod reduction will mostly completely
unrolled and outer loop relies on SLP discovery for the vectorization can
didn't do good job here.
