https://gcc.gnu.org/bugzilla/show_bug.cgi?id=61171
Bug ID: 61171
Summary: vectorization fails for a reduction in presence of
subtraction
Product: gcc
Version: 4.9.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: tree-optimization
Assignee: unassigned at gcc dot gnu.org
Reporter: vincenzo.innocente at cern dot ch
give this code
cat bug.cc
float px[1024];
float xx, vv;
unsigned int N=1024;
void ok() {
for (auto j=0U; j<N; ++j) {
auto ax = px[j]-xx;
vv-=ax;
}
}
void noOk() {
for (auto j=0U; j<N; ++j) {
auto ax = xx-px[j];
vv+=ax;
}
}
c++ -std=c++11 -Ofast -S bug.cc -march=haswell -fopt-info-vec-missed; cat
bug.s
ok vectorize, noOk does not because
bug.cc:13:3: note: reduction: not commutative/associative: vv.6_9 = _2 - _6;
even if the code is clearly identical (at least with Ofast)
