https://gcc.gnu.org/bugzilla/show_bug.cgi?id=82004
Bug ID: 82004
Summary: SPEC CPU2017 628.pop2_s miscompare
Product: gcc
Version: 8.0
Status: UNCONFIRMED
Severity: normal
Priority: P3
Component: middle-end
Assignee: unassigned at gcc dot gnu.org
Reporter: andrey.y.guskov at intel dot com
Target Milestone: ---
I`m seeing a miscompare in 628.pop2_s compiled with the following option set:
-m64 -march=core-avx2 -mfpmath=sse -Ofast -funroll-loops -fconvert=big-endian
-fopenmp
The guilty revision is r251230. Despite exp(log(c)*x) being faster, it seems to
lack the precision of pow(c,x).
