[PATCH] D148674: ValueTracking: Handle frem in computeKnownFPClass

[Kevin P. Neal via Phabricator via llvm-commits]

kpn added inline comments.


================
Comment at: llvm/lib/Analysis/ValueTracking.cpp:4680
+          KnownRHS.isKnownNeverLogicalZero(*F, Op->getType())) {
+        Known.knownNot(fcNan);
+      }
----------------
754-2019 section 6.1 says "remainder(subnormal, Inf) signals underflow." Does that mean it produces a nan or a zero? I'm guessing zero, but don't hold me to that. Anybody?

Section 7.2 agrees with your comment and code here.


CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D148674/new/

https://reviews.llvm.org/D148674