The dcbrt and dtanh and implementations (see below) currently use decimal literals within constant definitions, but it makes more sense to use hex literals to represent floating point values. Also, the other math intrinsics use hex literals.
1. src/hotspot/cpu/x86/stubGenerator_x86_64_cbrt.cpp
2. src/hotspot/cpu/x86/stubGenerator_x86_64_tanh.cpp
1. src/hotspot/cpu/x86/stubGenerator_x86_64_cbrt.cpp
2. src/hotspot/cpu/x86/stubGenerator_x86_64_tanh.cpp
- caused by
-
JDK-8338694 x86_64 intrinsic for tanh using libm
-
- Resolved
-