[NSAN] wrong value in "hex" for the float128
Open
compiler-rt:nsan
- Dominant language
- LLVM
- Stars
- 40.5k
- Forks
- 18.7k
- PR merge metrics
- PR metrics pending
Description
example
WARNING: NumericalStabilitySanitizer: inconsistent shadow results while checking call argument #0
double precision (native): dec: -0.00013593408999037200 hex: -0x1.1d130f11c72c70000000p-13
__float128 precision (shadow): dec: -0.00013593438856027846 hex:-0x8.e899c0d62e8031700000p-16
shadow truncated to double : dec: -0.00013593438856027845 hex: -0x1.1d13381ac5d000000000p-13
Relative error: 0.00021964265968929818% (2^33 epsilons)
it seems to me that "hex" value in the second line is wrong (it should start with the very same digits of the other two as the decimal values do)
Contributor guide
Assessment
This issue has not been assessed yet.