diff options
| author | Aina Niemetz <aina.niemetz@gmail.com> | 2021-03-19 10:28:14 -0700 |
|---|---|---|
| committer | GitHub <noreply@github.com> | 2021-03-19 17:28:14 +0000 |
| commit | 58e219362b2e9a7d7b9b9b526760c392cd50e878 (patch) | |
| tree | edc5b6e8f4307942a914d13c218197328c8cb6bf /src/util/floatingpoint.cpp | |
| parent | 95f5be640fd362098a54e222d56d49144fe0efbe (diff) | |
BitVector: Change setBit to set the bit in place. (#6176)
Creating BitVectors (and deleting them) is in general expensive because
of the underlying multi-precision Integer. If possible, unnecessary
constructions and desctructions of BitVectors should be avoided.
The most common use case for `setBit` is that for an existing BitVector,
a given bit should be set to a certain value. Not doing this in place
generates unnecessary constructions and destructions of BitVectors.
Diffstat (limited to 'src/util/floatingpoint.cpp')
| -rw-r--r-- | src/util/floatingpoint.cpp | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/src/util/floatingpoint.cpp b/src/util/floatingpoint.cpp index 42f411ab2..cebdbbb29 100644 --- a/src/util/floatingpoint.cpp +++ b/src/util/floatingpoint.cpp @@ -361,8 +361,8 @@ FloatingPoint FloatingPoint::makeMaxNormal(const FloatingPointSize& size, bool sign) { BitVector bvsign = sign ? BitVector::mkOne(1) : BitVector::mkZero(1); - BitVector bvexp = - BitVector::mkOnes(size.packedExponentWidth()).setBit(0, false); + BitVector bvexp = BitVector::mkOnes(size.packedExponentWidth()); + bvexp.setBit(0, false); BitVector bvsig = BitVector::mkOnes(size.packedSignificandWidth()); return FloatingPoint(size, bvsign.concat(bvexp).concat(bvsig)); } |