Skip to content

Difference (regression?) in floating point pow on 1.8 #44525

Closed

Description

Looking at some PkgEval logs a failure that occurs comes up from a changed result for 2^111.94. This used to be computed as

julia> v = 2^111.94
4.9807840732777406e33

which is in agreement with e.g. Python. The current value is

julia> 2^111.94
4.980784073277717e33

which to my naive eye seems to be significantly worse:

julia> v = 2^111.94; bv = big"2"^111.94; (v - bv) / bv # 1.7
2.09394118307...e-17

julia> v = 2^111.94; bv = big"2"^111.94; (v - bv) / bv # 1.8
-4.72427549745...e-15
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

Labels

No labels
No labels

Type

No type

Projects

No projects

Milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions