Releases: milankl/SoftPosit.jl
Releases · milankl/SoftPosit.jl
v0.5.2
SoftPosit v0.5.2
Closed issues:
- Precompilation warnings (#76)
Merged pull requests:
- conversion from irrationals (#75) (@milankl)
- remove duplicate int conversions (#77) (@milankl)
- fix
sin
,cos
, andexp
(#78) (@oscardssmith) - avoid overflow on
exp
and friends. (#79) (@oscardssmith)
v0.5.1
v0.5.0
v0.4.0
SoftPosit v0.4.0
Closed issues:
Merged pull requests:
- Install TagBot as a GitHub Action (#54) (@JuliaTagBot)
- Conversion (#56) (@milankl)
- bitstring testing with :split (#57) (@milankl)
- speedup with implicit sign with 2's compl (#58) (@milankl)
- Float32(::Posit16) without C, no tests (#59) (@milankl)
- no overflow rounding mode for conversion (#60) (@milankl)
- make new conversion default (#61) (@milankl)
v0.3.0
v0.3.0 (2019-12-04)
Merged pull requests:
v0.3: Merge pull request #50 from milankl/floorceil
Contains
- Floor and ceil functions for Posit8, Posit16 and Posit32
Trigonometric functions and documentation added
SoftPosit.jl now supports some trigonometric functions like exp, log, sin, cos, tan.
A documentation is added in the form of an IPython notebook.
SoftPosit.jl v0.1
SoftPosit.jl supports Posit8, Posit16, Posit32 for basic arithmetic operations and are ready to be used in algorithms that do not require extended functionality (trigonometric functions, etc.)
Quire8, Quire16 and Quire32 are also supported.
Other posit formats Posit8_2, Posit16_2, Posit24_2 (and similar _1 with 1 exponent bit) are only partly supported due to dependencies of the underlying SoftPosit C-library.
Not yet supported are exp, log, sin, cos, tan, pow, etc.