If you are interested in CEPHES, I have published a new version of it.
See:
https://github.com/jacob-navia/qlib.git

This is a first batch of code I will publish. The lcc source code will follow.

jacob

On 11/3/2022 12:56 PM, Jacob Navia wrote:
> If you are interested in CEPHES, I have published a new version of it.
> See:
> https://github.com/jacob-navia/qlib.git
>
> This is a first batch of code I will publish. The lcc source code will follow.

typedef struct _float128_t{
unsigned long long low, high;
} float128_t;

Reminds me of an old piece of code from an old friend:

https://groups.google.com/g/comp.lang.asm.x86/c/FScbTaQEYLc/m/2lbw55j9UYgJ

Le jeudi 3 novembre 2022 à 21:08:05 UTC+1, Chris M. Thomasson a écrit :
> On 11/3/2022 12:56 PM, Jacob Navia wrote:
> > If you are interested in CEPHES, I have published a new version of it.
> > See:
> > https://github.com/jacob-navia/qlib.git
> >
> > This is a first batch of code I will publish. The lcc source code will follow.
> typedef struct _float128_t{
> unsigned long long low, high;
> } float128_t;
>
> Reminds me of an old piece of code from an old friend:
>
> https://groups.google.com/g/comp.lang.asm.x86/c/FScbTaQEYLc/m/2lbw55j9UYgJ

Well, representing 128 bit floats as two 64 bit numbers must have been used by ... 99% of the people that wrote code to handle those numbers!
But the version of t he CEPHES library I am presenting uses 8 BYTE numbers (mantissa of 448 bits). The 128 bit numbers are used to seed the newton iteration of some routines. Instead of starting with a seed of 64 bit, it is better to start with a seed of 128 ones, then pass to 448 bit arithmetic when needed