There is a that QD project:
https://www.davidhbailey.com/dhbsoftware/
Lisence is:
https://www.davidhbailey.com/dhbsoftware/LBNL-BSD-License.docx
In attachment is a limited version (DD) of it in C-language.
Just fix it and make an assembler version of it, but follow a license.
PS.
I don't have to be involved in this project any more.
Interesting stuff for folks who have a use for high precision maths. We have a member here Ray Filiatrault who has a background in maths and I remember long ago Ray mentioning a technique of fixed point maths that could do unusual things.
if anyone is interested in double-double or quad-double precision using floating-point arithmetic Google gives a lot of hits, here's a PDF describing the algorithms http://web.mit.edu/tabbott/Public/quaddouble-debian/qd-2.3.4-old/docs/qd.pdf and https://portal.nersc.gov/project/sparse/xiaoye-web/arith15.pdf
Quote from: hutch-- on December 03, 2020, 12:05:05 PM
Interesting stuff for folks who have a use for high precision maths. We have a member here Ray Filiatrault who has a background in maths and I remember long ago Ray mentioning a technique of fixed point maths that could do unusual things.
as long as masm64 sdk started,I have wanted to try out if 64bit version fixed point and time it vs real8's
but I prefer the real4's and real8's flexibility and together with SIMD its fast
scalar it seems like its going to be a race between conversion between float/ints using floats vs extra bitshifting when using fixed points
As asmc support 128 bit, a library with it could be usable.