[Libre-soc-isa] [Bug 817] Big Integer Math (sv.adde, sv.subfe, sv.madded, 128 by 64-bit -> 64-bit div/rem, maybe more...)

[bugzilla-daemon at libre-soc.org]

https://bugs.libre-soc.org/show_bug.cgi?id=817

--- Comment #8 from Jacob Lifshay <programmerjake at gmail.com> ---
I implemented a WIP version of goldschmidt division, for the 128x64->64-bit
division instructions. I didn't yet figure out the correct values for the
additional internal precision needed, but I found a paper describing how to
calculate them ... they showed that the parameters used for AMD's K8 were too
loose, meaning the K8 has more internal precision than needed, wasting gates.

https://git.libre-soc.org/?p=soc.git;a=commitdiff;h=cc2c51cf2cd7b22f7d18871fbfb73e0b7707d91a

The paper I found:
Even, G., Seidel, P. M., & Ferguson, W. E. (2003).
A Parametric Error Analysis of Goldschmidt's Division Algorithm.
https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.90.1238&rep=rep1&type=pdf

-- 
You are receiving this mail because:
You are on the CC list for the bug.