this to be done using an adaptation (to Base 4) of the binary method: https://en.wikipedia.org/wiki/Methods_of_computing_square_roots#Binary_numeral_system_(base_2) adaptation to base 4 is 2-bits at a time. this requires only 3 extra comparators and one extra adder (the adder being used to calculate a multiply by 3). multiply by 0 is clearly 0. multiply by 1 is clearly the original. multiply by 2 is a fixed-width shift. multiply by 3 is an add (of number shifted added to original number).
http://lists.libre-riscv.org/pipermail/libre-riscv-dev/2019-March/000926.html
*** Bug 56 has been marked as a duplicate of this bug. ***
Plan for thr further FPU development: - create sqrt function - create test for sqrt function - run the tests for the whole FPU
what i recommend instead is to create a milestone "preliminary exploratory software emulation of FP SQRT" milestone. context: http://lists.libre-riscv.org/pipermail/libre-riscv-dev/2019-April/001221.html
(In reply to Aleksandar Kostovic from comment #3) > Plan for thr further FPU development: > > - create sqrt function > - create test for sqrt function > - run the tests for the whole FPU so if you approve the idea "create preliminary software-only exploratory version" can you please create a new bugreport, and set its "Blocks" field to 43 (this bug)? also remember to "take ownership" of the bug, and to set the NLnet milestone field. i need you to do more of this because i'm getting RSI and it's now very painful to type.
> > paper) into python, which we will *need anyway as part of the unit > > tests*. > > > Yes i will convert that code to python. great. look fwd to seeing it. always wanted to impl.
Hi Aleksander I moved fsqrt.py to its own subdir, it is still the same filename so use find . -name fsqrt.py ok?
>Hi Aleksander I moved fsqrt.py to its own >subdir, it is still the same filename so use >find . -name fsqrt.py ok? Thats fine. Will check it out