[libre-riscv-dev] new update written for NLnet grant and progress

[Jacob Lifshay]

On Mon, May 27, 2019 at 9:09 PM Luke Kenneth Casson Leighton
<lkcl at lkcl.net> wrote:
>
> On Tue, May 28, 2019 at 4:43 AM Jacob Lifshay <programmerjake at gmail.com> wrote:
> >
> > When the multiplier is used in the intended way, it will have 2x
> > Mul8_16_32_64 for a total of 128-bits per argument/result bus, so will
> > support 2 64x64 multiplications per cycle. Supporting 64-bit
> > multiplications requires approximately 1.5x the number of gates when
> > compared to the 128-bit wide 32/16/8-bit multiplier.
>
>  ok i did a sort-of summary version of that.
I think having almost the same long sentence twice may not be the best.
It might be better as follows, since that way the exact number is not
specified, but the ratios are:

... handle a batch of 8-bit operations, or half the number of 16-bit
operations, or a quarter of the number of 32-bit operations or just
one-eighth of the number of 64-bit operations. This way, a lot less
gates are required, compared to having several separate units.
The unit tests demonstrate that the code that Jacob has written provide ...

Jacob