Groups | Search | Server Info | Login | Register

Groups > comp.arch.arithmetic > #65

Re: Napier's Bones versus Booth Encoding

Show all headers | View raw

Quadibloc wrote:
> 128K bytes of memory will contain a 256 by 256 multiplication table,
> and FF times FF is FE01.
>
> Thus, with a table lookup for every 8 bits of the multiplicand, two
> input terms can be generated for *eight* bits of the multiplier,
> which *doubles* the efficiency achieved with Booth encoding!
>
> And it's not as if you need an associative memory either, the two
> numbers being multiplied just index into the table.
>
> I am not so presumptuous as to think that this is a *new* idea on my
> part; no doubt others have thought of it, and it has a name, but I
> haven't noticed anyone mentioning it.

Name: Table lookup?

I bet there was a bunch of old machines without HW multipliers which did 
just this, probably based on bcd digits, i.e. 160 bytes (with 60 empty 
slots) give you the regular multiplication table.

I don't see where Booth give you any additional gains though, please 
explain!

Terje

-- 
- <Terje.Mathisen at tmsw.no>
"almost all programming can be viewed as an exercise in caching"

Back to comp.arch.arithmetic | Previous | Next — Previous in thread | Next in thread | Find similar

Thread

Napier's Bones versus Booth Encoding Quadibloc <jsavard@ecn.ab.ca> - 2014-06-15 07:00 -0700
  Re: Napier's Bones versus Booth Encoding Terje Mathisen <terje.mathisen@tmsw.no> - 2014-06-15 16:30 +0200
    Re: Napier's Bones versus Booth Encoding Quadibloc <jsavard@ecn.ab.ca> - 2014-06-15 09:28 -0700

csiph-web