Re: World's Fastest Processor? For Now, Anyway

On Wed, 23 May 2007 16:03:31 -0700, Chip Coldwell <coldwell@xxxxxxxxxxxxxxxxx> wrote:

On Wed, 23 May 2007, Dan Foster wrote:

In article <66dfp8Idh60R@xxxxxxxxxxxxxxxxxxxxxxxx>, Larry Kilgallen <Kilgallen@xxxxxxxxxxx> wrote:
>>
>> What kind basis do they have for making this claim that it's the first
>> one to do it in hardware? Haven't there been a number of
>> microprocessors that did this long ago?
>
> Perhaps they were not _UNIX_ microprocessors, like this one is :-)

That's true; but the Motorola 68040 had built-in FP support in hardware
some 17 years ago.

Ah, but that was *binary* floating-point. What IBM is describing is
*decimal* floating-point.

Decimal floating-point is standard in pocket calculators, because user's
get annoyed when 0.1 (which has no exact binary representation) gets
rounded to the nearest binary floating-point equivalent.

IBM had decimal arithmetic in hardware in their mainframes in the '60s,
but that may have been fixed-point.

That was scaled fixed decimal. so you might, for example, declare in your
program something like

declare foo fixed decimal(34,15);

which would give you 19 digits left of the decimal point and 15 right

you are free to choose p and q in fixed decimal(p,q) subject to implementation
limits


Chip




--
Using Opera's revolutionary e-mail client: http://www.opera.com/mail/
.

Relevant Pages

  • Re: Naval Railgun: Barrel Design or Power Supply?
    ... Instead of a railgun, you should use a coilgun. ... induction motor" style launcher is contactless, ... railgun then there is less control of the output velocity. ...
    (sci.physics)
  • Re: Worlds Fastest Processor? For Now, Anyway
    ... but the Motorola 68040 had built-in FP support in hardware ... but that was *binary* floating-point. ... Decimal floating-point is standard in pocket calculators, ...
    (comp.os.vms)
  • Re: curious about 8 byte LOGICAL...
    ... fast 128-bit hardware impractical. ... much slower than 64-bit floating-point on the same machine. ... If Intel ever builds a 128-bit wide ALU, ... The SSE and AVX instructions are SIMD instructions. ...
    (comp.lang.fortran)
  • Re: Damn you, FEDEX! or Nikon D40 lost in Springfield, MO blackhole.
    ... the 2 mp Mavica he had been using with a Nikon D40. ... After shopping around, he got me to order one for him. ... The shipper had it insured, but from what I have read it could take weeks to sort this crap out. ... You may get your insurance from FedEx and a couple weeks later they find it and deliver it. ...
    (alt.photography)
  • Re: Gentler Decimal Floating-Point
    ... floating-point support. ... admit that in most computers binary floating-point is the only attempt ... As long as DFP is regarded as not speed-critical, ... hardware, at the cost of an extra digit of precision. ...
    (comp.arch.arithmetic)