Re: tech snippets

Eugene Leitl (
Thu, 19 Dec 1996 17:09:15 +0100 (MET)

On Mon, 16 Dec 1996, James Rogers wrote:

> >It seems, MMX multimedia instructions performance is distinctly inferior
> >to vanilla PowerPC.
> >
> I agree. I think MMX is an excuse to have mediocre floating point
> performance. I have always had reservations about specialty, high-level

Very few vanilla applications actually require floats, actually. Some new
games do, but this is sloppy programming: fixed-point or scaled integers,
especally xxl integers can substitute any float. Even scientific number
crunching can be done without floats.

FPUs make dies much larger, and hence more costly. Otoh, Intel price
politics has very little with production costs... My next CPU (I am
giving back my lent SNI (Ob)Scenic 120 MHz, since it sucks so
insufferably ) will be the new AMD, because it has excellent integer
performance and superior power management (if running a proper OS, that
is). Moreover, it is _much_ cheaper than an equivalent Pentium, at slightly
better overall performance (besides floats).

Don't support monopolies, you always have to pay for that, either in cash
or in an other form, sooner or later. Trouble is, you don't even notice
ruined marketplaces/architectures, because you can't examine an
alternative present, where this particular monopoly does not exist. It is
extremely difficult to argument with customers using alternative courses
of development "See, you could have XY for XZ $$'s just now, if only...".

> instructions on a CPU. But then again, Intel has been doing this since the
> i860 (which they are finally retiring). PowerPC has a solid floating point

Monopolies are not interested in serving their customers. Why should
they, since there is virtually zero competition (by definition)?

> architecture borrowed from IBMs Power 2 series, which also offered superb
> floating point performance.

Both PentiumPro and PowerPC, especially in cheap MSP machines are serious
candidates to obsoletify workstations in number crunching applications.
After mainframe has sunk below the surface of the tar pit, workstation's
death knell is beginning to toll, already.

> I think the P6 is Intel's first x86 CPU with adequate (although modest)
> floating point performance. If I was going to do serious floating point
> number crunching though, I would definitely go with PA-RISC, Alpha, or
> PowerPC architectures. I think as Intel's architectures get faster, they

Are you sure you get the most bang for the $$? PA-RISC is horrendously
expensive, as are the better Alphas. Don't know much about PowerPC,
though, but then there are not many vendors shipping them (Apple is
specializing in crippled architectures, as usual).

> are going to find it harder and harder to sell their chips if they don't
> seriously upgrade their floating point hardware. Fabulous integer
> performance isn't enough for most high-end, computation heavy software,
> especially in the realm of multimedia where you have a lot of realtime
> signal processing and transforms.

The more reason for doing it in 128 bit integers, and with a maspar
pipeline (one CPU for each processing stage).


| |transhumanism >H, cryonics, |
| |nanotechnology, etc. etc. |
| |"deus ex machina, v.0.0.alpha" |
|icbmto:N 48 10'07'' E 011 33'53'' | |