NetNews Offline 2/NetNews Offline Volume 2.iso/news/comp/sys/amiga/misc-part1/9394

Internet Message Format | 1996-08-05 | 5.6 KB

Path: pn.itnet.it!not-for-mail
From: "Stefano Agostinelli" <[e-mail]>
Newsgroups: comp.sys.amiga.networking,comp.sys.amiga.misc,comp.sys.amiga.advocacy,comp.sys.amiga.datacomm
Subject: Re: New Press Release!
Date: Sat, 23 Mar 1996 13:57:04 +0100
Organization: ITnet
Message-ID: <[e-mail]>
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
NNTP-Posting-Host: agos001.pn.itnet.it
X-NewsReader: IntuiNews 1.3 (27.8.95)

Steve Koren wrote in comp.sys.amiga.networking about "Re: New Press Release!":
> >  There's little intelligence in comparing a 100MHz cpu with a 25MHz
>
> Yes, exactly.  That's what I've been trying to say :-), but you keep
> insisting on saying, "well, *if* they ran at the same clock speed, the
> 68K would be faster!".  While true, that is meaningless unless you can
> actually purchase a 68060 at the same clock speed.

  Yes and no. I can still compare "computing power" by normalizing tests;
  yes 060 is available only @50MHz while the Pentium is available in the
  range 75-166MHz, however I can still normalize speed results obtained
  by Pentiums and compare with speed results obtained from the 060,doing
  so you'll probably find a 060@50 == P5@75.

> >  Well it means cause if cpu A provides 2x power than cpu B at the same
> > clock
> >  rate I tend to think that cpu A is clearly superior in terms of "power".
>
> No, it doesn't mean that at all.  "At the same clock rate" is
> meaningless when the CPUs cannot run at the same clock rate.  The DEC
> Alpha run at 266 MHz.  Even if it would be as slow as a 68060 if it ran
> at 50 MHz, it is still quite safe to say it is a more powerful chip.
> One has to compare reality, not some hypothetical situation where you
> make the clock speeds the same.

 Indeed.I've said from the beginning that all that matters is how well
 a cpu performs in the real world...the Pentium clearly wins hands down.
 However this is cause it's available in higher clock rate flavours...
 DEC Alpha cpus are so fast cause they run at those high clock rates,
 this does not necessarily imply that DEC Alpha are "powerful chips".

> If my LW render or a complex symbolic integration takes many times
> longer on my 68060, I care little that it *would* be as powerful as the
> P6 *if* it could run at 200 MHz.

 Of course.But this does not imply that the P6 is "better" than the 060,
 it only implies that P6 is available at higher clock rates.

> >  However it seems that Pentium and 060 at same clock rate are about as
> >  fast.
>
> Again, only in theory, because in real life the clocks speeds are not
> the same.  166 MHz Pentiums are available now, and 200s are either out
> or close.  Would you like to bet on the chance of there ever being a 166
> MHz 68060?

 You're misunderstanding my thoughts.68K line is dead,I know.All I was
 trying to say is that the 060@50MHz is about as fast as Pentium as
 75MHz.The problem is that Pentiums are available at higher clock rates,
 that's all.

> > > answer is always the same: the Intel chips have higher absolute FP *and*
> > > Int performance than 680x0.
> >
> >  No this is wrong.Motorola tests say the 040 is twice as fast than a 486
> >  at the same clock rate.Pentium is instead just as fast than a 060 at the
> >  same clock rate.
>
> I said the Intel chips have a higher absolute FP and int performance.
> This is not wrong.  I did *not* make any claims about any specific
> chips.  Of course, you can find specific comparisons where Motorola is
> faster than Intel.  You can find specific comparisons where Zilog is
> faster than Intel, too.

 I'm pretty sure.040 is about 1.5 times faster than a 486 ASCR at integer
 math,2 times faster than a 486 at fp math.These are not specific tests,are
 averaged tests.

> My claim was only that the Intel chips are the faster architecture,
> because there are no 68K chips that run as fast as even the middle of
> the road Pentiums, let alone the P6's.

 Agree 100%

> I hate Intel chips because they have breathtakingly poor non-orthogonal
> instruction sets, induced a huge amount of legacy 16 bit code, and many
> other reasons.  But it makes little sense to hate them for being slow,
> because they are a great deal faster than the architecture I like (68k).

 I don't hate them cause they are slow.I hate them cause they are the
 now de-facto standard while being so poorly designed (from the
 programmer's point of view).

> >  I'm not posting this crap..I've just stated that the problem is that Intel
> >  chips are available at higher clock rates.
>
> Right.  If it was easy to get the 68K up to 200 MHz, it would have been
> done.  There are many factors that determine the clock speed at which a
> chip can be run, and one cannot always just increase the clock rate
> arbitrarily or cheaply.  Chip designers must design for high clock
> speeds, and even then, yield at the higher rates is often poor.

 Indeed.What hurts me the most is Motorola's attitude at this competition.

 Cheers
Stefano

+--------------------------+-------------------------------------------+
| Stefano Agostinelli      | Now developing for You ARM :              |
| Physics Dep. - Genoa     | the state of art Amiga Role Master System |
+--------------------------+-------------------------------------------+
| IRC: arm                 | WWW: www.geocities.com/SiliconValley/3630 |
| E-M: [e-mail] | Look at Stefano.html and ARM.html         |
+--------------------------+-------------------------------------------+