Path: ...!2.eu.feeder.erje.net!feeder.erje.net!eternal-september.org!feeder3.eternal-september.org!news.eternal-september.org!.POSTED!not-for-mail
From: anton@mips.complang.tuwien.ac.at (Anton Ertl)
Newsgroups: comp.arch
Subject: Re: is Vax addressing sane today
Date: Fri, 04 Oct 2024 15:07:17 GMT
Organization: Institut fuer Computersprachen, Technische Universitaet Wien
Lines: 95
Message-ID: <2024Oct4.170717@mips.complang.tuwien.ac.at>
References: <2024Oct3.085754@mips.complang.tuwien.ac.at> <memo.20241003234930.19028I@jgd.cix.co.uk>
Injection-Date: Fri, 04 Oct 2024 19:28:15 +0200 (CEST)
Injection-Info: dont-email.me; posting-host="e836c6a0a468f0be540ff82074643594";
	logging-data="336965"; mail-complaints-to="abuse@eternal-september.org";	posting-account="U2FsdGVkX1+oSBAQpWeTdQgVcIoG9tH5"
Cancel-Lock: sha1:BNDIzeqGnTCBofSYViB3CP3YA0s=
X-newsreader: xrn 10.11
Bytes: 5620

jgd@cix.co.uk (John Dallman) writes:
>In article <2024Oct3.085754@mips.complang.tuwien.ac.at>,
>anton@mips.complang.tuwien.ac.at (Anton Ertl) wrote:
>
>> If the RISC companies failed to keep up, they only have themselves to 
>> blame. It seems to me that a number of RISC companies had difficulties
>> with managing the larger projects that the growing die areas allowed.
>
>Another contributing factor was Itanium, which was quite successful at
>disrupting the development cycles of the RISC architectures.

That's the question.  It seems to me that many struggled even before,
and jumped ship to IA-64 ASAP.

>Alpha suffered from DEC's mis-management, which led to DEC being taken
>over by Compaq. They killed Alpha when Itanium first became to work, and
>before it was clear that it was a turkey.

Alpha suffered before.  The 21264 was late, and did not keep up in the
clock race.  While they had higher clock rates than the competition up
to the EV56 (1996), the OoO EV6 appeared with a lower clock rate than
the in-order EV56 (while the OoO Pentium Pro had a higher clock rate
than the in-order Pentium available at the same time), and did not
scale as well with smaller processes as the Intel and AMD CPUs, which
were making huge strides in those years.  Intel then had the 2000MHz
Pentium 4, and AMD the 1200MHz Athlon in 2000 (and 1400MHz by the time
Alpha was canceled); unfortunately, release dates for EV6 variants at
different clock rates are not documented on Wikipedia, so
unfortunately I cannot make a table of Alpha vs. Intel and AMD clock
rates by year.

>PA-RISC was intended by HP to be replaced by Itanium. They managed that,
>but their success was limited because Linux on x86-64 was so much more
>cost-effective.

Reportedly they thought early on that they could not afford to keep
their own line competetive, so they started the IA-64 project with
Intel.  Interestingly, they also designed the OoO PA-8000, which was
introduced at the same time as the Pentium Pro, and they used the same
microarchitectur until they introduced the PA-8900 almost 10 years
later, which showed a more evolutionary approach than most others used
in those years.

>IBM kept POWER development going through the Itanium period, which is a
>significant reason why it's still going.

With the Power 4+ (2003) it also got competetive clock rates
(although, judging by the PowerPC 970, I wonder what the IPC was).

>SGI went into Itanium hard and neglected MIPS development, which never
>recovered. It had been losing in the performance race anyway.

The followon project "Beast" for the R10000 failed (was canceled), and
then SGI management was happy to jump ship to Itanium, and in the
meantime they only respun the R10000 into R12000, R14000, R16000.

>Sun kept SPARC development going, but made a different mistake, by
>spreading their development resources over too many projects. The ones
>that succeeded did so too slowly, and they fell behind.

Intel, HP, SGI and AMD went to OoO in 1995/1996, Alpha in 1998, Power
at the latest with Power3 in 1998, only Sun kept doing in-order stuff,
and took until 2011 to finally get an OoO CPU out the door in the form
of the SPARC T4 (their Rock project was also OoO, but was canceled).
They also had relatively low clock rates before that (which changed
with the SPARC T5).  Fujitsu managed better, introducing the OoO
SPARC64 V in 2002, and also with competetive clock rates.

>Also, Linux ate
>their web-infrastructure market rather quickly.

Well, SPARC survived much longer than most others, despite being
technically a lot behind.

Power still survives, maybe only because it has a common basis with
iSeries (or whatever it is called now).  Similarly, s390x survives
because of its software legacy.

>Linux could not have had the success it did without the large range of
>powerful and cheap hardware designed to run Windows. 

It was first developed on a 386, and many of the early co-developers
also had IA-32 machines.  But the 386 certainly was not designed to
run Windows.  The 386 project was finished before Windows 1.0 was
released in November 1985, and nobody used Windows 1.0 or 2.0, so why
would anybody design a processor for those?  Windows became only
popular with 3.0 in 1990 (after the release of the 486, which was
therefore not designed for Windows, either).  When I bought my first
PC (with a 486) in 1993, it ran DOS (for games) and Linux (for
everything else).

- anton
-- 
'Anyone trying for "industrial quality" ISA should avoid undefined behavior.'
  Mitch Alsup, <c17fcd89-f024-40e7-a594-88a85ac10d20o@googlegroups.com>