Intel strikes back with a parallel x86 design

  • Thread starter Thread starter Jim Brooks
  • Start date Start date
I don't understand how you can have a 2GHz 'gap'. What's so good about
a high clock rate, anyway?

Robin
 
YKhan said:
was in fact pushing it at that time. IBM was also fully behind the
later 386.
Click to expand...

IBM was only behind the 386 to the extent that it didn't cannibalise any of
their existing products markets.
 
Workstations at $10k+ a pop are never going to let you rule the world.
Click to expand...

You could buy several 68K-based ones for c. $5K, which was only
40% more than the $3.5K you needed for an IBM PC, and they could
do several times as much. Or could have done, if the software
actually worked.
Only a little bit later? PowerPC was introduced in 91 or 92.
Click to expand...

You are posting as an outsider. I am not. My point is that IBM
could have delivered a lot earlier than they did.
You're seriously trying to feed me this crap about IBM having PowerPCs
around 1985, when the 386 was first introduced, and you expect anyone
to take you seriously? By the time PowerPC was introduce, it was
competing against 486's and the first of the Pentiums.
Click to expand...

(a) See above and (b) read what I posted. For a system to be viable,
it has to have software as well as hardware. Not merely was the
80386 late, but the software to take advantage of it took ages to
arrive.

The PowerPC was not competing against the 80386 as something to run
Windows 3.1 on, but as something to run a more serious operating
system. And that didn't happen until the very end of the 80386's
life, except in a few specialist systems. I take your point that
the real competition was the 486, but that changes little.


Regards,
Nick Maclaren.
 
Jim Brooks wrote On 09/23/05 11:11,:

The multithread model already runs each thread opportunistically.
Timer interupts already occur on instruction boundaries.
Click to expand...


You're thinking in terms of sw threads and at the OS level.

hw threads are a set of instruction streams
the CPU itself schedules to hw execution units.

Move a broken priority model from software to hardware. For what
speed advantage ?
Click to expand...


If two hw threads (two instructions from different threads)
are contending for one currently available hw execution unit,
the hw scheduler should select either based on dynamic thread
priorities (similar to an OS scheduler).

- sub-threads

Support for parallel programming.
A reduced 80386 Task-State Segment (TSS) will be defined
(avoid saving unnecessary registers such as ES/FS/GS)
A variant of JUMP [TSS] with a new Thread bit defined in the TSS
will spawn a sub-thread (analoguous to a UNIX child process).
The sub-thread can stop by IRET [TSS].
A new WAIT [TSS] will synchronize the parent with its sub-thread.

- thread exceptions

A thread can raise exceptions to end or suspend itself.
Click to expand...

These are all software features. There is no reason to move
the slowest parts of thread management to hardware, such as scheduling
of same.
Click to expand...


<sigh>

Think about how to maximize chip die area.
And how to actually _take_advantage_ of the pipeline stalls
that occur in a single-threaded superscalar core.

Doubt it, but hey...
Click to expand...


Else the threads will thrash the cache
as does Sun's putrid plagiarization of C++.

SUN. :-)


Yea, lets optimize the least used section of the instruction set.
Click to expand...


FADD/FMUL/FLDST mean floating-point adder, multiplier, load/store units,
resp. Not to be confused with eponymous x87 instructions (sorry).
 
Don't be silly.


Either you are very young, or your memory is failing. Let me remind
you.

Back in the early 1980s, there were several chips fighting it out
for the workstation market. Intel's 8086 wasn't up to it and even
the 80286 was pretty marginal - the 68000 and 68010 were rapidly
becoming the dominant chips in the high-end market. What stopped
them from becoming dominant was that no company produced a 68K-based
workstation that was both relatively cheap and with working software.
Sun and Apollo established themselves because their systems
more-or-less worked. There were much cheaper 68K boxes, but their
software was crap, and they didn't have the third-party support.
There were at least a dozen companies in the world who could have
put together a decent 68K-based system over a year before the 80386
became viable. None did.
Click to expand...

I'm certainly not young and I have a very good memory of my own experience.
Remember (did you even know) Definicon? They sold a PC add-in card with
68020, 68882 and 1MB(2MB option, IIRC) memory - it was reasonably priced
and before 80386 was a viable route to 32-bit address space and what we
thought was great performance. Admittedly it was saddled with the ISA-16
bus to talk to the PC system for I/O but the compilers worked well and
produced reasonably efficient code - I checked.

I still remember the day I got our first 16MHz 80386/80387, with 2MB memory
and ran a comparison of our LP software against the 20MHz Definicon
68020... excitedly hoping for confirmation of the superiority of the 68020.
What a let down - the 80386/8037, even at 16MHz blew the 20MHz 68020/68882
away... no contest.

The 68020 failed because it was not competent.

The Intel CPUs succeeded because Intel supplied a "kit" which worked - they
documented all the layout & bus loadings down to resistors and capacitors,
supplied the support chips and compared with Moto 68020 and especially
NS16032/32032, it was like painting by numbers to build a working system.
Only a little bit later, IBM and Motorola set up the PowerPC project.
Let's skip over the long and complex details, but the facts of the
matter were that IBM had viable systems a full year before the 80386
became viable, but wouldn't release them. And the PowerPC consortium
was really quite big at that stage, including Apple and most of the
Tier 2 vendors at least having taken out options. And please note
that Intel and Microsoft combined were small compared to IBM back in
the days of the 80286.
Click to expand...

I think you have your dates a bit muxed ip - we had 80386 and Phar Lap's
DOS Extender in 1986; I didn't get a sniff of even a Risc/6k until 1990.
Are you sure you're not thinking of the PC RT/Risc systems? They were not
PowerPC but some weird CPU IBM had laying around - the 801 IIRC. ISTR a
weird virtual address cache arrangement. It was a non-starter
performance-wise and I didn't pay much more attention to it after an
initial disappointing try.
 
(a) See above and (b) read what I posted. For a system to be viable,
it has to have software as well as hardware. Not merely was the
80386 late, but the software to take advantage of it took ages to
arrive.

The PowerPC was not competing against the 80386 as something to run
Windows 3.1 on, but as something to run a more serious operating
system. And that didn't happen until the very end of the 80386's
life, except in a few specialist systems. I take your point that
the real competition was the 486, but that changes little.
Click to expand...

You are missing one point: the cost of software was prohibitive (aka,
priced assuming rich business customers). In a slightly later time frame
(early 90s), my father bought a Win 3.1 PC+CAD s/w+pspice to do (analog)
circuit design (as a consultant). He considered workstations, but the cost
of the equivalent s/w was many thousands of dollars higher (more like in
$10'000 increments, but I can't remember the precise numbers). The PC s/w
was more in the low thousands of dollars price range.
 
|>
|> You probably know this but
|>
|> If my memory serves me well, 2 of those were not failed, at least not
|> during their heyday they were quite successful selling in the millions
|> mark when a million actually meant something. One of them is alive and
|> well inside your settop box (that must mean many many millions at 70%
|> market share) but don't ask ST to name it, it hurts too much to say the
|> word.

Yes. What the x86 fanatics miss is that there are a large number
of designs that could perfectly well have prevented its rise, or
toppled it from its perch and taken over during one of its more
vulnerable periods. Its success was always more a matter of luck
(and incompetence by the opposition) than merit.
Click to expand...

True. ...until the critical mass was reached (long ago). IBM had the exact same
problem thirty years ago with FS, vs. S/360. The installed software was
worth far more than the hardware and any benefits the hardware *could*
have were far offset by the inertia of the installed base. We're now stuck
with x86 and the heroic efforts to make it work. Intel forgot this
lesson, though they've had it handed to them a few times.
In addition to those systems and the Alpha, there was the 68K range and
PowerPC, which both came VERY close to blocking the rise of the x86 and
toppling it, respectively. We know why they didn't, too, and the
reasons were not architectural.
Click to expand...

Ancient history. 68K had a chance, though they blew it. PowerPC as an
x86 killer was a dream. Remember 615?
Nowadays, with the patent system preventing innovation by new companies
and established companies not being prepared to tackle new
general-purpose architectures, I doubt that anything could make headway
until the x86 collapses of its own accord. Unless, of course, that
China says "sod you" to the USA over patents and starts innovating
itself.
Click to expand...

Perhaps. However, Intel has had more fumbles than the NFL allows. AMD
now has the ball.
 
never saw you without long pants.
Click to expand...

To be truthfull, you never seen me in them either. ;-)
Too cold for shorts up there in the green mts?
Click to expand...

Cold? Nah, that's a flat-lander tundra thing. It won't get even get cool
here until at least midnight.
Or you just can't spell?
Click to expand...

I repeat; I thought you knew me better. ;-)
 
Chris said:
IBM was only behind the 386 to the extent that it didn't cannibalise any of
their existing products markets.
Click to expand...

Which market exactly was the 386 going to cannibalize? The 286 market?

Yousuf Khan
 
YKhan said:
Which market exactly was the 386 going to cannibalize? The 286 market?
Click to expand...

there was period (fall '88) where far east clone manufacturers built
up an enormous inventory of 286 systems in preparation for fall season
buying frenzy ... and cut-rate 386sx systems come on the market and
wiped out the 286 demand. all those 286 systems eventually flooded the
market at fire sales.

this has oblique mention to 386sx hasten end of 286 era in the
fall of 88.
http://www.pcmag.com/article2/0,1895,1162496,00.asp
 
Nick said:
You could buy several 68K-based ones for c. $5K, which was only
40% more than the $3.5K you needed for an IBM PC, and they could
do several times as much. Or could have done, if the software
actually worked.
Click to expand...

You see, this is exactly my point about the comp.arch beatniks -- they
prefer to live in a peppermint world it seems. Something that's "only
40% more" than something else is nothing short of a dismal failure.
Especially if it's got no software actually working.

You want software that actually works? Well you won't get that by
paying only 40% more, you have to pay 185% more to actually get working
sofware.
You are posting as an outsider. I am not. My point is that IBM
could have delivered a lot earlier than they did.
Click to expand...

Right, "I swear IBM could've delivered this hardware six years before
they actually delivered it. They were just holding back."

(a) See above and (b) read what I posted. For a system to be viable,
it has to have software as well as hardware. Not merely was the
80386 late, but the software to take advantage of it took ages to
arrive.
Click to expand...

The viability of 386's is not in question here, they were viable the
day they arrived on the shelves. They ran all of the thousands of
software designed for x86 up until that point. PowerPC was not viable
because it had no software base at all during that time.
The PowerPC was not competing against the 80386 as something to run
Windows 3.1 on, but as something to run a more serious operating
system. And that didn't happen until the very end of the 80386's
life, except in a few specialist systems. I take your point that
the real competition was the 486, but that changes little.
Click to expand...

So the PowerPC ran more serious operating systems, I suppose? Like
what, MacOS? The 386's ran everything from Windows to Netware, to Unix
including Linux.

Yousuf Khan
 
on the other hand ... here are some raw chip numbers from 1990 (as
opposed to pc systems)

CA's CA's CA's CA's CA's SHARE
CPU 1986 1987 1988 1989 1990 1990
INTEL 80186/286 971622 1840928 2506136 2924733 3077693 36%
INTEL 808x 2374099 2546252 2232084 1913469 1456441 17%
INTEL 80386 19859 231767 625134 893610 949963 11%
INTEL 80486 0 0 0 301 64699 1%
MOTOROLA 68xxx 1921562 1787042 1354593 1378959 1273736 15%
UNKNOWN/OTHER 0 813485 664137 570155 653013 697567 8%
(includes Intel)
=============== 6100627 7070126 7325753 8103435 8620257 100%
 
I'm certainly not young and I have a very good memory of my own experience.
Remember (did you even know) Definicon? They sold a PC add-in card with
68020, 68882 and 1MB(2MB option, IIRC) memory - it was reasonably priced
and before 80386 was a viable route to 32-bit address space and what we
thought was great performance. Admittedly it was saddled with the ISA-16
bus to talk to the PC system for I/O but the compilers worked well and
produced reasonably efficient code - I checked.
Click to expand...

Yes. But those systems were too complex to take off.
I still remember the day I got our first 16MHz 80386/80387, with 2MB memory
and ran a comparison of our LP software against the 20MHz Definicon
68020... excitedly hoping for confirmation of the superiority of the 68020.
What a let down - the 80386/8037, even at 16MHz blew the 20MHz 68020/68882
away... no contest.

The 68020 failed because it was not competent.

The Intel CPUs succeeded because Intel supplied a "kit" which worked - they
documented all the layout & bus loadings down to resistors and capacitors,
supplied the support chips and compared with Moto 68020 and especially
NS16032/32032, it was like painting by numbers to build a working system.
Click to expand...

That was a factor, and made it hard for garage shops to build a
68K-based system, but I don't think that is relevant. If one HAD
been built to take off, it would have been by a fairly serious
vendor. Apple, Sun, Apollo and others had few problems.
I think you have your dates a bit muxed ip - we had 80386 and Phar Lap's
DOS Extender in 1986; I didn't get a sniff of even a Risc/6k until 1990.
Click to expand...

Yes and no. I responded on this in another posting. It was a few
years later, and was at the stage when the x86 systems were trying
to break into the serious workstation and server market. While
most of the action occurred when the 80386 was Intel's dominant
CPU, I shouldn't have mentioned that, on the grounds of it being
confusing. And my point is that IBM didn't release when they could
have done.
Are you sure you're not thinking of the PC RT/Risc systems?
Click to expand...

No. As you say, they were ghastly.


Regards,
Nick Maclaren.
 
You are missing one point: the cost of software was prohibitive (aka,
priced assuming rich business customers). In a slightly later time frame
(early 90s), my father bought a Win 3.1 PC+CAD s/w+pspice to do (analog)
circuit design (as a consultant). He considered workstations, but the cost
of the equivalent s/w was many thousands of dollars higher (more like in
$10'000 increments, but I can't remember the precise numbers). The PC s/w
was more in the low thousands of dollars price range.
Click to expand...

No, I am not missing that. One of the objectives of the PowerPC
project was to bring the prices of such 'quality' software down
to comparable to the x86-based stuff. IBM had plans to twist
arms of the third party people, but never got their systems out,
and never proceeded to that stage.

As I say, the failure had nothing to do with the merits of any
x86-based solution, but was entirely within IBM.


Regards,
Nick Maclaren.
 
Ancient history. 68K had a chance, though they blew it. PowerPC as an
x86 killer was a dream. Remember 615?
Click to expand...

The 615 was part of the way that IBM blew it. You are wrong if you
think that project wasn't viable, in principle, but it relied on
IBM keeping its nerve, putting enough resources in to maintain
schedules, and taking pragmatic, technically competent decisions.
IBM failed in all three ....


Regards,
Nick Maclaren.
 
You see, this is exactly my point about the comp.arch beatniks -- they
prefer to live in a peppermint world it seems. Something that's "only
40% more" than something else is nothing short of a dismal failure.
Click to expand...

Well, when you have graduated from your comforter to peppermints,
please let us know.

You might like to ask yourself why SCSI disks and, later, Fibrechannel
ones ever sold - they are considerably more than 40% more expensive
than the IDE etc. equivalents and always have been. Perhaps you don't
work in a field where people buy SCSI and Fibrechannel systems, but
I and many other posters here do.

The significant history of the 80386 and its successors was Intel's
long haul out of the cheap and nasty market (with high turnover but
no margins) into the medium and large business market (with much
lower turnover but vastly larger margins). The IA64 project was an
attempt to cement this but, in THAT case, it is Intel that blew it!

The point about the PowerPC project is not that it would have
competed with Intel's dominance of the cheap desktop market, but
that it could have dominated both the performance desktop and
(more importantly) small server market. Intel would then have
been trapped in a ghetto of negligible margins, and would then have
been gradually squeezed as the prices of PowerPC systems came down.

As history relates, IBM blew it.


Regards,
Nick Maclaren.
 
Only a little bit later, IBM and Motorola set up the PowerPC project.
Let's skip over the long and complex details, but the facts of the
matter were that IBM had viable systems a full year before the 80386
became viable, but wouldn't release them. And the PowerPC consortium
was really quite big at that stage, including Apple and most of the
Tier 2 vendors at least having taken out options. And please note
that Intel and Microsoft combined were small compared to IBM back in
the days of the 80286.
Click to expand...

Coulda woulda shoulda? It failed to dislodge x86 because they tried
to force a hardware solution without software to support it. How many
applications ever got ported to Windows/PowerPC?

Hardware = Cheap. Software = Expensive. This is why x86 dominated
the market once it got it's HUGE lead in software.
If I were to get $1 billion together, design a CPU that outperformed
the x86 10:1 and build it into a system, what chance would I have of
selling it without being smothered in lawsuits? Get real.
Click to expand...

What chances would you have to sell it, period? With no software
support you're already dead in the water. Best look towards the
embedded market for you're design and hope that you can get the power
consumption down.

Everyone looking to revolutionize the computer world by forcing a new
software model onto people have doomed themselves to failure. Once
people realize this then they might stop wasting their money and start
innovating in ways that people will actually use.
 
Nick said:
For a system to be viable, it has to have software as well as
hardware. Not merely was the 80386 late, but the software to
take advantage of it took ages to arrive.
Click to expand...

True, and when Microsoft finally did deliver an OS for the x86
family, it was also pitched at MIPS, Alpha and (later?) PowerPC.

However, the thing that killed all these (and OS/2, and UNIX on
the x86 until very recently, and almost killed NT (MS still though
it was worth launching Windows ME nearly 10 years after NT's debut)
was the market insisting that "it has to run that crappy DOS device
driver and that crappy game I like".

I *think* we are now past that point and consequently the x86 may
be more vulnerable now than at any point in the last 20 years.
On the other hand, it doesn't look like the x86 ISA is actually the
limiting factor if you are trying to climb memory walls or write
moderately parallel code, so perhaps (vulnerable though it is) it
is no longer worth beating.
 
Back
Top