x86 vs The World...

On Thu, 05 Jun 2014 11:24:22 +0200, David Brown wrote:

> It was low-power and low-size from the beginning.

A large part of this stems from having a shoe-string budget.

The small size of the development team limited complexity. The
power consumption had to be kept below 1W so that a plastic package
could be used; a ceramic package would have cost more than the chip itself.

David Brown wrote:
<snip>
>
> I think there are a mix of reasons - some technical, some non-technical.
>
> The x86 ISA is crap.  It was viewed as limited and old-fashioned the day
> it was created, and the IBM engineers designing the first PC wanted
> nothing to do with it (they picked the 68000).


I can't place the link, but everyone who touched the PC @ IBM was doomed 
by it.

>  But the chip was
> relatively cheap, and the 8-bit bus needed meant cheaper memory and main
> boards, so the PHB's pushed it as the cpu for the IBM PC.  Fast-forward
> to today and we have fantastic implementations of the same crap ISA.


For a while, PCI roamed the Earth. It's still around in forms.

I'm not so sure that the original ISA bus wasn't the master stroke - it
provided for all those third-party peripherals. The various forms of PC
provided for doing systems integration at a then-lower cost. This was
a significant business model from say, 1985 until 1995 or so. Perhaps
even past that.

To an extent, the Lisa was the computer everyone wished the PC was. It 
sank under its price tag.


> (amd64 is somewhat better, but still not a nice ISA compared to MIPS,
> PPC, or ARM.)  If the same clever folks at Intel had used the same time
> and resources on making a better implementation of MIPS or PPC, we would
> have faster, cheaper and lower-power cpus in the mainstream.
>

Strangely enough, there are still PC104 bus computers around.

I think you are perhaps underestimating the cost component of the 
barrier to entry for SGI and other "better" computers. They were
often the price of a newish used car.

<snip>

-- 
Les Cargill

On 2014-06-26, Les Cargill <lcargill99@comcast.com> wrote:
> David Brown wrote:
><snip>
>>
>> The x86 ISA is crap.  It was viewed as limited and old-fashioned the day
>> it was created, and the IBM engineers designing the first PC wanted
>> nothing to do with it (they picked the 68000).
>
>
> I can't place the link, but everyone who touched the PC @ IBM was doomed 
> by it.
>
>>  But the chip was
>> relatively cheap, and the 8-bit bus needed meant cheaper memory and main
>> boards, so the PHB's pushed it as the cpu for the IBM PC.  Fast-forward
>> to today and we have fantastic implementations of the same crap ISA.
>
>
> For a while, PCI roamed the Earth. It's still around in forms.
>
> I'm not so sure that the original ISA bus wasn't the master stroke - it
> provided for all those third-party peripherals. The various forms of PC
> provided for doing systems integration at a then-lower cost. This was
> a significant business model from say, 1985 until 1995 or so. Perhaps
> even past that.
>

I think David meant ISA as in "Instruction Set Architecture", not the
ISA bus.

I agree with him, BTW. :-)

Simon.

-- 
Simon Clubley, clubley@remove_me.eisner.decus.org-Earth.UFP
Microsoft: Bringing you 1980s technology to a 21st century world

On 6/26/2014 6:42 AM, Nobody wrote:
> On Thu, 05 Jun 2014 11:24:22 +0200, David Brown wrote:
>
>> It was low-power and low-size from the beginning.
>
> A large part of this stems from having a shoe-string budget.
>
> The small size of the development team limited complexity. The
> power consumption had to be kept below 1W so that a plastic package
> could be used; a ceramic package would have cost more than the chip itself.

What processors at that time were cranking over 1 watt?  I recall lots 
of CPUs at that time with reasonable power levels.  The original ARM was 
no competition for a desktop type CPU like the x86.

The real test of a CPU is the power for a given amount of work.  There 
should be a power oriented Dhrystone type test.  Measure the energy used 
while running the test and divide by the number of iterations in the 
test.  Joules per Dhrystone.

But most processors spend a lot of time doing nothing waiting for 
something else to happen.  PCs are the worst.  Then the only thing that 
matters can be measured by a Watt meter.

-- 

Rick

rickman wrote:

> On 6/26/2014 6:42 AM, Nobody wrote:
>> On Thu, 05 Jun 2014 11:24:22 +0200, David Brown wrote:
>>
>>> It was low-power and low-size from the beginning.
>>
>> A large part of this stems from having a shoe-string budget.
>>
>> The small size of the development team limited complexity. The
>> power consumption had to be kept below 1W so that a plastic package
>> could be used; a ceramic package would have cost more than the chip
>> itself.
> 
> What processors at that time were cranking over 1 watt?  I recall lots
> of CPUs at that time with reasonable power levels.  The original ARM was
> no competition for a desktop type CPU like the x86.
> 
> The real test of a CPU is the power for a given amount of work.  There
> should be a power oriented Dhrystone type test.  Measure the energy used
> while running the test and divide by the number of iterations in the
> test.  Joules per Dhrystone.
> 
> But most processors spend a lot of time doing nothing waiting for
> something else to happen.  PCs are the worst.  Then the only thing that
> matters can be measured by a Watt meter.

But the newer one are waiting faster :-)

-- 
Reinhardt

On 26/06/14 21:57, rickman wrote:
> On 6/26/2014 6:42 AM, Nobody wrote:
>> On Thu, 05 Jun 2014 11:24:22 +0200, David Brown wrote:
>>
>>> It was low-power and low-size from the beginning.
>>
>> A large part of this stems from having a shoe-string budget.
>>
>> The small size of the development team limited complexity. The
>> power consumption had to be kept below 1W so that a plastic package
>> could be used; a ceramic package would have cost more than the chip
>> itself.
>
> What processors at that time were cranking over 1 watt?  I recall lots
> of CPUs at that time with reasonable power levels.  The original ARM was
> no competition for a desktop type CPU like the x86.
>

On the first BBC Archimedes, one of the standard applications was a 
80286 emulator that could boot standard MS-DOS and run DOS applications. 
  It was a little slower than a real 80286 - but it was not 
significantly slower than a standard PC of two or three years earlier. 
That was just using a rather simple software emulator - when running 
native code, the ARM in the Archimedes rang rings round top-model PC's 
of the time.

One of the demonstration programs was a Mandelbrot set renderer - full 
screen (640 x 480 pixels) renderings took 7 seconds, while PC's took 
minutes to complete.  Another was a 3D spaceship game - IIRC, it was 
written in interpreted Basic on the Archimedes, but it had a level of 
graphics you didn't get in hand-coded assembly on PC's.  And the gui on 
the Archimedes had features like anti-aliased fonts and full window 
movement that were not seen on PC's until perhaps 8 years later.

Obviously these are best-case examples, and it is hard to do an 
apples-to-apples comparison of systems from 25 years ago, but I believe 
the early ARM chips completely outclassed the available x86 chips in 
pure processor speed.  And the Acorn software engineers and the RiscOS 
operating system completely outclassed anything MS (or Apple) could come 
up with.

(Acorn and ARM were not alone in this respect, historically.  For 
example, the Commodore Amiga with its 68k processor was far faster, and 
had much better software, than PC's of the time.)


> The real test of a CPU is the power for a given amount of work.  There
> should be a power oriented Dhrystone type test.  Measure the energy used
> while running the test and divide by the number of iterations in the
> test.  Joules per Dhrystone.
>
> But most processors spend a lot of time doing nothing waiting for
> something else to happen.  PCs are the worst.  Then the only thing that
> matters can be measured by a Watt meter.
>

Op 27-Jun-14 13:50, David Brown schreef:
> On 26/06/14 21:57, rickman wrote:
> (Acorn and ARM were not alone in this respect, historically.  For
> example, the Commodore Amiga with its 68k processor was far faster, and
> had much better software, than PC's of the time.)

I don't know much about the Acorn, but the 68k on Amiga was assisted by 
blitters and such which did a lot of the heavy lifting and gave the 
impression of relativity fast system, whereas on PC of that era the 8086 
had to do most of the work all by itself. A 8 MHz 68K wasn't that fast, 
though it did have a much, much nicer instruction set than the 8086 and 
80286.

Simon Clubley wrote:
> On 2014-06-26, Les Cargill <lcargill99@comcast.com> wrote:
>> David Brown wrote:
>> <snip>
>>>
>>> The x86 ISA is crap.  It was viewed as limited and old-fashioned the day
>>> it was created, and the IBM engineers designing the first PC wanted
>>> nothing to do with it (they picked the 68000).
>>
>>
>> I can't place the link, but everyone who touched the PC @ IBM was doomed
>> by it.
>>
>>>   But the chip was
>>> relatively cheap, and the 8-bit bus needed meant cheaper memory and main
>>> boards, so the PHB's pushed it as the cpu for the IBM PC.  Fast-forward
>>> to today and we have fantastic implementations of the same crap ISA.
>>
>>
>> For a while, PCI roamed the Earth. It's still around in forms.
>>
>> I'm not so sure that the original ISA bus wasn't the master stroke - it
>> provided for all those third-party peripherals. The various forms of PC
>> provided for doing systems integration at a then-lower cost. This was
>> a significant business model from say, 1985 until 1995 or so. Perhaps
>> even past that.
>>
>
> I think David meant ISA as in "Instruction Set Architecture", not the
> ISA bus.
>

Ah; my bad.

> I agree with him, BTW. :-)
>
> Simon.
>

-- 
Les Cargill

David Brown <david.brown@hesbynett.no> wrote:
> Another was a 3D spaceship game - IIRC, it was written in interpreted 
> Basic on the Archimedes, but it had a level of graphics you didn't 
> get in hand-coded assembly on PC's.

The game was Zarch by David Braben (of Elite fame), but the story about 
it being written in Basic is false. IIRC it had a loader written in 
Basic, which some journalists mistook for the game itself.

<http://en.wikipedia.org/wiki/Zarch>

-a

2014-06-13 05:18, Przemek Klosowski skrev:
> On Fri, 06 Jun 2014 13:59:46 +0000, Michael Koenig wrote:
>
>> One thing a lot of people forget is that ARM was actually the first
>> commercial RISC processor. Yes, it was inspired by the work of the
>> Berkeley-RISC and Stanford-MIPS teams, but their commercial results,
>> namely SPARC and MIPS, came a little bit later.
>>
>
> Not quite: MIPS and RISC came out in 1983 and ARM-1 in 1985. Berkeley RISC
> did not really have a commercial followup, and of course MIPS was the
> target of the Hennesy Patterson book that implanted the RISC ideology in
> the industry.
>


Might remember wrong, but I think the SPARC is following the Berkley 
RISC approach.

BR
Ulf Samuelsson