View Single Post
  #7  
Old November 26th 12, 04:51 AM posted to alt.comp.hardware.pc-homebuilt,comp.sys.ibm.pc.hardware.chips,comp.sys.intel,alt.comp.hardware.amd.x86-64
Michael Black[_2_]
external usenet poster
 
Posts: 164
Default The end of the road for the DIY PC?

On Sun, 25 Nov 2012, GMAN wrote:

In article , "geoff" wrote:
Apple was/is like that, limited options in changing out hardware. If Intel
completely removes the DIY aspect of a PC then they are handing business
over to Apple. Also, a lot of third party vendors will probably close shop.

Fixed hardware + a Bing OS (aka Windows 8) = a fast declining pc industry.


Sounds like the 1990's Atari ST, AMIGA all over again.

Really, every computer. Sure you could buy an S100 bus systemin the early
days, but there was limited ability to upgrade despite all the boards
plugging into a motherboard that only had sockets.

It was easy to move to the Z80 from the 8080. But the bus was very much
related to the 8080, so "foreign" CPUs took a lot of adapting. Even the
front panel on the Altair was too specific to the 8080 to be useful with
another CPU. The standardization was often because of CP/M the operating
system, since it was written to keep the I/O in a small section, one could
fairly eaily adapt it to other hardware (as long as it used the 8080).

So the real upgrade path was the 16bit CPU, preferably the 8088 or 8086.
But then there were other issues besides differring buss signals, such as
lack of address lines for more RAM. There were various schemes to deal
with that, but it took time before standardization set in, and then it was
mostly too late.

When MITS came out with a 6800 based computer in the fall of 1975, they
put a different bus on it, and when SWTP put out their computer (which was
far more successful 6800 system than the MITS 6800 system) it used a
different bus (though that bus tended to be used by other 6800 based
computers).

The DIgital Group that was more like a hobby trying to turn into a
commercial product, it used its own bus which made it easy to have
different CPU boards, but they never went further than the Z80 and maybe
the 6502.

The Apple II wsa fairly flexible, so one could get Z80 cards for it, then
later 6809 cards, and at some point 68000 cards. But they were
workarounds and usually the 6502 did the I/O.

Let's not forget that the original IBM PC was no different from that Amiga
or Atari. ALl three had CPUs in sockets, but there was no plug in
replacement that made things faster. You could workaround that, but it
would need a whole board. And you'd be stuck with the existing clock
frequency unless you had complicated timing methods (to run the CPU faster
but the bus at its regular rate).

It was only with time that the "IBM PC" became more flexible. And that
was more a crossover between the CPUs and the motherboard manufacturers.
So you could put in a faster CPU, but that's because the motherboard
company anticipated faster speeds and put in jumpers. That meant the CPU
companies had to keep the other companies informed of where they were
going.

In the 386 era there was some level of variability, so you could get a
cheaper one that had no math coprocessor built in (and oddly then find a
math coprocessor to add later).

It was really in much more recent times that a motherboard had some hope
to be useable over time, and that was because the CPUs generally stopped
changing that much, the speed being the key factor. If the motherboard
anticipated upgrades, and the CPU kept the same package and other
features, then you could use the motehrboard for a few years. Usually a
new motherboard was needed if the databus bumped up in size, the exception
being eventually with the 32-64 upgrade.

Otherwise, it would be no different from the Amiga or Atari, except by
that point nobody was making CPUs to plug into the expansion bus (I once
found an 80286 card that did that), so you had to replace the motehrboard.
But then, the motherboard probably cost as much as one of those plug in
upgrade boards in the past, but the new motherboard didn't have to
compromise. The only good thing was the case was generally generic so the
new motherboard fit (well so long as the area for connectors at the back
matched up or could be replaced).

Michael