Curious--your favorite architecture?

Most people here use the S/360+ architecture for their
assembly language programming.  I was curious what people
think of other common architectures, both previous and
subsequent to S/360.

For instance...

AS/400 LIC (whatever that is)?
1401/1410 Autocoder/SPS?
709/7090 (word oriented)
x86 (all variants through Pentium)
1130
S/360+ supervisory state
CDC 6xxx series
Univac 1100 series


(S/360 includes all subsequent models)

hancock4@bbs.cpcn.com wrote:
> Most people here use the S/360+ architecture for their
> assembly language programming.  I was curious what people
> think of other common architectures, both previous and
> subsequent to S/360.

> For instance...

> AS/400 LIC (whatever that is)?
> 1401/1410 Autocoder/SPS?
> 709/7090 (word oriented)
> x86 (all variants through Pentium)
> 1130
> S/360+ supervisory state
> CDC 6xxx series
> Univac 1100 series

> (S/360 includes all subsequent models)

<delurk>

My first machine was the IBM 1620, which was fun, but I really liked the
CDC 3600 (and 3800) I wrote code for at the NASA MSC, these 40 years
ago. The CDC 3200, the 3600's little brother, was nice, too, but the
short (24-bit) word meant that the 3[68]00's fullword opcodes had to be
implemented as macros or subroutines, where necessary. As the 3200s mostly
did I/O for the 3[68]00, or did donkey-engine work while the 3[68]00 dud
intensive number-crunching, there wasn't much need to use the 3600-specific
stuff on the smaller machines. The 1620's appeal was that it was so cery
easy to do string-mode ops on.

I've written for the 709x and 704x, but thought that they were clunky, as
though they were interim designs -- stops on the way to a real solution, as
it were.

The CDC 6600 was fun, too, but I didn't get to do much with it.

For _real_ arcana, the seminar on advanced programming at ou.edu in 1973
was near the top: writing code for ILLIAC IV, and getting to test it.

<lurk>

--
Mike Andrews, W5EGO
mikea@mikea.ath.cx
Tired old symin since 1964

Mike Andrews wrote:

> I've written for the 709x and 704x, but thought that they were
clunky, as
> though they were interim designs -- stops on the way to a real
solution, as
> it were.

I thought the 7094 was the same as a 7090 (which was the same as
the 709); that the 7094 just was a little faster.  So, I thought
the machine language would be all the same.

hancock4@bbs.cpcn.com wrote:
> Most people here use the S/360+ architecture for their
> assembly language programming.  I was curious what people
> think of other common architectures, both previous and
> subsequent to S/360.

Well, the main reason for following this group, which doesn't get
much traffic, is that I like the architecture.  It is fairly simple,
yet powerful.   After less than two years of Fortran, a copy of the
Principles of Operation, and some sample assembler programs I started
writing Fortran callable subroutines.  (So I didn't have to learn I/O
in the beginning.)

> AS/400 LIC (whatever that is)?

No comment on this one.

> 1401/1410 Autocoder/SPS?
> 709/7090 (word oriented)
> x86 (all variants through Pentium)

x86 is complicated in a variety of ways.  Some are the shortage
of registers and the special use that many have.  Complicated
addressing modes that don't simplify things so much.  Also,
complications due to the assembler and linker conventions which
aren't really part of the architecture but make it harder, anyway.

> 1130
> S/360+ supervisory state
> CDC 6xxx series
> Univac 1100 series

I never had a chance to use the older IBM machines, nor the CDC or
Univac machines.

The second machine I did assembler programming on was the PDP-10,
which isn't so hard to learn after S/360.  Indirect addressing with
the indirect bit is a little different, but there are still 16
registers, and many of the same operations.   It is word addressed
complicating work with characters.

VAX isn't so different, though it has many addressing modes, and,
somewhat like S/360 was designed in part to make assembly programming
easier.  (Unlike most RISC architectures which are designed for compiler
generated code.)

Considering the S/360 architecture is now over 40 years old and still
viable says something about its design!

-- glen

hancock4@bbs.cpcn.com wrote:
> Most people here use the S/360+ architecture for their
> assembly language programming.  I was curious what people
> think of other common architectures, both previous and
> subsequent to S/360.

> For instance...

> AS/400 LIC (whatever that is)?
> 1401/1410 Autocoder/SPS?
> 709/7090 (word oriented)
> x86 (all variants through Pentium)
> 1130
> S/360+ supervisory state
> CDC 6xxx series
> Univac 1100 series

> (S/360 includes all subsequent models)

<delurk>

My first machine was the IBM 1620, which was fun, but I really liked the
CDC 3600 (and 3800) I wrote code for at the NASA MSC, these 40 years
ago. The CDC 3200, the 3600's little brother, was nice, too, but the
short (24-bit) word meant that the 3[68]00's fullword opcodes had to be
implemented as macros or subroutines, where necessary. As the 3200s mostly
did I/O for the 3[68]00, or did donkey-engine work while the 3[68]00 dud
intensive number-crunching, there wasn't much need to use the 3600-specific
stuff on the smaller machines. The 1620's appeal was that it was so cery
easy to do string-mode ops on.

I've written for the 709x and 704x, but thought that they were clunky, as
though they were interim designs -- stops on the way to a real solution, as
it were.

The CDC 6600 was fun, too, but I didn't get to do much with it.

For _real_ arcana, the seminar on advanced programming at ou.edu in 1973
was near the top: writing code for ILLIAC IV, and getting to test it.

<lurk>

--
Mike Andrews, W5EGO
mikea@mikea.ath.cx
Tired old symin since 1964