On Mon, 9 Aug 2004, emx wrote:
> Here is a post that was made on one of the web sites dedicated to John
> Titor.
I will comment on this post:
> I'm not sure if anyone else has posted anything about the 5100, but
> this machine was very interesting. In the early 80's, I worked on a
> slightly upgraded version of the 5100 called the 5110. The 5100/5110
> machine executed code intended for IBM mainframes of the time (System
> 360/370) via a simple processor that emulated the old 360/370 CPU.
This is wrong. The PALM processor (as used in the 5100/5110) has its own
design and its own microcode and machine code. It does NOT execute any
other microcode other than that programmed into a ROS on the CPU card. The
"frontend" languages APL or BASIC are indeed written in other machine
code. In order to run these interpreters, a program written in PALM
machine code is run (part of the Executable ROS) to interprete the
interpreter. That's why BASIC/APL is running so slowly on these machines.
The PALM CPU itself is a full 16 bit processor with 8 bit ALU and 18 bit
data path (incl. parity bits for each byte).
> Languages, such as APL or VS BASIC were programs written in machine
> language, that were, in turn, interpreted by the CPU in a lower level
> language called 'Microcode'
If microcode == machine code, then yes.
> Microcode allowed IBM to modify or enhance basic machine language
> instructions at will after the mainframe was delivered to the
> customer. The concept was 'ship an initial version of the 'machine',
> but allow upgrades as they were developed without the wholesale
> replacement of hardware'. To change a basic instruction in
> microprocessors today usually requires a physical change in the chip,
> since all instructions are 'hard-wired' into the design of chip.
You can't change the microcode of the PALM processor. That would be like
changing the microcode of e.g. a MC68000.
And today's processors are all microcoded, i.e. you DON'T have to change
the wiring of the processor, but only the so called microprogram for
exactly this one instruction. Well, the problem is that the microprogram
ROM is located on the CPU chip...
> The core design of the 5100 series machine was an emulator of the old
> IBM 360/370. The CPU of these machines was designed to execute the
No, the SCAMP was designed to run a subset of the IBM 1130 processor. The
PALM processor really has nothing in common with either the 1130 or any
360/370.
> mainframe. It was not a fast machine, but it worked, and the
> performance of the old 5100 was good enough to run many of the same
> applications we had developed on our $10 million twin IBM 370
In my opinion the PALM processor is quite fast, e.g. around 500k
instr./sec.
> mainframe installation. APL and VS BASIC apps ran with little or no
> modification, except for storage device definitions, which were
> limited to 300KB 8" floppies or 150KB streaming tape cartridges.
Well, the 300kB are 1,1MB (DS/DD) and the 150kB are more like 220kB (or
500kB with DC600 cartridges).
Again: For more information have a look at
http://computermuseum-stuttgart.de/dev/ibm_5110/technik/en/index.html
Christian Corti
Received on Mon Aug 16 2004 - 04:02:37 BST