[Simh] On number bases
Richard Cornwell
skyvis at sky-visions.com
Fri Feb 6 09:01:18 EST 2015
On Fri, 6 Feb 2015 09:35:35 +0100 (CET)
Andreas Davour <ante at Update.UU.SE> wrote:
> On Thu, 5 Feb 2015, dave porter wrote:
>
> > Octal makes sense for the PDP11 since the instruction
> > format is divided into 3-bit fields (operand modes,
> > register specifiers, etc).
> >
> > Hex makes sense for the VAX since the instruction
> > format is divided into 4-bit fields (operand modes,
> > register specifiers, etc).
>
> How about PDP10, PDP8, PDP15 and other fun architectures?
>
The PDP 8 has a 3 bit opcode, 2 bits of flags, and 7 bits of address.
So Octal makes more sense then Hex.
The PDP 15 a 4 bit opcode, 2 bits of flags and 12 bits of address. It
could go either way.
The PDP 10 has a 9 bit opcode, 2 four bit register fields, indirect
field and 18 bit address. However it does not split nicely along
either Hex or Octal lines. However 2 18 bit octal numbers are
probably the best way to represent this. The High order 3 bits of the
opcode determines the group and the format for how the rest of the
opcode is encoded. The next 4 bits generally the operation, or bits
to encode the operation this is really true for the Bit Testing 6xx
and Half Word 5xx, instructions. For my KA10 simulator I grouped
looked at the top 6 bits to decide how to handle the operation. But
many opcodes where handled in one batch, like the Booleans.
The IBM 360 instructions might make more sense in Octal, however the
addressing fields are based on 4 bit register 12 bit offset. The 4
bit field is what messes up octal encoding.
The CDC6600 uses a 6 bit opcode, 3 3 bit register fields or 2 3 bit
register fields and a 18 bit offset. This makes most sense to use
octal.
I think it really comes down to which you are most comfortable with
and which format the manuals come in.
Rich
--
==========================================================================
Richard Cornwell
skyvis at sky-visions.com
http://sky-visions.com
==========================================================================
More information about the Simh
mailing list