Re: Bit twiddling

Chris Croughton wrote:
On Fri, 03 Jun 2005 16:54:58 -0700, glen herrmannsfeldt <gah@xxxxxxxxxxxxxxxx> wrote:


*** T. Winter wrote:


In article <ln1x7jb47t.fsf@xxxxxxxxxxxxxxx> Keith Thompson <kst-u@xxxxxxx> writes: 

> Whether bit 0 is the least or most significant bit is probably
> system-specific; the bits in a 32-bit word might be numbered 0..31 or
> 31..0.  Which way they're numbered is probably more a matter of
> terminology than actual semantics, unless the CPU provides
> instructions that take bit numbers as arguments.


And if the instructions are consistent.  That is not the case with all
processors...


IBM machines derived from S/360 consistently number them with the
MSB as zero.  The terminology gets a little confusing now that the
architecture has been extended to 64 bits.  The old instructions now
operate on bits 32-63 of the general registers.



That's a good reason for numbering with 0 as LSB!  European comms
specifications had the 'standard' of numbering 0 as MSB, I believe that
it got into some early RFCs as well (I've certainly seen it in some
non-European spec.s).  Most confusing, especially when dealing with
multiple specifications in the same project (the lower layers have bit 0
as most significant, whereas the higher layers have bit 0 as LSB -- or
is it the other way round?  Madness!)

Chris C

In the end, happily, it doesn't matter. It's just a matter of notation. It doesn't change the way the machine works, only how it's described in the docs.

More interesting is a bit's value in time. Think of async serial bytes: start bit, 8 data bits and stop bit. It is of crucial importance that we know whether the bit after the start bit is the lsb or the msb of the byte. I used to know. Really. :-)

--
Joe Wright                            mailto:joewwright@xxxxxxxxxxx
"Everything should be made as simple as possible, but not simpler."
                    --- Albert Einstein ---
.