On Mon, 08 Apr 2013 12:59:26 +0100, Dave Liquorice wrote:

On Mon, 08 Apr 2013 12:03:40 +0100, polygonum wrote:

The use of K/M and Ki/Mi is just trying to make it worse IMO.

I still think that a better approach would have been something like
Ksubscript 2 or Ksubscript 10) when people were being specific -

Why K? The SI prefix for a multiplier near 1000 is k.

As with bits and bytes (b/B). Why not k and K (1000/1024)?

As it happens, it's M (1000) and Mi (1024).

And k (1000) and Ki (1024).

The small k is a historical anomaly.


--
Use the BIG mirror service in the UK: http://www.mirrorservice.org
My posts (including this one) are my copyright and if @diy_forums on
Twitter wish to tweet them they can pay me £30 a post
*lightning surge protection* - a w_tom conductor

On 08/04/13 11:29, dennis@home wrote:
On 08/04/2013 00:51, The Natural Philosopher wrote:
On 08/04/13 00:09, Jeremy Nicoll - news posts wrote:
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for those
types
of memory that use one (not most domestic machines, so far as I know).



a BYTE is always 8 bits by definition. parity bits and other protocol
overheads so on are not part of the byte.



No it isn't.
Show your definition.

http://en.wikipedia.org/wiki/Byte

--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 12:10, bm wrote:
"dennis@home" wrote in message
.com...
On 08/04/2013 00:20, bm wrote:
"Jeremy Nicoll - news posts"
wrote
in message nvalid...
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for those
types
of memory that use one (not most domestic machines, so far as I know).
I think Dennis is including the start/stop bits. If he wrote his own
software UART he could have pretty much however many 'bits' (transmitted)
that he wants but he'll find there are 8 bits in a byte, 4 in a nibble
and
16 in a word (without the transport).


No, modems are measured in baud which is symbols per second not bits per
second. A symbol can contain many bits.
Well blow me down, for 40+ years i've understood baud-rate to be bits per
second, what a dummy.

snip


well of course it is. Dennis is talking out of his arse again.

he probably looked it up on wiki where someone who doesn't understand
communication theory has written an erroneous article.

As originally conceived it applied to telegraphy. It then morphed into
simple modems, but the whole concept became blurred when different
modulations schemas to which a level transition in a carrier state (or
baseband) could not be said to uniquely identify a single bit of
information,

Even his (and wikis) definition of 'symbol' means almost nothing these
days with QAM style modulations over multiple carrier frequencies. Or
worse still, spread spectrum convolutions.

which is why we now talk in terms of bps, not baud. baud as a concept
smply doesnt apply to ADSL Or any oher modern QAM or even weirder
modulation schemas.

--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 12:11, polygonum wrote:
On 08/04/2013 12:08, dennis@home wrote:

But there you are a bytes is not defined as 8 bits, it may be frequently
used to refer to 8 bits, it doesn't make it correct.

WHich part of this did you have difficulty in understanding:

"With ISO/IEC 80000-13, this common meaning was codified in a formal
standard."

http://www.iso.org/iso/catalogue_detail?csnumber=31898

you can buy the PDF for an exhorbitant sum and see for yourself.


--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 12:20, Dave Liquorice wrote:
On Mon, 08 Apr 2013 01:09:55 +0100, The Natural Philosopher wrote:

But I am currently getting 5282 net download, which if I applied 77%
to, would imply a synch speed of 6890!

I am synching at 6624 and I doubt my BRAS is over 6000. More like 5500.
which would be 96% 'efficiency'.
BRAS is set in 512 kbps steps from 1 Mbps or there abouts up. Or was I'm
not sure they still do that.

http://www.kitz.co.uk/adsl/IPprofile.htm

Depends on the service you have.

well no need to have BT tell me, because at my synch speed my BRAS is
5500 or less,

And that's pretty much what I seem to have as described above.

whether speed testes are adding protocol to give raw bps or its a cooked
bps that apples to actual payload transferred I cannot say.


--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 12:59, Dave Liquorice wrote:
On Mon, 08 Apr 2013 12:03:40 +0100, polygonum wrote:

The use of K/M and Ki/Mi is just trying to make it worse IMO.
I still think that a better approach would have been something like
Ksubscript 2 or Ksubscript 10) when people were being specific -
Why K? The SI prefix for a multiplier near 1000 is k.

As with bits and bytes (b/B). Why not k and K (1000/1024)?

I guess as a great many people don't seem able to grasp the significance
of the case of the prefixes that isn't a good idea. M or m?


or p and P
....



--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 14:26, Andy Champ wrote:
On 08/04/2013 11:17, dennis@home wrote:
A byte isn't eight bits.
An octet is 8 bits.
A byte is just a group of adjacent bits that may or may not have eight
of them.

I've worked on systems that didn't have 8 bit characters. We called
the 6-bit units characters on one of them, and the other one didn't
have an equivalent concept - a word was 36-bits. (and a character in a
file could be sixbit or ascii)

In neither case did we call them bytes.

Andy
exactly. the tendency was to call anything 8 bit a BYTE and 3anything
else a n bit word.

apart from 4 bit nibbles.


--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/2013 12:59, Dave Liquorice wrote:
On Mon, 08 Apr 2013 12:03:40 +0100, polygonum wrote:

The use of K/M and Ki/Mi is just trying to make it worse IMO.

I still think that a better approach would have been something like
Ksubscript 2 or Ksubscript 10) when people were being specific -

Why K? The SI prefix for a multiplier near 1000 is k.

As with bits and bytes (b/B). Why not k and K (1000/1024)?

I guess as a great many people don't seem able to grasp the significance
of the case of the prefixes that isn't a good idea. M or m?


My bad...

I agree.

--
Rod

On 08/04/2013 17:02, The Natural Philosopher wrote:

well of course it is. Dennis is talking out of his arse again.

he probably looked it up on wiki where someone who doesn't understand
communication theory has written an erroneous article.

As originally conceived it applied to telegraphy. It then morphed into
simple modems, but the whole concept became blurred when different
modulations schemas to which a level transition in a carrier state (or
baseband) could not be said to uniquely identify a single bit of
information,

Even his (and wikis) definition of 'symbol' means almost nothing these
days with QAM style modulations over multiple carrier frequencies. Or
worse still, spread spectrum convolutions.

which is why we now talk in terms of bps, not baud. baud as a concept
smply doesnt apply to ADSL Or any oher modern QAM or even weirder
modulation schemas.



And yet again TNP shows his ignorance.
Makes you wonder why nobody cares about his philosophy.

On 08/04/2013 16:50, The Natural Philosopher wrote:
On 08/04/13 11:29, dennis@home wrote:
On 08/04/2013 00:51, The Natural Philosopher wrote:
On 08/04/13 00:09, Jeremy Nicoll - news posts wrote:
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for those
types
of memory that use one (not most domestic machines, so far as I know).



a BYTE is always 8 bits by definition. parity bits and other protocol
overheads so on are not part of the byte.



No it isn't.
Show your definition.

http://en.wikipedia.org/wiki/Byte


Guess where this line appears?

"The term octet was defined to explicitly denote a sequence of 8 bits
because of the ambiguity associated at the time with the term byte.[4]"

On 08/04/2013 20:37, dennis@home wrote:
On 08/04/2013 16:50, The Natural Philosopher wrote:
On 08/04/13 11:29, dennis@home wrote:
On 08/04/2013 00:51, The Natural Philosopher wrote:
On 08/04/13 00:09, Jeremy Nicoll - news posts wrote:
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for those
types
of memory that use one (not most domestic machines, so far as I know).



a BYTE is always 8 bits by definition. parity bits and other protocol
overheads so on are not part of the byte.



No it isn't.
Show your definition.

http://en.wikipedia.org/wiki/Byte


Guess where this line appears?

"The term octet was defined to explicitly denote a sequence of 8 bits
because of the ambiguity associated at the time with the term byte.[4]"

Do you not recognise past tenses when they leap out, grab you by the
neck, and stuff you down a wire at any bit/baud/symbol or other rate?

--
Rod

On 08/04/13 20:37, dennis@home wrote:
On 08/04/2013 16:50, The Natural Philosopher wrote:
On 08/04/13 11:29, dennis@home wrote:
On 08/04/2013 00:51, The Natural Philosopher wrote:
On 08/04/13 00:09, Jeremy Nicoll - news posts wrote:
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for
those
types
of memory that use one (not most domestic machines, so far as I
know).



a BYTE is always 8 bits by definition. parity bits and other protocol
overheads so on are not part of the byte.



No it isn't.
Show your definition.

http://en.wikipedia.org/wiki/Byte


Guess where this line appears?

"The term octet was defined to explicitly denote a sequence of 8 bits
because of the ambiguity associated *at the time* with the term byte.[4]"

You can weaselbut you cant hide. later on it became standardised.


--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On 08/04/13 20:33, dennis@home wrote:
On 08/04/2013 17:02, The Natural Philosopher wrote:

well of course it is. Dennis is talking out of his arse again.

he probably looked it up on wiki where someone who doesn't understand
communication theory has written an erroneous article.

As originally conceived it applied to telegraphy. It then morphed into
simple modems, but the whole concept became blurred when different
modulations schemas to which a level transition in a carrier state (or
baseband) could not be said to uniquely identify a single bit of
information,

Even his (and wikis) definition of 'symbol' means almost nothing these
days with QAM style modulations over multiple carrier frequencies. Or
worse still, spread spectrum convolutions.

which is why we now talk in terms of bps, not baud. baud as a concept
smply doesnt apply to ADSL Or any oher modern QAM or even weirder
modulation schemas.



And yet again TNP shows his ignorance.

To the possesor of very little knowledge, far greater knowledge is
indistinguishable from ignorance.


Makes you wonder why nobody cares about his philosophy.


--
Ineptocracy

(in-ep-toc-ra-cy) €“ a system of government where the least capable to lead are elected by the least capable of producing, and where the members of society least likely to sustain themselves or succeed, are rewarded with goods and services paid for by the confiscated wealth of a diminishing number of producers.

On Mon, 08 Apr 2013 20:37:52 +0100, dennis@home wrote:

On 08/04/2013 16:50, The Natural Philosopher wrote:
On 08/04/13 11:29, dennis@home wrote:
On 08/04/2013 00:51, The Natural Philosopher wrote:
On 08/04/13 00:09, Jeremy Nicoll - news posts wrote:
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.
Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for
those types of memory that use one (not most domestic machines, so
far as I know).



a BYTE is always 8 bits by definition. parity bits and other protocol
overheads so on are not part of the byte.



No it isn't.
Show your definition.

http://en.wikipedia.org/wiki/Byte


Guess where this line appears?

"The term octet was defined to explicitly denote a sequence of 8 bits
because of the ambiguity associated at the time with the term byte.[4]"

"at the time" - pre-dating the standard.



--
Use the BIG mirror service in the UK: http://www.mirrorservice.org
My posts (including this one) are my copyright and if @diy_forums on
Twitter wish to tweet them they can pay me £30 a post
*lightning surge protection* - a w_tom conductor

http://en.wikipedia.org/wiki/Byte

- which starts by saying that a byte "most commonly consists of eight
bits.", but then goes on to give examples of when it does not...

--
Jeremy C B Nicoll - my opinions are my own.

Email sent to my from-address will be deleted. Instead, please reply
to replacing "aaa" by "284".

The Natural Philosopher wrote:
On 08/04/13 20:33, dennis@home wrote:
[snip]

And yet again TNP shows his ignorance.

To the possesor of very little knowledge, far greater knowledge is
indistinguishable from ignorance.

Pennis doesn't have very little knowledge. He's a rare case of negative
knowledge. Every "fact" he states will be wrong.

--
€¢DarWin|
_/ _/

Well like I pointed out even Cisco are inconsistent and use B for bit on
some products

Is this on their pro products or consumer stuff?

And is it in the technical literature (and consistently so, not just a
random typo), or in marketing literature?


--
Jeremy C B Nicoll - my opinions are my own.

Email sent to my from-address will be deleted. Instead, please reply
to replacing "aaa" by "284".

Jeremy Nicoll - news posts wrote:
Well like I pointed out even Cisco are inconsistent and use B for bit on
some products

Is this on their pro products or consumer stuff?

And is it in the technical literature (and consistently so, not just a
random typo), or in marketing literature?

It's some marketing ****e. But Pennis spends all his life grasping at
straws.

--
€¢DarWin|
_/ _/

On 08/04/2013 12:08, dennis@home wrote:
On 08/04/2013 11:36, polygonum wrote:
On 08/04/2013 11:17, dennis@home wrote:


A byte isn't eight bits.
An octet is 8 bits.
A byte is just a group of adjacent bits that may or may not have eight
of them.
Just because its a popular misconception doesn't mean its correct, it
just means you have to allow for the misinformed getting it wrong.
It doesn't really matter to 99.9% of people but it sure does if you are
designing kit that has to interoperate with other kit.

Of course the famous and rather expensive error was what does m mean,
metres or miles and just how much can you miss Mars by?

BTW a word isn't 16 bits either, its whatever width the computer was
designed to use. Popular word sizes have been 14 bits, 32 bits, 24 bits
and a few others.

If you today used the term 'byte' for anything other than eight bits,
you'd have to have a positively perverse aim of causing confusion.

Wiki suggests "The size of the byte has historically been hardware
dependent and no definitive standards existed that mandated the size.
The de facto standard of eight bits is a convenient power of two
permitting the values 0 through 255 for one byte. With ISO/IEC 80000-13,
this common meaning was codified in a formal standard."

As usual, the flaming standard seems not to be accessible. But willing
to believe Wiki on this matter.

If you ever use 'byte' to mean anything other than eight bits, it is up
to you as author to state that and explain. An unadorned, unexplained
'byte' simply IS eight bits.


Well I did say a professional wouldn't use those terms

Indeed, and you would be wrong.

To be fair it will depend a little on the particular area in which the
professional is working. If dealing with protocols at layer 2 and below,
then "octets" become useful to differentiate 8 bit sequences from coded
8 bit bytes which may be different sizes. However anyone using the term
"byte" without further qualification should have no expectation of it
being interpreted as anything other than 8 bits.

But there you are a bytes is not defined as 8 bits, it may be frequently
used to refer to 8 bits, it doesn't make it correct.

A byte historically could be other sizes. Now it is formally defined as
8 bits.



--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/

On 08/04/2013 12:10, bm wrote:
"dennis@home" wrote in message
.com...
On 08/04/2013 00:20, bm wrote:
"Jeremy Nicoll - news posts"
wrote
in message nvalid...
"dennis@home" wrote:

BTW maybe you can explain how many bits there are in a byte,
I will give you a hint it isn't eight.

Depends on the machine architecture, doesn't it?

It'll also depend on whether you're counting the parity bit, for those
types
of memory that use one (not most domestic machines, so far as I know).

I think Dennis is including the start/stop bits. If he wrote his own
software UART he could have pretty much however many 'bits' (transmitted)
that he wants but he'll find there are 8 bits in a byte, 4 in a nibble
and
16 in a word (without the transport).



No, modems are measured in baud which is symbols per second not bits per
second. A symbol can contain many bits.

Well blow me down, for 40+ years i've understood baud-rate to be bits per
second, what a dummy.

Just to make matters more complex, the term comes from information
theory, which mathematically quantifies the amount of information
transferred during an exchange. Hence its based on ratios of
probabilities of being able to guess the result both before and after
the information transfer. Since stuff dealing with ratios tends to be
expressed in logarithmic units they have their own units of information
quantity. If using natural logs, the unit is called a "nat", for base
ten the "hartley", and just to add confusion, the base two log unit is
called the "bit".

So baud, can legitimately be defined as "bits" per second, but the bits
in question, are not the 1/8th fragments of a byte that we traditionally
know and love! This can yield interesting discussion including phrases
like the "number of bits per bit". So you can see why datacomms guys go
for terms like "symbols" or "constellations" ;-)


--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/

On 08/04/2013 14:33, Andy Champ wrote:
On 08/04/2013 12:10, bm wrote:
Well blow me down, for 40+ years i've understood baud-rate to be bits per
second, what a dummy.

you are far from alone.

Another example is a parallel bus such as SCSI or a printer bus which
sends whole characters in one go. The baud rate is the rate at which
each wire sends bits - but the data rate is 8 (or 16, or 32) times higher.

Same applies to any parallel bus...


--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/

On 09/04/2013 13:15, polygonum wrote:
On 09/04/2013 10:38, John Rumm wrote:
On 08/04/2013 14:33, Andy Champ wrote:
On 08/04/2013 12:10, bm wrote:
Well blow me down, for 40+ years i've understood baud-rate to be bits
per
second, what a dummy.

you are far from alone.

Another example is a parallel bus such as SCSI or a printer bus which
sends whole characters in one go. The baud rate is the rate at which
each wire sends bits - but the data rate is 8 (or 16, or 32) times
higher.

Same applies to any parallel bus...


Each wire has a baud rate, that much I can agree. But am starting to
question the meaning of a baud rate for multiple wire connections - what
does it mean? Surely it will either be the sum of the individual wire
baud rates or, umm, something else, quite what does not immediately
appear in my mind?

The baud rate only tells you the signalling rate... not the number of
data bits carried per signal, or the number of additional symbols
carried by additional parallel data paths.

Hence the baud rate of 100Mb ethernet is the same as that of 1000Mb
ethernet. The higher data rate is achieved from coding more bits per
symbol, and also using four parallel paths instead of a single path.

So baud rates are useful when assessing spectral bandwidth requirements
for a channel, even if they are not the whole picture from the data
throughput perspective.



--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/

On 09/04/2013 17:17, John Rumm wrote:

8


The baud rate only tells you the signalling rate... not the number of
data bits carried per signal, or the number of additional symbols
carried by additional parallel data paths.

Hence the baud rate of 100Mb ethernet is the same as that of 1000Mb
ethernet. The higher data rate is achieved from coding more bits per
symbol, and also using four parallel paths instead of a single path.

Its two paths xmit and two paths receive isn't it.
Its also gigabit x two as its full duplex.

On 09/04/2013 20:51, dennis@home wrote:
On 09/04/2013 17:17, John Rumm wrote:

8


The baud rate only tells you the signalling rate... not the number of
data bits carried per signal, or the number of additional symbols
carried by additional parallel data paths.

Hence the baud rate of 100Mb ethernet is the same as that of 1000Mb
ethernet. The higher data rate is achieved from coding more bits per
symbol, and also using four parallel paths instead of a single path.

Its two paths xmit and two paths receive isn't it.

IIRC, its all 4 both ways with some nifty signal processing to excise
your sent traffic from what you receive... or am I thinking of 10gig?

Its also gigabit x two as its full duplex.

Yup, full duplex has been at least optional from pretty much the start
on twisted pair.


--
Cheers,

John.

/================================================== ===============\
| Internode Ltd - http://www.internode.co.uk |
|-----------------------------------------------------------------|
| John Rumm - john(at)internode(dot)co(dot)uk |
\================================================= ================/

On 10/04/2013 11:21, John Rumm wrote:
On 09/04/2013 20:51, dennis@home wrote:
On 09/04/2013 17:17, John Rumm wrote:

8


The baud rate only tells you the signalling rate... not the number of
data bits carried per signal, or the number of additional symbols
carried by additional parallel data paths.

Hence the baud rate of 100Mb ethernet is the same as that of 1000Mb
ethernet. The higher data rate is achieved from coding more bits per
symbol, and also using four parallel paths instead of a single path.

Its two paths xmit and two paths receive isn't it.

IIRC, its all 4 both ways with some nifty signal processing to excise
your sent traffic from what you receive... or am I thinking of 10gig?

Ah.
There are/were two standards 1000baseT and 1000baseTX.
The T variant uses four pairs at the same time, the TX variant doesn't.

It looks like most implementations are 1000baseT so do indeed use all
four pairs.

It looks like the baud rate is 125MHz and they transmit an octet at a
time (even though there are only four pairs).