Hardware modem query
Peter McVries
buying a controllerless/hardware based modem instead of using the onboard
one on my motherboard.
I understand that hardware modems do not require the cpu to do work for them
and so on, but surely on a reasonable pc (1Ghz cpu) this should be fairly
unimportant?
I'd be grateful if anyone with more knowledge about these modems could
provide me with info about why hardware modems are so much better to have -
if indeed they really are.
TIA.
Nigel Wade
It really depends on what you want them for.
IMHO, the first reason to use an external modem is that they have a power
switch which allows you to turn them off and be sure your modem can only
use the phone line when you want it to.
The second main reason is that pretty much any hardware modem is supported
in Linux where only a limited number of Winmodems are. If you don't care
about Linux this is irrelevent.
--
Nigel Wade
GSV Three Minds in a Can
Nigel Wade <n...@ion.le.ac.uk> said
The third reason is so you can cycle the power if the modem gets in a
knotted state, and the fourth is so you can watch der-blinken-lites and
detect whether the modem is really online or not, whether data is
actually moving, and what dumb connect speed it has fallen back to.
--
GSV Three Minds in a Can
Outgoing Msgs are Turing Tested,and indistinguishable from human typing.
Jim Noble
"GSV Three Minds in a Can" <GSV@[127.0.0.1]> wrote in message
news:xMPrBrB5...@from.is.invalid...
> the fourth is so you can watch der-blinken-lites and
> detect whether the modem is really online or not, whether data is
> actually moving, and what dumb connect speed it has fallen back to.
All of the above plus:-
Fifthly, when you forget to check and initiate a dial up when er
indoors is already using the phone, you can quickly cut the power and
avoid hearing disapproving words :).
Jim
Peter McVries
:-)
Sorry, but are you guys saying that *all* hardware modems are external?
I was under the impression it was possible to buy controllerless(hardware)
modems that were internal.
Ralph Wade Phillips
"Peter McVries" <just...@reply.com> wrote in message
news:b6shbs$ndt$1...@newsg2.svr.pol.co.uk...
> Sorry, but are you guys saying that *all* hardware modems are external?
Most all made today are - as a hint, if it's PCI, it's most likly a
controllerless, or software, modem.
>
> I was under the impression it was possible to buy controllerless(hardware)
> modems that were internal.
Eh. A full hardware modem has a controller - a controllerless modem
is, by definition, software-based.
But there are even DSP-less modems. I have one in my machine for
faxing ONLY. I'd not use it for any serious telecommunications, because it
sucks bigtime (planetary masses through stir sticks, I think.)
RwP
Alex Fraser
>
> I was under the impression it was possible to buy controllerless(hardware)
> modems that were internal.
Controllerless modems are usually sold as hardware modems; DSP-less modems
are sold as software modems. A controllerless modem still requires quite a
bit of software (but negligible CPU time) to operate, so many people, like
me, don't call them a hardware modem.
Alex
GSV Three Minds in a Can
Peter McVries <just...@reply.com> said
>Jim Noble wrote:
>> "GSV Three Minds in a Can" <GSV@[127.0.0.1]> wrote in message
>> news:xMPrBrB5...@from.is.invalid...
>>> the fourth is so you can watch der-blinken-lites and
>>> detect whether the modem is really online or not, whether data is
>>> actually moving, and what dumb connect speed it has fallen back to.
>>
>> All of the above plus:-
>>
>> Fifthly, when you forget to check and initiate a dial up when er
>> indoors is already using the phone, you can quickly cut the power and
>> avoid hearing disapproving words :).
>
>:-)
>Sorry, but are you guys saying that *all* hardware modems are external?
No, but all external modems are hardware modems .. which is yet another
reason they are preferable. Oh, one more, they only use a serial port
which was probably going to waste anyway, rather than expecting the OS
to find I/O addresses, IRQs, etc for yet another device (COM3, most
often).
--
GSV Three Minds in a Can
Jonathan Buzzard
"Ralph Wade Phillips" <ral...@techie.com> writes:
> Howdy!
>
> "Peter McVries" <just...@reply.com> wrote in message
> news:b6shbs$ndt$1...@newsg2.svr.pol.co.uk...
>
>> Sorry, but are you guys saying that *all* hardware modems are external?
>
> Most all made today are - as a hint, if it's PCI, it's most likly a
> controllerless, or software, modem.
>
>>
>> I was under the impression it was possible to buy controllerless(hardware)
>> modems that were internal.
>
> Eh. A full hardware modem has a controller - a controllerless modem
> is, by definition, software-based.
Really, explain which bit of the definition of a modem says that it
has to have an RS232 interface with a Hayes compatible AT command
processor? I can remember genuine hardware modems (not your
pseudo hardware ones you get today) that had neither.
JAB.
--
Jonathan A. Buzzard Email: jonathan at buzzard.me.uk
Northumberland, United Kingdom. Tel: +44 1661-832195
Jonathan Buzzard
But you would be of course wrong. Techincally no hardware modems have
been manufactured in years. They have all have been using D/A and A/D
convertors coupled to a DSP for ages.
Ralph Wade Phillips
"Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
news:kcvs6b...@192.168.42.254...
> In article <b6ske1$8e3lt$4...@id-81734.news.dfncis.de>,
> "Ralph Wade Phillips" <ral...@techie.com> writes:
> > Howdy!
> >
> > "Peter McVries" <just...@reply.com> wrote in message
> > news:b6shbs$ndt$1...@newsg2.svr.pol.co.uk...
> >
> >> Sorry, but are you guys saying that *all* hardware modems are external?
> >
> > Most all made today are - as a hint, if it's PCI, it's most
likly a
> > controllerless, or software, modem.
> >
> >>
> >> I was under the impression it was possible to buy
controllerless(hardware)
> >> modems that were internal.
> >
> > Eh. A full hardware modem has a controller - a controllerless
modem
> > is, by definition, software-based.
>
> Really, explain which bit of the definition of a modem says that it
> has to have an RS232 interface with a Hayes compatible AT command
> processor? I can remember genuine hardware modems (not your
> pseudo hardware ones you get today) that had neither.
And who said anything about a RS-232C interface?
Or Hayes AT command processors?
And, I know of a couple of modems that had neither - but in a PC
world, if it doesn't have something that fakes a 8250-family UART at a
standard address, it's going to take special software to talk to it anyway.
Thereby failing the "pure hardware" test of "Boot off of a DOS, Unix, Linux,
or some other command line OS diskette and pop text to it to control it".
RwP
Ralph Wade Phillips
"Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
news:5evs6b...@192.168.42.254...
> In article <b6splh$8m5o6$1...@id-149533.news.dfncis.de>,
> "Alex Fraser" <m...@privacy.net> writes:
> > "Peter McVries" <just...@reply.com> wrote in message
> > news:b6shbs$ndt$1...@newsg2.svr.pol.co.uk...
> >> Sorry, but are you guys saying that *all* hardware modems are external?
> >>
> >> I was under the impression it was possible to buy
controllerless(hardware)
> >> modems that were internal.
> >
> > Controllerless modems are usually sold as hardware modems; DSP-less
modems
> > are sold as software modems. A controllerless modem still requires quite
a
> > bit of software (but negligible CPU time) to operate, so many people,
like
> > me, don't call them a hardware modem.
> >
>
> But you would be of course wrong. Techincally no hardware modems have
> been manufactured in years. They have all have been using D/A and A/D
> convertors coupled to a DSP for ages.
<looks at USRobotics v.92 external modem in box> Really? Wonder
what THAT piece of equipment is, then.
Or the Robotics Performance Pro internal modem.
Jonathan, you happen to be wrong here. There ARE hardware modems
being made even today.
RwP
Alex Fraser
> "Alex Fraser" <m...@privacy.net> writes:
> > Controllerless modems are usually sold as hardware modems; DSP-less
> > modems are sold as software modems. A controllerless modem still
> > requires quite a bit of software (but negligible CPU time) to operate,
> > so many people, like me, don't call them a hardware modem.
>
> But you would be of course wrong.
To not call them a hardware modem?
Alex
Alex Fraser
news:b6udh1$8tbut$1...@ID-81734.news.dfncis.de...
> Howdy!
>
> "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> news:5evs6b...@192.168.42.254...
> > But you would be of course wrong. Techincally no hardware modems have
> > been manufactured in years. They have all have been using D/A and A/D
> > convertors coupled to a DSP for ages.
>
> <looks at USRobotics v.92 external modem in box> Really? Wonder
> what THAT piece of equipment is, then.
Most likely a few bits for the line and serial interfaces, an A/D, a D/A, a
DSP, a microcontroller or processor of some sort, and some firmware. Does
that make it hardware or software?
How about a broadband router with various Ethernet-related
components/subsytems plus a RISC CPU running some embedded OS running NAT
routing software?
> Or the Robotics Performance Pro internal modem.
>
> Jonathan, you happen to be wrong here. There ARE hardware modems
> being made even today.
Seems to me there's fundamentally a continuous line between hardware and
software, and Jonathan has picked an arbitrary point along it. I guess
(partly from your other post) what you mean by "hardware" is "does not
require software running on the attached computer's CPU" - whether
technically inaccurate or not, that seems a fairly sensible definition to
me.
Alex
Jonathan Buzzard
GSV Three Minds in a Can <GSV@[127.0.0.1]> writes:
[SNIP]
>
> No, but all external modems are hardware modems .. which is yet another
> reason they are preferable. Oh, one more, they only use a serial port
> which was probably going to waste anyway, rather than expecting the OS
> to find I/O addresses, IRQs, etc for yet another device (COM3, most
> often).
>
In fact even this is an inaccurate statement.
Jonathan Buzzard
"Ralph Wade Phillips" <ral...@techie.com> writes:
[SNIP]
>
> <looks at USRobotics v.92 external modem in box> Really? Wonder
> what THAT piece of equipment is, then.
>
> Or the Robotics Performance Pro internal modem.
>
> Jonathan, you happen to be wrong here. There ARE hardware modems
> being made even today.
>
No there are not. Inside that box, is a D/A converter a A/D convertor,
a RS232 level converter a *DSP chip* and an EEPROM, with a few
passives and a DC voltage regulator.
The modem works by software running on that DSP chip. It is no
different from a PCI card that has a DSP chip on it that does
all the work, but instead of using the outdated RS232 interface
and running a Hayes compatible AT command processor as well,
decides to use the PCI bus and a software driver on the computer
to simulate the AT command processor (which by the way is only
used these days for telling the modem which number to dial).
However once upon a time, long ago in the days of 14k4, 9600 and
older modems, there where custom hardware silicon circuits that
did the MODulation and DEModulation, no sign of a DSP and not
a line of code anywhere in sight.
Jonathan Buzzard
"Alex Fraser" <m...@privacy.net> writes:
[SNIP]
>
> Seems to me there's fundamentally a continuous line between hardware and
> software, and Jonathan has picked an arbitrary point along it.
No I have picked the only correct point on it. Any other point chosen
can only be done when not in possesion of the full facts.
> I guess
> (partly from your other post) what you mean by "hardware" is "does not
> require software running on the attached computer's CPU" - whether
> technically inaccurate or not, that seems a fairly sensible definition to
> me.
>
Explain to me the difference between a serial port driver and a PCI
driver? The answer is none. All modems require software drivers to
interface to them. Just because one type use a long standing,
defacto standard, *does* not differentiate them in a technical
perspective from one that does not.
Jonathan Buzzard
Why should a modem have to use a legacy hardware interface to
pass your incorrect definition of "pure hardware"?
GSV Three Minds in a Can
Jonathan Buzzard <jona...@buzzard.org.uk> said
>In article <A0XnPBCa...@from.is.invalid>,
> GSV Three Minds in a Can <GSV@[127.0.0.1]> writes:
>
>[SNIP]
>>
>> No, but all external modems are hardware modems .. which is yet another
>> reason they are preferable. Oh, one more, they only use a serial port
>> which was probably going to waste anyway, rather than expecting the OS
>> to find I/O addresses, IRQs, etc for yet another device (COM3, most
>> often).
>>
>
>In fact even this is an inaccurate statement.
Feel free to expound further .. btw, I was talking =Serial Port=
external modems, so don't drag in USB/firewire jobbies to confuse
matters.
--
GSV Three Minds in a Can
Ralph Wade Phillips
"Alex Fraser" <m...@privacy.net> wrote in message
news:b6ujf2$9a7hr$1...@ID-149533.news.dfncis.de...
> "Ralph Wade Phillips" <ral...@techie.com> wrote in message
> news:b6udh1$8tbut$1...@ID-81734.news.dfncis.de...
> > Howdy!
> >
> > "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> > news:5evs6b...@192.168.42.254...
> [snip]
> > > But you would be of course wrong. Techincally no hardware modems have
> > > been manufactured in years. They have all have been using D/A and A/D
> > > convertors coupled to a DSP for ages.
> >
> > <looks at USRobotics v.92 external modem in box> Really?
Wonder
> > what THAT piece of equipment is, then.
>
> Most likely a few bits for the line and serial interfaces, an A/D, a D/A,
a
> DSP, a microcontroller or processor of some sort, and some firmware. Does
> that make it hardware or software?
Hardware. It takes RS-232 in, shoves analog tones out. It also
takes analog tones in, and shoves RS-232 out. It works wonderfully with a
dumb printer as a remote printing terminal.
What else DO you expect it to mean?
It doesn't rely on a Windows driver to emulate a UART and/or a DSP,
in case you're wondering.
>
> How about a broadband router with various Ethernet-related
> components/subsytems plus a RISC CPU running some embedded OS running NAT
> routing software?
Doesn't count in this discussion. Might as well ask about the
optical modems used on optical "fly-by-wire" systems in jet fighters. Has
NOTHING to do with dialup access.
>
> > Or the Robotics Performance Pro internal modem.
> >
> > Jonathan, you happen to be wrong here. There ARE hardware
modems
> > being made even today.
>
> Seems to me there's fundamentally a continuous line between hardware and
> software, and Jonathan has picked an arbitrary point along it. I guess
> (partly from your other post) what you mean by "hardware" is "does not
> require software running on the attached computer's CPU" - whether
> technically inaccurate or not, that seems a fairly sensible definition to
> me.
And it's the best one there is.
RwP
Ralph Wade Phillips
"Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
news:ge7v6b...@192.168.42.254...
> In article <b6udh1$8tbut$1...@id-81734.news.dfncis.de>,
> "Ralph Wade Phillips" <ral...@techie.com> writes:
>
> [SNIP]
> >
> > <looks at USRobotics v.92 external modem in box> Really?
Wonder
> > what THAT piece of equipment is, then.
> >
> > Or the Robotics Performance Pro internal modem.
> >
> > Jonathan, you happen to be wrong here. There ARE hardware
modems
> > being made even today.
> >
>
> No there are not. Inside that box, is a D/A converter a A/D convertor,
> a RS232 level converter a *DSP chip* and an EEPROM, with a few
> passives and a DC voltage regulator.
And requires no fancy drivers on the host computer or device. And
can even work with a RS-232 interfaced terminal. Or printer. Or printing
terminal.
>
> The modem works by software running on that DSP chip. It is no
> different from a PCI card that has a DSP chip on it that does
> all the work, but instead of using the outdated RS232 interface
> and running a Hayes compatible AT command processor as well,
> decides to use the PCI bus and a software driver on the computer
> to simulate the AT command processor (which by the way is only
> used these days for telling the modem which number to dial).
More than that, on some installs.
>
> However once upon a time, long ago in the days of 14k4, 9600 and
> older modems, there where custom hardware silicon circuits that
> did the MODulation and DEModulation, no sign of a DSP and not
> a line of code anywhere in sight.
Even by the 14K4 days, DSPs were in use. Even if they were
dedicated modem-only DSPs. That was just about the only way to do
quadrature phase modulation - it's rather hard to tune a batch of RC filters
for that.
RwP
Jonathan Buzzard
"Ralph Wade Phillips" <ral...@techie.com> writes:
[SNIP]
>
> And requires no fancy drivers on the host computer or device. And
> can even work with a RS-232 interfaced terminal. Or printer. Or printing
> terminal.
>
Only because a standard has been agreed for the interface that allows
the OS to include a driver that conforms to that standard.
If your computer does not have a serial port that looks like an 8250
derivative then you need a fancy driver on the host computer...
Jonathan Buzzard
GSV Three Minds in a Can <GSV@[127.0.0.1]> writes:
> Bitstring <g15v6b...@192.168.42.254>, from the wonderful person
> Jonathan Buzzard <jona...@buzzard.org.uk> said
>>In article <A0XnPBCa...@from.is.invalid>,
>> GSV Three Minds in a Can <GSV@[127.0.0.1]> writes:
>>
>>[SNIP]
>>>
>>> No, but all external modems are hardware modems .. which is yet another
>>> reason they are preferable. Oh, one more, they only use a serial port
>>> which was probably going to waste anyway, rather than expecting the OS
>>> to find I/O addresses, IRQs, etc for yet another device (COM3, most
>>> often).
>>>
>>
>>In fact even this is an inaccurate statement.
>
> Feel free to expound further .. btw, I was talking =Serial Port=
> external modems, so don't drag in USB/firewire jobbies to confuse
> matters.
>
Yep, as I understand Apple did a serial port external modem that
was effectively just a D/A and A/D converter coupled to the telephone
line, with all the DSP being done on the host computer. It did not
last long however.
Alex Fraser
> In article <b6ujf2$9a7hr$1...@id-149533.news.dfncis.de>,
> "Alex Fraser" <m...@privacy.net> writes:
> [SNIP]
> > Seems to me there's fundamentally a continuous line between hardware and
> > software, and Jonathan has picked an arbitrary point along it.
>
> No I have picked the only correct point on it.
And that is?
[snip]
> Explain to me the difference between a serial port driver and a PCI
> driver? The answer is none.
I agree. The drivers for both a controllerless and DSP-less modem do a lot
more than shuffe data as-is to/from the application from/to a serial port,
it's this "extra" software I was referring to.
Alex
Alex Fraser
> Howdy!
>
> "Alex Fraser" <m...@privacy.net> wrote in message
> news:b6ujf2$9a7hr$1...@ID-149533.news.dfncis.de...
> > "Ralph Wade Phillips" <ral...@techie.com> wrote in message
> > news:b6udh1$8tbut$1...@ID-81734.news.dfncis.de...
> > > <looks at USRobotics v.92 external modem in box> Really?
> > > Wonder what THAT piece of equipment is, then.
> >
> > Most likely a few bits for the line and serial interfaces, an A/D, a
> > D/A, a DSP, a microcontroller or processor of some sort, and some
> > firmware. Does that make it hardware or software?
>
> Hardware. It takes RS-232 in, shoves analog tones out. It also
> takes analog tones in, and shoves RS-232 out. It works wonderfully with a
> dumb printer as a remote printing terminal.
>
> What else DO you expect it to mean?
There's a fair amount of software to make it work, how can it be considered
"pure hardware" as you put it elsewhere?
[snip]
> > How about a broadband router with various Ethernet-related
> > components/subsytems plus a RISC CPU running some embedded OS running
> > NAT routing software?
>
> Doesn't count in this discussion.
It's perhaps a more obvious example than a modem that the distinction isn't
clear-cut.
Alex
Jonathan Buzzard
"Alex Fraser" <m...@privacy.net> writes:
> "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> news:vh7v6b...@192.168.42.254...
>> In article <b6ujf2$9a7hr$1...@id-149533.news.dfncis.de>,
>> "Alex Fraser" <m...@privacy.net> writes:
>> [SNIP]
>> > Seems to me there's fundamentally a continuous line between hardware and
>> > software, and Jonathan has picked an arbitrary point along it.
>>
>> No I have picked the only correct point on it.
>
> And that is?
>
> [snip]
>> Explain to me the difference between a serial port driver and a PCI
>> driver? The answer is none.
>
> I agree. The drivers for both a controllerless and DSP-less modem do a lot
> more than shuffe data as-is to/from the application from/to a serial port,
> it's this "extra" software I was referring to.
>
I am lost with your logic. On the one had we have modems with serial
ports, that require software drivers to suffle information over
the said serial port to the modem in order for them to work. On
the other hand we have PCI card modems which have onboard DSP's
that does all the work, which also have software drivers that
shuffle the information this time over the PCI bus to the modem.
What's the difference? Well as far as I can tell from a technical
perspective none really.
In fact if you are into gaming a PCI modem with a DSP would
be better than one with a serial port interface as they should
have much lower latency.
I will also add in when you consider ISDN modems most people seem
to prefer internal modems, which do exactly what a PCI modem with
a DSP does, and drops the serial port interface and implements
the Hayes AT command processor in the driver.
Alex Fraser
> In article <b70nuf$9qjsl$1...@id-149533.news.dfncis.de>,
> "Alex Fraser" <m...@privacy.net> writes:
> > "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> > news:vh7v6b...@192.168.42.254...
> >> In article <b6ujf2$9a7hr$1...@id-149533.news.dfncis.de>,
> >> "Alex Fraser" <m...@privacy.net> writes:
> >> [SNIP]
> >> > Seems to me there's fundamentally a continuous line between hardware
> >> > and software, and Jonathan has picked an arbitrary point along it.
> >>
> >> No I have picked the only correct point on it.
> >
> > And that is?
No answer?
[snip]
> I am lost with your logic.
[snip]
I'll try to make it more obvious then.
With a normal serial port modem, the software on the computer shuffles the
data around as-is, no additional processing. Couldn't be simpler.
With a DSP-less modem, the software on the computer performs signal
processing, implements the AT command interpreter, and handles all the
link-level protocols including connection negotiation, speed shifts, error
correction and data compression. Plus oddments like going off/on-hook and
generating DTMF tones (and loop disconnect dialing, for that matter).
A controllerless modem is the same, except the signal processing is
offloaded to a DSP. Reading your posts, you seem to believe that a
controllerless modem only has the command interpreter in the driver - this
is not so.
> In fact if you are into gaming a PCI modem with a DSP would
> be better than one with a serial port interface as they should
> have much lower latency.
Why, as a matter of interest? It's not the case in practice - the best of
each that I have seen are near enough identical, and in general,
controllerless PCI modems have slightly higher latency. There are bigger
differences between different modems, and the same modem can have much
better/worse latency depending on the ISP equipment.
Alex
Conor
says...
> The modem works by software running on that DSP chip. It is no
> different from a PCI card that has a DSP chip on it that does
> all the work, but instead of using the outdated RS232 interface
> and running a Hayes compatible AT command processor as well,
> decides to use the PCI bus and a software driver on the computer
> to simulate the AT command processor (which by the way is only
> used these days for telling the modem which number to dial).
>
<snip>
Still knackered if you lose the driver disks though and its not
supported in Windows...
--
________________________
Conor Turton
conor_...@hotmail.com
ICQ:31909763
________________________
Conor
says...
> If your computer does not have a serial port that looks like an 8250
> derivative then you need a fancy driver on the host computer...
>
HAHAHAHAHAHAHA....
Ralph Wade Phillips
"Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
news:r8j07b...@192.168.42.254...
> In article <b6vn09$9kbgi$1...@id-81734.news.dfncis.de>,
> "Ralph Wade Phillips" <ral...@techie.com> writes:
>
> [SNIP]
> >
> > And requires no fancy drivers on the host computer or device.
And
> > can even work with a RS-232 interfaced terminal. Or printer. Or
printing
> > terminal.
> >
>
> Only because a standard has been agreed for the interface that allows
> the OS to include a driver that conforms to that standard.
And, what OS does a Genicom printer run? Or a Wyse-50 terminal?
You keep bouncing around and don't seem to even be able to read what
I've typed.
Oh - I've got one for you - DecWriter III. Praytell, what OS THAT
puppy runs?
>
> If your computer does not have a serial port that looks like an 8250
> derivative then you need a fancy driver on the host computer...
Which was part of my point. Which said point you want to argue
with.
Numbnuts.
Crawl back under your bridge.
RwP
Jonathan Buzzard
Conor <conor_...@hotmail.com> writes:
> In article <ge7v6b...@192.168.42.254>, jona...@buzzard.org.uk
> says...
>
>> The modem works by software running on that DSP chip. It is no
>> different from a PCI card that has a DSP chip on it that does
>> all the work, but instead of using the outdated RS232 interface
>> and running a Hayes compatible AT command processor as well,
>> decides to use the PCI bus and a software driver on the computer
>> to simulate the AT command processor (which by the way is only
>> used these days for telling the modem which number to dial).
>>
> <snip>
> Still knackered if you lose the driver disks though and its not
> supported in Windows...
>
Didn't say otherwise, and it is not relevant to the question.
DayDragon
> I will also add in when you consider ISDN modems most people seem
> to prefer internal modems, which do exactly what a PCI modem with
> a DSP does, and drops the serial port interface and implements
> the Hayes AT command processor in the driver.
>
use a modem, it uses a terminal adapter (or is it adaptor?).
Jonathan Buzzard
No but the point was that they look on the software side much like
modems. Most for example will accept AT commands etc. etc. without
actually having an Hayes AT command processor onboard...
Jonathan Buzzard
"Ralph Wade Phillips" <ral...@techie.com> writes:
[SNIP]
>
> And, what OS does a Genicom printer run?
None, but it is irrelevant, as they don't have a PCI slot. Are you
saying the ISA and PCI internal modems that I have which have
serial UART's and a Hayes AT command processor don't qualify either.
However it does have hardware that has been specially designed to
talk to the RS232 port.
> Or a Wyse-50 terminal?
This does not run an OS per say, but it *does* have a terminal program
that runs on an internal processor, which has software written
that can drive whatever the serial UART is in the machine.
>
> Oh - I've got one for you - DecWriter III. Praytell, what OS THAT
> puppy runs?
See above.
>>
>> If your computer does not have a serial port that looks like an 8250
>> derivative then you need a fancy driver on the host computer...
>
> Which was part of my point. Which said point you want to argue with.
Just because the fancy driver comes with the OS or built into the
hardware does not change the facts.
Jonathan Buzzard
"Alex Fraser" <m...@privacy.net> writes:
> "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> news:5st07b...@192.168.42.254...
>> In article <b70nuf$9qjsl$1...@id-149533.news.dfncis.de>,
>> "Alex Fraser" <m...@privacy.net> writes:
>> > "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
>> > news:vh7v6b...@192.168.42.254...
>> >> In article <b6ujf2$9a7hr$1...@id-149533.news.dfncis.de>,
>> >> "Alex Fraser" <m...@privacy.net> writes:
>> >> [SNIP]
>> >> > Seems to me there's fundamentally a continuous line between hardware
>> >> > and software, and Jonathan has picked an arbitrary point along it.
>> >>
>> >> No I have picked the only correct point on it.
>> >
>> > And that is?
>
> No answer?
A true hardware modem is one which does not have a DSP in it. Anything
else is really a software modem of some description.
> [snip]
>> I am lost with your logic.
> [snip]
>
> I'll try to make it more obvious then.
>
> With a normal serial port modem, the software on the computer shuffles the
> data around as-is, no additional processing. Couldn't be simpler.
Not quite.
> With a DSP-less modem, the software on the computer performs signal
> processing, implements the AT command interpreter, and handles all the
> link-level protocols including connection negotiation, speed shifts, error
> correction and data compression. Plus oddments like going off/on-hook and
> generating DTMF tones (and loop disconnect dialing, for that matter).
>
> A controllerless modem is the same, except the signal processing is
> offloaded to a DSP. Reading your posts, you seem to believe that a
> controllerless modem only has the command interpreter in the driver - this
> is not so.
Well yes they sometimes do things like v42 and v42.bis in the driver.
However the reason for this is that they are basically defunct in
99.99% of modems which are are used exclusively for internet access
and the ppp protocol provides better error correction and compression.
So lets see how do you dial a number, well on an external serial port
modem you send some AT command. On a PCI one you fiddle with some
registers over the PCI bus. Conceptually there is no difference.
>> In fact if you are into gaming a PCI modem with a DSP would
>> be better than one with a serial port interface as they should
>> have much lower latency.
>
> Why, as a matter of interest?
Because the latency of the PCI bus is *much* lower than that of
a serial port. In a modern PC the serial port is sitting of an
ISA/PCI bridge. So a byte arrives at the modem, in one it is
shunted over the serial port at 115200bps, then squeezed through
the ISA/PCI bridge before reaching the CPU. On the other it is
squirted directly through the PCI bus...
> It's not the case in practice - the best of
> each that I have seen are near enough identical, and in general,
> controllerless PCI modems have slightly higher latency. There are bigger
> differences between different modems, and the same modem can have much
> better/worse latency depending on the ISP equipment.
Quite probably, I said 'should' which is an euphonism for 'in theory'.
Alex Fraser
> In article <b7154b$9vast$1...@id-149533.news.dfncis.de>,
> "Alex Fraser" <m...@privacy.net> writes:
> > With a normal serial port modem, the software on the computer shuffles
> > the data around as-is, no additional processing. Couldn't be simpler.
>
> Not quite.
The crucial points are "no additional processing" and that is that it's
tough to imagine a driver that's less complex.
> > With a DSP-less modem, the software on the computer performs signal
> > processing, implements the AT command interpreter, and handles all the
> > link-level protocols including connection negotiation, speed shifts,
> > error correction and data compression. Plus oddments like going
> > off/on-hook and generating DTMF tones (and loop disconnect dialing, for
> > that matter).
> >
> > A controllerless modem is the same, except the signal processing is
> > offloaded to a DSP. Reading your posts, you seem to believe that a
> > controllerless modem only has the command interpreter in the driver -
> > this is not so.
>
> Well yes they sometimes do things like v42 and v42.bis in the driver.
I don't know of one that doesn't.
> However the reason for this is that they are basically defunct in
> 99.99% of modems which are are used exclusively for internet access
> and the ppp protocol provides better error correction and compression.
What error correction? How many ISPs support compression?
> So lets see how do you dial a number, well on an external serial port
> modem you send some AT command. On a PCI one you fiddle with some
> registers over the PCI bus. Conceptually there is no difference.
At one level, yes. But the difference is in the scale of the operation. One
driver feeds the number for the modem to dial, the other tells it to go off
hook, then use the DSP to help it detect the dialtone, then tell it to
produce a couple of tones, then stop, then produce another couple of tones,
then stop, ...
And that's just dialing. Do you get the picture?
> >> In fact if you are into gaming a PCI modem with a DSP would
> >> be better than one with a serial port interface as they should
> >> have much lower latency.
> >
> > Why, as a matter of interest?
>
> Because the latency of the PCI bus is *much* lower than that of
> a serial port.
Sure, but do you have figures? If my mental arithmetic is right, the serial
port could have a hundred times the latency and it wouldn't be significant
(a few milliseconds at most).
Alex
Ralph Wade Phillips
"Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
news:ulu17b...@192.168.42.254...
> In article <b71g6e$95gff$1...@id-81734.news.dfncis.de>,
> "Ralph Wade Phillips" <ral...@techie.com> writes:
>
> [SNIP]
> >
> > And, what OS does a Genicom printer run?
>
> None, but it is irrelevant, as they don't have a PCI slot. Are you
> saying the ISA and PCI internal modems that I have which have
> serial UART's and a Hayes AT command processor don't qualify either.
> However it does have hardware that has been specially designed to
> talk to the RS232 port.
No - YOU claimed that the EXTERNAL USRobotics v.92 modem !I!
mentioned was !NOT! a hardware modem.
Shall I go back and requote what you handily snipped, to avoid
'fessing up that you are busy trying to sound like an idiot?
>
> > Or a Wyse-50 terminal?
>
> This does not run an OS per say, but it *does* have a terminal program
> that runs on an internal processor, which has software written
> that can drive whatever the serial UART is in the machine.
And that affects the modem how? Oh, I see - the modem has to have a
batch of hardware, and can't be just software diddling the DSP now. LIKE
THE USRobotics EXTERNAL I MENTIONED. THAT YOU CLAIMED ISN'T MADE ANYMORE,
AND HASN'T BEEN FOR YEARS.
>
> >
> > Oh - I've got one for you - DecWriter III. Praytell, what OS
THAT
> > puppy runs?
>
>
> See above.
See above again.
>
> >>
> >> If your computer does not have a serial port that looks like an 8250
> >> derivative then you need a fancy driver on the host computer...
> >
> > Which was part of my point. Which said point you want to argue
with.
>
> Just because the fancy driver comes with the OS or built into the
> hardware does not change the facts.
What facts? That you seem to be so anxious to split hairs, that you
want to look like an idiot?
RwP
Jonathan Buzzard
"Alex Fraser" <m...@privacy.net> writes:
[SNIP]
>
> What error correction? How many ISPs support compression?
Most I have used. Simple answer is though that most binaries are
already compressed so they don't compress anymore. However text
does compress. If I download lots of text (say emails or usenet)
I frequently get sustained transfer rates of over 10KBps. Simply
not possible with dialup without compression.
>> So lets see how do you dial a number, well on an external serial port
>> modem you send some AT command. On a PCI one you fiddle with some
>> registers over the PCI bus. Conceptually there is no difference.
>
> At one level, yes. But the difference is in the scale of the operation. One
> driver feeds the number for the modem to dial, the other tells it to go off
> hook, then use the DSP to help it detect the dialtone, then tell it to
> produce a couple of tones, then stop, then produce another couple of tones,
> then stop, ...
> And that's just dialing. Do you get the picture?
No, I don't because the first operation with most of these DSP
based controllerless modems drivers is to say hey, here is your
firmware, run that. Then to dial a number you send a few fairly
simple commands to the DSP.
Now with a serial port, you have to put the serial port into
the correct mode, and then send a series of characters one at
a time, in a specially formatted manner.
Still don't see the difference conceptually. It is just you are
used to one.
>> >> In fact if you are into gaming a PCI modem with a DSP would
>> >> be better than one with a serial port interface as they should
>> >> have much lower latency.
>> >
>> > Why, as a matter of interest?
>>
>> Because the latency of the PCI bus is *much* lower than that of
>> a serial port.
>
> Sure, but do you have figures? If my mental arithmetic is right, the serial
> port could have a hundred times the latency and it wouldn't be significant
> (a few milliseconds at most).
>
Yes but gaming types seem to obsesse about a few milliseconds.
Lets face what is the point of having a computer that does 120fps
in Unreal Tournament?
Jonathan Buzzard
"Ralph Wade Phillips" <ral...@techie.com> writes:
> Howdy!
>
> "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> news:ulu17b...@192.168.42.254...
>> In article <b71g6e$95gff$1...@id-81734.news.dfncis.de>,
>> "Ralph Wade Phillips" <ral...@techie.com> writes:
>>
>> [SNIP]
>> >
>> > And, what OS does a Genicom printer run?
>>
>> None, but it is irrelevant, as they don't have a PCI slot. Are you
>> saying the ISA and PCI internal modems that I have which have
>> serial UART's and a Hayes AT command processor don't qualify either.
>> However it does have hardware that has been specially designed to
>> talk to the RS232 port.
>
> No - YOU claimed that the EXTERNAL USRobotics v.92 modem !I!
> mentioned was !NOT! a hardware modem.
Just because you cannot see the software does not mean it does
not exist...
>
> And that affects the modem how? Oh, I see - the modem has to have a
> batch of hardware, and can't be just software diddling the DSP now. LIKE
> THE USRobotics EXTERNAL I MENTIONED. THAT YOU CLAIMED ISN'T MADE ANYMORE,
> AND HASN'T BEEN FOR YEARS.
Because inside that modem there is a DSP chip running a program.
You can probably go the USRobotics website, and download an
update for the program to upgrade it to the latest version. It
is technical a software modem, in that it opperates by running
software.
Way back when, modems had dedicated filter chips and hardwired
digital logic chips inside them, with not a DSP's in sight.
These are the *ONLY* true hardware modems, and nobody has made
a modem like this for around a decade.
Alex Fraser
> In article <b729vt$ainpo$1...@id-149533.news.dfncis.de>,
> "Alex Fraser" <m...@privacy.net> writes:
> [SNIP]
> > What error correction? How many ISPs support compression?
>
> Most I have used.
You've removed the context of my question... I'm not sure whether you are
answering about PPP-based compression or V.42bis. I meant the former.
Assuming you meant the latter, which ISPs?
[snip]
> Still don't see the difference conceptually.
As I have already said, at one level there isn't. At that level, you pretty
much generalise all peripherals, PCI cards, and so on that you attach to a
PC. The difference is in the details that are specific to a modem, visible
in the specific nature of the hardware/driver interface, which corresponds
to much lesser/greater complexity in the driver itself.
[snip]
> >> Because the latency of the PCI bus is *much* lower than that of
> >> a serial port.
> >
> > Sure, but do you have figures? If my mental arithmetic is right, the
> > serial port could have a hundred times the latency and it wouldn't be
> > significant (a few milliseconds at most).
>
> Yes but gaming types seem to obsesse about a few milliseconds.
Of course, that doesn't mean there's actually a difference. To put it
another way, the variation in latency from moment to moment will generally
be similar or even larger (particularly with a dial-up modem because of
errors causing retransmissions).
> Lets face what is the point of having a computer that does 120fps
> in Unreal Tournament?
Actually, there is some truth to the benefit of apparently overkill
framerates. If you normally get 120fps, the chances are the framerate will
never be low enough to be an issue. The framerate inevitably gives rise to
some latency between your input and the visible response, subject to
vertical refresh.
Alex
Jonathan Buzzard
"Alex Fraser" <m...@privacy.net> writes:
> "Jonathan Buzzard" <jona...@buzzard.org.uk> wrote in message
> news:tpk47b...@192.168.42.254...
>> In article <b729vt$ainpo$1...@id-149533.news.dfncis.de>,
>> "Alex Fraser" <m...@privacy.net> writes:
>> [SNIP]
>> > What error correction? How many ISPs support compression?
>>
>> Most I have used.
>
> You've removed the context of my question... I'm not sure whether you are
> answering about PPP-based compression or V.42bis. I meant the former.
> Assuming you meant the latter, which ISPs?
Nobody uses V.42bis because PPP provides better compression. Most ISP's
seem to provide PPP with compression as far as I can tell.
>
>> Lets face what is the point of having a computer that does 120fps
>> in Unreal Tournament?
>
> Actually, there is some truth to the benefit of apparently overkill
> framerates. If you normally get 120fps, the chances are the framerate will
> never be low enough to be an issue. The framerate inevitably gives rise to
> some latency between your input and the visible response, subject to
> vertical refresh.
Perhaps but as the eye is not going to see anything much beyond
30fps it is pointless, so 120fps is just "willy waving".
Alex Fraser
> Nobody uses V.42bis because PPP provides better compression.
They are broadly similar. The fact that serial ports still typically work at
115200bps can be a limit to V.42bis, but not as much as you might
immediately expect because of latency and incompressible data.
> Most ISP's seem to provide PPP with compression as far as I can tell.
The situation may have changed in the past few years, but that /was/ not the
case. All that I have used supported V.42bis.
[snip]
> > Actually, there is some truth to the benefit of apparently overkill
> > framerates. If you normally get 120fps, the chances are the framerate
> > will never be low enough to be an issue. [...]
>
> Perhaps but as the eye is not going to see anything much beyond
> 30fps it is pointless, so 120fps is just "willy waving".
See above regarding typical and minimum framerates. I don't know why a film
in the cinema at 24fps looks fine yet playing a game at the same framerate
is horrible, but my best guess would be a combination of the inevitable
latency and the lack of motion blur.
Alex