1237a5f84e16
tech



Hello Jim - 

CA>>> People think the 8250s and 16450s are less capable than
CA>>> they are. From my testing 57600 is not a problem for
CA>>> either of them. 

--8<--cut 

JH> No, my personal experience isn't relevant - - by the time I
JH> got a 14,400 external modem, the card I had it attached to
JH> had 15650's. 

You mean 16550As? 

CA>> I think just about anyone here knows your book is a bunch
CA>> of nonsense and I have personally tested them in the
CA>> manner I have stated with the results I have given. 

JH> I saw where you said you got 57K thru a null modem cable.
JH> Is that relevant to what I had been talking to Wayne C.
JH> about? 

The cable from the serial port to the external modem is in no
way going to alter the results of a null modem cable attached
to the same port. 

JH> What does your personal experience indicate with respect to
JH> external modems running under Windows or other, and using
JH> browser software to browse websites? 

Short answer: all of my personal modems are internals 

I was doing the null modem cable testing for another user who
communicated with me for a year or two regarding zmodem
transfer speeds and that he used his external modem exclusively
for telecom using DOS and W9x and was wondering if the older
UART was slowing his transfers. It was when he used a 4mhz 8088
machine but not by any significant percentage. 

JH> The book is now in its 14th edition, and has sold well over
JH> 2 million copies. The author has been saying that about
JH> 8250 chips for quite awhile (in my personal experience, I
JH> saw it in the 6th edition (1996) and in the 12th edition
JH> (2000)). If what I quoted above is truly drivel wrt using
JH> an external modem, seems like some sharp-shooting
JH> gunslinger would have blown the author out of the saddle
JH> over that by now. 

And was promptly ignored. Want more untested unrealistic
information? Ask Linux users what the minimum hardware required
for an install is. 

The 'giveaway' is the mention of 9600 baud which was actually a
limitation of the DOS INT14h that was designed to drive serial
printers and was never intended as a telecom interface. The
original W31 comm.drv was limited to this speed but can be
replaced by third party drivers that can go all the way to
115200. 

Many who never do what they write about make this mistake as
often as thinking the KERMIT transfer is properly set lacked at
a 120 byte packet size because they did not read _all_ of the
documentation, just the begining. ;-) 

JH> But its not too late for you to put a notch on your
JH> six-shooter - - in the book, it is suggested that - "If you
JH> have any questions about PC hardware, suggestions for the
JH> next version of the book, or any comments in general, send
JH> them to Scott via email at scottmueller{at}compuserve.com I'm
JH> sure he'd be glad to hear from you. 

Maybe I could rewrite parts of his book and become a
'footnote'! I think not. 

JH> After you finish that, there is more error out there which
JH> seems to need fixing, since I read this week at USRobotics
JH> website that "In general, if you have a 28.8 Kbps modem or
JH> faster, you will need a 16550 UART." So you may as well
JH> straighten them out also, while you're at it. 

Whoever wrote that has never tried it. 

I thought USR was bought out by 3COM then 3COM said there was
no money in modems and wasn't going to make them anymore? 

CA>> You need to get a better book or actually test some of
CA>> these things yourself. 

JH> I got myself into this by cautioning Wayne C. about going
JH> out and buying an- - "external 56 serial modem, TBOMK the
JH> 8250 won't support that speed. Corrections welcomed if I'm
JH> wrong." 

JH> So, perhaps you are correct about the null modem connection
JH> and I'm correct about the external 56K modem?? 

No, I'm right and you are wrong (and so is your book and USR's
website). 

JH> In any event, I haven't heard from you or anyone else here
JH> about successfully using browser software and surfing the
JH> net with an external 56K modem connected to an 8250 UART. 

What demands do you envision a browser will place upon the UART
that GSZ doing a zmodem transfer wouldn't? I don't get your
point? 

If GSZ can maintain theoretical max transfer rates using the
8250 and 16450 UARTs why do you think a lowly 'browser' is
going to increase the 'load'???? 

If it is your assumption that a graphical OS and browser will
rob CPU cycles and cause the UART to 'fault' then it's the CPU
and/or OS inability to properly multitask that is at fault and
not the UART. 

The 8250 and 16450 can manage 57600 as I said they can. 

JH> So, I don't really consider myself as standing corrected -
JH> - at least not yet. 

Suit yourself. I took the time to find out for myself. 

>
>        ,                          ,
>      o/      Charles.Angelich      \o       ,
>       __o/
>     / >          USA, MI           < \   __\__
 

___ * ATP/16bit 2.31 * 
... DOS the Ghost in the Machine! http://www.undercoverdesign.com/dosghost/

--- Maximus/2 3.01