-=> /* Quoting Bob Juge to Frank Sexton */ <=-
 > Have a look at an interesting article on the x2 /
 > K56Flex / v.90 state of affairs at Boardwatch magazine's
 > Web page:
 DH>> http://www.boardwatch.com/mag/98/mar/bwm24.html
 
 FS>> I've seen it. It says that x2 is much better than
 FS>> K56Flex. Doesn't mean much now that v.90 is here.
 
 BJ>> Read it again.
 
 FS> What are you getting at?
 BJ> Reread this part of the article:
 BJ> === Cut ===
 BJ> V.90 STANDARD IMPLICATIONS
 BJ> [...]
 BJ> But we're persuaded that most of the differences will remain
 BJ> proprietary. Basically, V.90 specifies how the modems will
 BJ> talk to each other to make complex decisions about how to
 BJ> treat the variety of digital networks across the land. And
 BJ> this is almost entirely client modem driven.
 BJ> [Image]   Broadly, once the two modems have established that
 BJ> they are V.90 modems and can do this trick, the
 BJ> client modem uploads a Digital Impairment Learning Descriptor
 BJ> (DILD). This basically tells the server modem what type of
 BJ> test tone to transmit over the network. The server modem
 BJ> dutifully transmits the tone, and the client modem compares
 BJ> the received tone to its internal reference tone. The client
 BJ> modem uses this comparison to calculate the best data point
 BJ> constellation to use and notifies the server precisely what
 BJ> constellation configuration to use in transmitting data
 BJ> downstream. The upstream link is still the 33.6 Kbps V.34 at
 BJ> best.
 BJ> [Image]   V.PCM describes how to upload the DILD, what to
 BJ> respond with, and how to communicate the
 BJ> constellation. The magic lies in the client modem calculating
 BJ> what constellation to use for any given set of digital network
 BJ> conditions. This is NOT specified at all in V.90 and will vary
 BJ> entirely between US Robotics, Rockwell Semiconductor Systems,
 BJ> or Lucent Technologies.
 BJ> The result is that we will have interoperable modems compliant
 BJ> with the V.90 specification and able to talk to each other.
 BJ> But it appears that large disparities in achievable connection
 BJ> speeds will depend on which client modem you are using, and
 BJ> which server modem you connect to. Our testing would indicate
 BJ> these disparities may be enormous - far beyond anything we've
 BJ> seen with previous modem standards. In the previous round of
 BJ> V.34, the Rockwell chipset was ubiquitous and modem
 BJ> performance was fairly level across the universe of available
 BJ> modems. V.34 was basically V.34. In the coming world of V.90,
 BJ> we would expect to see an almost implausible range of
 BJ> operating performance from modems all purporting to be V.90
 BJ> compatible. We'll have a standard, but it won't be very
 BJ> standard with regard to performance.
 BJ> [...]
 BJ> -------------------------------------------------------
 BJ> Editor: Jack Rickard - Volume XI: Issue 3 - ISSN:1054-2760 - March
 BJ> 1998 Copyright 1998 Jack Rickard - ALL RIGHTS RESERVED
 BJ> [Image]Fable Of Contents
 BJ> === Cut ===
 BJ> This part of the article's findings has NOT been disputed.
 BJ> Bob
Ah... yes, the full implications of this didn't really
strike home with me till you posted this select bit. Kinda
scary... makes me glad I use Couriers.
-Frank
(fsexton@ibm.net - http://www.concentric.net/~fsexton)
--- Blue Wave/OS2 v2.30
---------------