The V.34 code still has trouble, but its less serious than I've seen before
in previous beta loads.
I believe, however, that I have isolated the actual cause.
The V.34 problem appears to be that Lucent finally enabled 64D constellation
sizes for speeds beyond 28800 (this NEVER used to be the case; you would get
an asymmetric 64D/16D connection, which is why 33.6 was basically impossible
99% of the time). This TOTALLY screws older Rockwell chipsets (the RCI288
series) which can't handle this, even if they claim to have "working" 33.6
firmware. I've been able to confirm that symmetric 64D connections are now
attempted (using an older Courier which will tell me what is being
negotiated, and that knows how to do it as well).
However, older RCI288 Rockwell chipsets can't actually run this
constellation configuration properly.
The consequence is that in the absence of V.90 or K56Flex the PM3 will try
to negotiate a quality 33.6 or 31200 link. And, if it is talking to one of
the defective Rockwell chipsets, you'll get a link - that runs at about 100
bps.
This is still kinda serious. What we're telling people is to lock affected
modems to speeds of no more than 28800. This seems to fix the throughput
problems, but the Lucent code SHOULD detect this on its own and not enable
unsupportable constellation configurations with modems that can't properly
handle it. It would seem rather trivial to figure out that this is going
on during a session and step down a notch or two.
We also have one modem, a Motorola Modemsurfer, which can't connect at all.
Finally, there ARE some situations under which V.90 is still unstable, but
ALL of them to date have involved Courier modems. The symptom on this
is almost constant renegotiations of the link speed, and, usually, a
connection failure 3-5 minutes into the call.
I have to believe that is a problem with the Courier firmware rather than the
Lucent box, because Sportster owners are happy. Now that USR has gone to this
insane "automated updater" you can't really do much with this - you used to
be able to flash back and forth, but that now appears to be impossible.
In my testing I got fucked twice by this - two LEGITIMATE Courier
V.everythings with X2 enabled got *downgraded* to 33.6 mode, and then the
goddamn thing wanted a credit card number (and $60 from it) to turn X2 and
V.90 back on! Further, whatever this frigging updater did, its permanent,
because my old flasher to turn X2 on no longer turns X2 on - the "I'm X2 ok"
bit is trashed in the flash, and so I now have a couple of formerly-X2-enabled
modems that are V.34+ only. I called USR's supposed "help" line and all I
got was the run-around - they claimed that the modems in question never were
X2 enabled, and so I couldn't have possibly seen what I really saw. Yeah,
right.
I gave up. I've always hated US Robotics. This latest stunt was the last
straw for me; I will never buy another one of their products.
---- Karl Denninger (karl@MCS.Net)| MCSNet - Serving Chicagoland and Wisconsin http://www.mcs.net/ | T1's from $600 monthly / All Lines K56Flex/DOV | NEW! Corporate ISDN Prices dropped by up to 50%! Voice: [+1 312 803-MCS1 x219]| EXCLUSIVE NEW FEATURE ON ALL PERSONAL ACCOUNTS Fax: [+1 312 803-4929] | *SPAMBLOCK* Technology now included at no cost
On Sat, Jun 27, 1998 at 11:09:43AM -0400, CYGNET Support wrote: > I know a lot of the list users respect Karl D. so I am wondering if he > tried b17. > > Karl if you tried it I would love to see your input. I liked the input you > gave on b15. > > How about it? > > > Thanks in advance > > > E. Grant > > > - > To unsubscribe, email 'majordomo@livingston.com' with > 'unsubscribe portmaster-users' in the body of the message. > Searchable list archive: <URL:http://www.livingston.com/Tech/archive/> - To unsubscribe, email 'majordomo@livingston.com' with 'unsubscribe portmaster-users' in the body of the message. Searchable list archive: <URL:http://www.livingston.com/Tech/archive/>