BG> I had originally suspected that the last V.34 chip may have been
BG> sub-standard, but what I've actually found is that all three of the
BG> earlier chips returned much the same results - all very ordinary -
BG> whilst the new V.34+ EPROM has provided an absolutely quantum leap
BG> in performance.  It's like running a completely different modem,
BG> as there's just no comparison between them.

RS> Its basically just the usual excising of modem inter operability warts
RS> on the handshake failures if they really have gone for good with the V43+
RS> roms. Presumably seen rather more on your type of very ordinary line.

BG> May also have been an inherent design fault,

Well, thats included in what I meant.

BG> considering that only the floating connects were
BG> occasionally failing. Once a specific link rate
BG> was forced, say 26400 or so, the problem went away.

Well, thats a bit harder to say given the variability of the fault occurrence.

BG> It could be that USR was trying too hard for full speed
BG> connects, ignoring the problems that showed up with faster
BG> link rates on poor or marginal lines.  Hard to say for sure.

I cant see that given the USR tendency to connect at a lower
rate and fall forward quite quickly after that. And it shouldnt
produce a symptom like that either, handshaking failure.

RS> The same thing happened with one rom change in the Supra
RS> V32bis roms too, helped the handshaking very substantially.

BG> Yeah, they were one of the few manufacturers to supply ROM upgrades, but
BG> most Yank companies were pretty bloody slack with their V.32bis modems.

True.

BG> Interestingly, I referred back to some of my press releases from
BG> USR, and they actually stated at the time that the V.34+ upgrade
BG> EPROM had been improved so dramatically, that users could expect
BG> at least one 2400bps step improvement in link rates, and in some
BG> cases, as much as 2 x 2400bps steps.  This actually correlates
BG> quite accurately with the results I'm seeing here too.

RS> Sure, but thats a completely separate issue to handshaking failures.

BG> Dunno, if the signal level or the modulation was somehow
BG> stronger, I think the two would go hand in hand.

Nope, the press release just talks about getting a higher thruput
out of a particular line. Nothing to do with handshake failure.

BG> But no matter whom I call at V.34 now, the modem categorically
BG> refuses to fail.  I'm certainhly not complaining though.

Thats how it should have been in the first place and adds to the evidence
that it just had a wart when calling non USR modems, visible with particular
types of not very good sessions, manifesting in handshake failure. Clearly
even the V32bis Sportster ALSO had it, tho not producing the same failure rate.

RS> AND its hard to say how much of that is poetic license on the
RS> speed stuff too, it may well be what actually happened is that
RS> its just fixed some blemishes in what they did before that. USR
RS> has always been very reluctant to admit they have fixed a stuffup.

BG> This improved Sportster V.34 code was put about at
BG> around the same time as they released the Courier
BG> flash-ROM which was giving everybody much faster connects,
BG> so I presume it's much the same revision of code.

Yeah, I assume so too.

Just think, if USR had had enough sense to make the Sportster
and Courier code compatible and had included a flashrom in the
Sportster, you could have had the improvement THEN as well |-)
@EOT:

---