Rod, at 09:55 on Jan 29 1996, you wrote to Bill Grimsley...

BG> How do you explain the vast majority of calls which actually
BG> start out at the maximum link rate of either 28800 or 33600 then?

RS> I'm not convinced that the 'vast majority' actually do.

BG> I was talking about MY outbound connects,

RS> Dunno, I could have sworn you actually said the reverse
RS> in Aust_Modem but dont have any easy way to check that out.

Couldn't have been me, as 99% of my oubounds are to Dave, and fewer than
10% of those initially connect at anything less than full speed both ways.

BG> and believe me when I tell you that 90%
BG> of them connect at 28800/28800 straight up.

RS> Maybe. I am still dubious about claim at 33600 tho.

At V.34+, the last call I made to Dave was at 31200/33600, so my lines
aren't quite perfect either.  Once the new EPROM turns up, I'll have a
better idea in the long term, but the above was with Russell Brooks'
US/Canada Courier. 

RS> Corse its possible split hairs and say that the vast bulk
RS> of your calls are to Dave and may possibly happen like
RS> that, but thats not really what we were talking about.

I realise that, and I was simply refuting your claim about not believing me
when I tell you that most of my calls connect at 28800 straight up.

BG> To me, that certainly is the absolute VAST majority.

RS> Not terribly relevant to the general question of what happens with the
RS> vast bulk of outgoing calls made by USR VEs by everyone tho. THATS whats
RS> relevant to the question we we actually discussing there, whether it tends
RS> to take the conservative route of a lower initial connect speed which then
RS> much more often INCREASES during the session than decreases. I'm damned
RS> sure that SOMEONE said that in Aust_Modems and I could have sworn it was 
RS> you.

Possibly because I agreed with the majority of VE user's observations that
the link rate does actually increase, often within seconds of the initial
connect.

BG> And no, I'm not complaining.  :)

RS> Oh sure, thats clearly the desirable result.

Especially when it does so instantly and transparently.

BG> In fact, on those calls which have started low, I used to
BG> do an ATY11 line frequency response dump, and it was quite
BG> obvious that the high end was marginal for full speed connects.

RS> Yes, but thats a separate question to whether you would expect
RS> the line conditions to generally IMPROVE during the call. You
RS> would normally expect that it was just as like to move down as
RS> up with a session which is marginal for full speed connects.

Indeed, on one or two of my poorer interstate connects, that's exactly what
has happened.  Quite obvious from watching the cps rates on large files
too.  Pity the stats only detail the initial connect and the highest
asymmetrical rate attained.  It'd certainly be useful to know how many
times the modem had speed shifted, and in what direction. 

BG> It may well do, but logged stats don't show this to be the case.

RS> Welp, someone said they did, and I could have sworn it was you.

Likely it was Russell, or that overzealous Steve Anderson (out near Poe).

BG> One of V.34's greatest advantages is its ability to
BG> speed-shift instantly and transparently (not at all
BG> like the abortions that V.32bis retrains are), so there's
BG> no way of knowing exactly what is happening during a call.

RS> Well, there is with a decent modem which displays those two speeds
RS> on the front of the modem during the call. And thats PRECISELY the
RS> value in that approach, you can see whats going on during the call.

Sure, couldn't agree more, but I only know of two brands which do this, and
even then only on their top models.  Given that LED or LCD displays are
quite cheap these days, it's unforgivable that they haven't been more
readily adopted.  I'd gladly pay another $10 or so for that capability.

BG> Judicious use of the on-line ATY11 dump shows that line conditions
BG> do vary quite considerably, even in the space of just a few seconds.

RS> Sure, no one disputes that they dont, the whole POINT of V34 is because
RS> they do and to keep wringing the best thruput out of the line as it 
RS> changes. Completely separate issue to it having a considerable tendency to 
RS> INCREASE during the session when you would in fact expect the change to be 
RS> random.

So you're basically suggesting that V.34 should be slightly more aggressive
in its negotiation, are you?  My initial reaction would be to agree, but I
can't help wondering why the manufacturers chose to do it the other way. 
And it ain't just USR either.  Quite a few modems start low, and shift
upwards.

RS> Even sillier with a V32bis call, the modems negotiate a 14400
RS> connect fine, and then the damned USR virtually IMMEDIATELY demands
RS> a retrain, on every single call I have made so far. Nuts IMO.

BG> Have you called many (or any) other USR V.34s?  What happens with those?

RS> Both produced IDENTICAL results, with about 5 calls to each. Both with
RS> the same mindless demand for a retrain straight after the connect and both
RS> with the same roughly 30% rate of falling flat on its face after that.

Easy to say that the Courier shouldn't be demanding a retrain immediately
following the connect, but if that's its normal operation, I'd have fully
expected to be hearing other complaints about that.  Also, one of the more
objectionable and time-wasting duties I had to perform at CHH was to test
EVERY single modem they put up for sale (although I suppose DOAs _can_ be a
tad embarrassing), and this was done by calling Mike's own BBS - equipped
with a Courier - and DLing a largish file.  I must have tested literally
100s of modems in this manner, the bulk of them being 14k4, yet I didn't
once see the problem you describe.  And there must be a HELL of a lot of
people calling Couriers with their Rockwell 14k4s too.

RS> Which conveniently eliminates any possibility of
RS> it being Pauls line, or a dud USR at his place, etc.

Dunno, I think it might be wise to grab another 14k4 at your end if
possible, and see what happens then.  If it fails the same way, I'd also be
pointing the finger at Couriers in general.  All you've proved so far is
that YOUR Supra appears to have a minor incompatibility problem with 2 or 3
Couriers.

BG> Of course, it couldn't be your Supra's ROM at fault, could it?  :)

RS> Nope, not when its only seen when calling a USR VE, particularly when
RS> the line quality to get a V32bis 14400 session isnt exactly what you
RS> might call demanding. Its totally fucked behaviour by the USR VE.

Maybe, maybe not.  Complain again when a different brand of modem does the
same thing at your end, and I'll likely be a little more sympathetic.  :)

BG> I have no faith at all in the builders' ability to get it right in such 
BG> edifices.

RS> Its not the builders that do it with phone wiring.

No, it's usually the electrician (with an Austel licence).  And I still
wouldn't trust all of them to get it right 100% of the time.

Regards, Bill

--- Msgedsq/2 3.20