Rod, at 09:07 on Mar 29 1996, you wrote to Bill Grimsley...

BG> While Paul had the Courier, I was seeing 28800s (with occasional
BG> 26400s, but don't forget that this was with the old Sportster ROM
BG> which couldn't even connect at all with the M34F most of the time),

RS> These are irrelevant, we at talking about how USR/USR performs
RS> compared with how M34F/M34F performs on calls from you to Paul.

BG> That's fair enough too, which is why I've basically ignored
BG> any data collected before using the current V.34+ EPROM.

RS> Yeah, its quite possible that even the USR/USRs would improve now too.

They certainly have, as it happens.  Poe's and Dave Hatch's lines are less
than immaculate, and reverting to the previous EPROM costs me between
4800-7200 bps on calls to either of them.  I'd like to continue testing,
but I'm a bit worried that I'll end up destroying the modem's PLCC socket. 
No matter though, the difference is not just noticeable, it's actually
quite horrific.

RS> I have gone back to your message after that and it doesnt
RS> say that. Most of what you said was about the %L value
RS> reports which we now know are completely meaningless.

BG> It's scrolled off the end of my limited message
BG> base here, but I did import the connect strings
BG> for those M34F-M34F calls, both of which were 26400.

RS> Nope, it aint in the message. The %L value stuff is.

Damn, I could have sworn that I imported the entire connect string.  I
certainly meant to, because of the 26400 indicated for both calls.

RS> Its now clear you did make too few now we know it was just two.

BG> you could also say that 100% of my M34F-M34F calls were 26400 while
BG> 99.9% of my USR-M34F calls are 28800.  But I won't, 'cause I also
BG> agree that's far too small a sample to say anything really useful.

RS> Yeah, the stats are useless to substantiate the claim right at the top.

What claim might that be then?  Nothing at the top of this message.  :)

RS> Which is all I ever said. It might even be true, but
RS> we dont actually have the evidence that allows that to be stated.

Yeah, I'm more than happy to agree with that.  Incidentally, there's one
thing I haven't yet tried, and it might be relevant to the USR's
inconsistent line probe stats with Rockwells, and that is checking to see
if the figures are the same when the calling modem is either a Sportster or
a Courier.  Until my Courier arrives, I'm not in a position to check this,
but Dave sure is with Jill's V.34 Sportster.  Must ask him to check that
out.

Regards, Bill

--- Msgedsq/2 3.20