[Info-vax] DE500 and autonegotiation

[B. Z. Lederman]

In article <kc249k$p7u$1 at usenet01.boi.hp.com>, "Robert A. Brooks" <rab at hp.com> writes:

My recollections are essentially the same as Hoffs.

When I worked in OpenVMS engineering, I had a problem with an Alpha
and one of the earlier PCI interfaces.  I consulted with >>> THE <<<
expert, and was told that, after it was implemented, some deficiencies
were found in the 100BaseT autonegotiation protocols.  Usually it worked,
but sometimes it didn't depending on which NIC you had and which
Router / Hub / Switch it was connecting to.  The solution most of the
time was to either change the network equipment (if possible), or,
on a VMS system, use the options that enable you to specify the
settings the NIC should use and not autonegotiate.  This is the
quickest solution as the external equipment shouldn't change
speed or duplex settings very often.

I was also told that the deficiencies were fixed in the 1000BaseT
protocols, so another solution is to change your external Router /
Hub / Switch to 1000BaseT.  My personal experience with some older
Apple Macintosh systems that didn't autonegotiate correctly at
100BaseT is that they do work correctly on a 1000BaseT hub.  And since 
nearly all of the equipment you buy today is 1000BaseT or "better", 
that's also an easy solution, even if your NIC and the rest of the
network are running at 100BaseT.

Bart.


>
>>> There is never any reason to have VMS boxes autonegotiate, as far as
>>> I've found out.
>
>I do not plan on starting a religious war, other than to pass along (one 
>again) the words of the guy who wrote and maintained the VMS ethernet 
>drivers for over 20 years (paraphrased).
>
>-----
>Autonegotiate should always work, assuming a properly-functioning switch 
>and card.  The VMS ethernet drivers are designed for autonegotiate to 
>work, and if there is a problem, we'll fix it.
>
>-------
>
>Of course, the crew maintaining the ethernet drivers has changed since 
>this opinion was issued, but the point is that the guy who wrote the 
>ethernet drivers disagrees with whoever made the assertion I've quoted.
>
>(The HP usenet feed went a bit nuts over the Christmas break, so I did 
>not see the original posting in this thread).
>
>                                          -- Rob
>