Re: Performance degradation over time on VAX...

[Arno Griffioen <arno.griffioen%ieee.org@localhost>]

On Tue, Jun 30, 2026 at 07:27:25AM -0500, Daniel Seagraves wrote:
> My machine is a MicroVAX 3100 model 20 with spinning rust SCSI disks,
> no emulated anything, and it started on the internal coax ethernet but
> switched to twisted pair on an AUI transceiver later because I suspected
> the coax might be influencing things. (It didn’t seem to make a
> difference but I need to run tests again to eliminate it as a variable)

On a MV3100 with the original Lance (AMD Am7990) it should not make any
difference in using either thin coax or an AUI to 10baseT adapter.

In both cases it will remain half-duplex in operation as the LANCE chip
itself does not support/handle full duplex operation.

If it did support FDX then with a suitable 10baseT AUI adapter you
could have run that and taken the randomising effect of collisions and
retransmissions mostly out of the equation.

As long as your coax segment is quiet or even used as a kind of
point-to-point link with only 2 machines then there should be no
difference in performance between built-in coax or AUI.

AUI adapter is just the same signal conditioning logic housed in an
external enclosure that the thin coax on-board circuits do as well.

If the network it connects to is busy with other machines/traffic
then an AUI to 10baseT linked to an ethernet switch would at least
reduce the 'chatter' seen by the microvax interface (only broadcast and
flooded/unknown-unicast from others will be seen) and thus less chance
of collisions on the switch<>microvax link.

But.. The above assumes things are in good working order.. 

If the internal transcever logic for the coax connection is iffy then
it may just perform sub-par because of lots of retransmissions.

Something I have on an ethernet VME card in an Atari TT where the coax
connection results in terrible performance because some of the transceiver
components on there are, lets say, 'not well'.. So I do run that with
an AUI adapter.

But that would be fairly easy to check with some data transfers across
either connection.

							Bye, Arno.