[Info-vax] web browsers for VMS on ALPHA

[John Wallace]

On Feb 22, 7:30 pm, hel... at astro.multiCLOTHESvax.de (Phillip Helbig---
undress to reply) wrote:
> In article
> <12cb4281-3554-4a65-b4a5-419145a3e... at d12g2000vbz.googlegroups.com>,
>
> John Wallace <johnwalla... at yahoo.co.uk> writes:
> > Usual responses to "it is too slow" apply - what resource limit is
> > causing the perceived slow performance? CPU?
>
> No---EV67.
>
> > Memory?
>
> No---2 GB.
>
> > Disk IO?
>
> I don't think so.  Bottleneck is the BA356, which limits speed to
> 20 MB/s (disks and controllers are faster).
>
> > Network
> > IO?
>
> The network card is 100 Mb/s and the DSL connection is 16 Mb/s.
>
> > And proceed from there. It would seem fairly unlikely that an EV6
> > doing simple web browsing would be compute bound, but confirming where
> > the bottleneck lies is what MONITOR is for.
>
> I haven't rebooted since bringing this machine into the cluster; some
> things are probably not optimal.  Will investigate more after the reboot
> (did an AUTOGEN after a few days of uptime).

More information (e.g. observations from MONITOR SYSTEM) and more
analysis needed please, at your leisure. At risk of preaching to the
choir, but...

Just because you have 2GB of memory does not mean you are making
effective use of it. You could (at least in principle) end up seeing
poor performance if the UAF/account/system parameters were set such
that not much of the 2GB was actually available to the process in
question; the OS could waste loads of time processing page faults
which hopefully would be soft faults (no IO involved) but there's
still overhead in processing them when it may not be necessary.
(Philip H also suggested this, amongst other things).

Best of luck.