Re: Canadian OpenVMS Seminar (07.02.20)


"Robert Deininger" <rdeininger@xxxxxxxxxxxxxxxxx> wrote in message
news:rdeininger-2402071048310001@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
l3.net...
<snip>
All marketing fluff should be taken with a grain of salt, and
fancy-sounding terms should be defined and understood before any
real-world value is assumed.

Agreed (as with the rest of your stuff).

Lots of interesting reading in your post, which unfortunately I may not be
able to get around to for a while.

In the meantime, are readers really supposed to conclude that HP's best
marketing (and maybe other) folks consider that "cosmic rays" is the
appropriate answer to the question of "What specific RAS-related features
are not available in currently available AMD64 servers, when compared with
features which are available in (comparably-priced?) currently available
Itanium/Integrity servers?".


Re: Solaris x86: Neil wrote about "running Solaris or LINUX on x86 to see
"if the hardware is flaky" or "is Windoz just junk". (I put my money on
Windoz being flaky which means HP should still start a Skunk Works to port
OpenVMS to x86-64)."

For this discussion, let's ignore Linux and stick to Solaris, which has a
Fault Management Architecture, and one of the folks working on it is Gavin
Maltby who has a blog at http://blogs.sun.com/gavinm/

It's no secret that AMD introduced Opteron Revision F last year, with
support for DDR2 memory and support for online spare memory chips, amongst
other stuff. Gavin's blog entry at
http://blogs.sun.com/gavinm/entry/fma_support_for_amd_opteron talks in
reasonable detail (far more than we've seen from HP so far) about other
RAS-related differences but his summary seems to be that "... there are no
new error types detected in these banks (icache, dcache, load-store unit,
bus unit aka l2cache) so no additional error reports for us to raise or any
change required to the diagnosis rules that consume those error reports."

C'mon HP, plenty of folks understand soft memory errors and the like at some
level of detail (which is presumably what the "cosmic rays" reference
related to). Plenty of people also know that if Proliant-class systems (not
just from HP, but from Dell, IBM, Sun, etc) couldn't already cope adequately
(for some value of "adequate") with that kind of thing, they wouldn't sell
in the
kind of volumes we see.

So, one more time: what specific RAS features are in Itanium systems that
aren't in comparable AMD64 systems (or the Intel equivalent, for anyone in
VMSland who for some reason might prefer to retain the exclusive commercial
relationship with Intel)?


The UK HP User Group has an Intel-sponsored seminar on the theme of
Integrity, scheduled for April 24. Maybe someone in HP or Intel could
arrange for a plausible answer to be available by then, but earlier would be
better.


regards
John


.

Relevant Pages