gauging interest in VAX 6000-530

From: allisonp_at_world.std.com <(allisonp_at_world.std.com)>
Date: Mon Oct 25 08:28:06 1999

> > Until you fill memory. then the contest reverses. The 486s of the time
> > were doing good at maybe 5meg bytes/sec and thats SLOW. The real test
> > then and now is what the system does with relational database of say
> > 100mb is size.
>
> It's a different test. I didn't say anything about relational database
> performance.

My users don't run drystones, they run databases and autocadd.

> I was trying to get a handle on -real numbers- for the 8650, based on
> numbers that are known. There are no performance numbers published for the
> 6500 that I am aware of.

Well don't use VUPs as interchangeable with MIPs, there is at best a weak
relationship. VUP is vax unit of performance and it's more a real this
is what it does with 50 users measure. You will find that some models of
vax are far hotter than the VUPS would indicate for MIPS or Drystones
compared to similar model with more vups. Why? optimized for math rather
than IO throughput or other parameter.

> Nor am I, I guess (is there an "average" retro geek?). I worked at DEC
> '73-'75 in Marlboro. (We built 10s and 20s there). I had Ken Olsen over
> for dinner and asked him about what it would take to get the PDP-11 out
> there competing with these newfangled microprocessors (y'all know what Ken
> thought about micros...). And I still have a copy of a memo I sent to Gordon
> Bell, suggesting that we put the PDP-10 on a board and sell -that- into the
> home market (and Gordon said: do it! and it was done, but several years
> later...). This was in 1975, mind you...

Ah, a ringer.

> Is this VS3100 actually an Alpha? If so, I have always realized the
> superiority of the Alpha. No contest.

Your out of the loop. VS3100s are what some call baby VAXen they used
varios version of the VAX chip starting with Cvax and wer mid 2s for VUP
rating though there are models with far more impressive numbers. The
VS3100m76 I have (pizza box) is rated 7.8vup and thats a 1992ish design
and with the dual SCSI bus and 16mb of ram it was the hot desktop (and
more).

> CI (at 32 MB/sec) is 256 Megabits/sec. Therefore, you can hang 16 CI bus
> adaptors onto a 66 MHz / 64-bit PCI bus.

Except that bytime we got to PCI CI was already over 10 years old and
there are fasters thing like fiber channel.

> Wait! I'm -not- saying that the PC world when the VAX 8650 was introduced
> was equivalent.

Right, the vax was ahead. that race may have ended but it's really a
quite recent event.

> ALL I'M TRYING TO DO is to put some perspective on CPU performance of this
> old obsolete VAX iron.

If your saying a 486 can beat a 780 the answer is, depends. for a lot of
stuff it is faster and for others it's a real dog.
 
> I'm -merely- trying to understand the single-user performance of that old
> iron in light of new iron (that I can actually benchmark).

Until you take into account the OS and the services running along whith
user tasks there is not compare save for those bare iron numbers that are
meaningless. As a single user on my M76 I have all the graphics (higer
res too) as the more common PCs and for many things its faster and some
slower. Then again the RZ2x drive I'm using arent that fast either and
I'm only running 16mb of ram... Put those constraints on you latest
PIII/4xx and see where it goes between reboots.

> Again, I dunno what kind of PCs you people were using, but I have not had
> PC reliability problems. Maybe I'm unusual in that respect.

I do better than average, I have uptimes that people tell me are unusual
for PCs with w95 and winNT but... I have a MicrovaxII I bought on exit
running VMS5.4-4 that was 4+ years old then thats still running and still
averages better uptime than any PC (power failures are the limits there!)

Sure it's slow, it's 1987 hardware but I don't worry if it's going to
crash before it's done or want another 16mb of ram to keep the OS happy.
I'm not saying PCs cant do this only that they generally dont achieve
this.

Then again not running anything from redmond helps. But that does not
correct inherent limitations.

> But for those old systems? Sooner or later, the maintenance costs will become
> too high, and the Itaniums will replace 'em, eh?

Life is like that and we die too.

Allison
Received on Mon Oct 25 1999 - 08:28:06 BST