[SunRescue] Re: Woah, a PCI Ultra 1

[Jonathan [no, I don't write for /.] Katz]

> Bill Bradford wrote:
> > I didnt beleive it till I saw pictures:
[ deletia ] 
> > http://cgi.ebay.com/aw-cgi/eBayISAPI.dll?ViewItem&item=275499088

I e-mailed with the guy cause I didn't beleive him until I saw the pictures.

> How many other 'alpha' test machines get outside Sun? How many machines
> get as far as prototypes, then get canned? Has anyone got one of the
> infamous Sun 486 machines??

When I was the Unix admin at Alteon Networks for a few months we had all 
kinds of funky Sun gear. Alteon is a partner with Sun in making their 
Gigabit Ethernet NICs and switches, so we needed all kinds of platforms 
to develop on and test their stuff. I remember in particular an 
UltraSparc2-based system which was nicknamed 'Quark' (this was in late 
1997.) In looking at Sun's current product line I think it would either 
be a true precursor to the U10 or U60. It had SCSI, but only room for one 
processor.

Any site that does any kind of OEM or partner work for Sun is a good 
place to start hunting for Alpha/Beta-type equipment. Most get that 
equipment free from Sun, devlop on it, and then "forget" to return it 
when the real products hit the market.... then the next cycle of 
development occurs, and that old box goes into some closet to collect dust.

> > I wonder why it was never released.  I would have bought one!  The Ultra1
> > is SO much better made than the '5.
> I was trying to find performance specs between U1s and U5s - anyone got
> comparison specs?

Well, the true "die-hard" workstation junkies (me included) only get 
warm-fuzzy feelings when the pizzabox on their desktop has SCSI in it. A 
lot of us were scared off by the U10 and U5 machines. 

In being the contractor/whore I was in Silly Valley I eventually wound up
at a site where there were a lot of hardware engineers who worked on both
the SS2, SS10, U5, and U10 boards. I flat out asked one "why the hell did
you go with IDE?" I was told because it was cheaper. When asked about the
performance implications (how IDE polls the CPU each time a read or write
is performed) he laughed, explaining to me that the IDE controller had a
seperate ASIC which handeled that polling, offloading that burden from the
CPU-- yielding no performance difference. AFAIK, the graphics on the U10
and U5 blow the TGXs that the U1s shipped with (which weren't Creator
2/3D) out of the water. 

No doubt that the U1s were better-made, but the U5s and U10s will run 
circles around it speed-wise. Even the buzz from Sun when the U5/U10s 
were produced (codenamed 'Darwin') was that they were made cheaply 
overseas. I've seen several (3 out of the 100 or so I've worked with) U5s 
get sent back to Sun because the NIC on the motherboard failed.

Cheezy NIC issues are old-hat for Sun. I was told to be suspect of the first 
production hme chips as found on the U1/170E's, as they were 10/100 chips 
which were released before the 100Mbit standard was finalized. I never 
had any issues with them (they seemed to fall into 100Mbit/HDX mode too 
easily sometimes) but I was warned to "stay away!"

Take Care!

-Jon  
Jonathan Katz =  J. Random BOFH = http://www.in.net/~jkatz = jkatz at in.net
 "Disproving myths about Corvette owners one girl at a time since 1997."