"Andrew Tyson" <asptysonATyahooD0TcomD0Tau> writes:

> I am contemplating purchasing a second hand Sun blade 1000 for personal use.
> I already own a number of older SBus based UltraSparc machines and they have
> always been reliable. However at work we have had three U10s die (with a
> known
> CPU fault)- so I am a bit wary of the newer stuff (I know that U10s are
> fairly ancient
> but it's all relative ;-)
>
> I like the idea of mixing CPU speeds on the SB1000, and also the FC-AL
> drives, memory expansion capability ...etc. I guess I am putting out the
> feelers to see if there are any potential problems with these
> machines.

I am a bit late to this thread. I loved my Blade 1000 (at work) when I
got it and for years afterwards. Built well, and it was the fastest
sparc I've ever had.

Now however I also have a Sun Blade 2500 (dual 1.28GHz cpus) also at
work and for my work the US-IIIi is definitely the CPU to have, so I'm
more ambivalent about the SB1000. The downside of 1000/2000 is they
were never good value, FC-AL disks - why? and actually the cpu
performance on normal (integer) workloads is not that great. I have
noticed a steep slide in the prices for SB1000 on Ebay recently, so
I've been assuming that others have noticed how much better the sparc
bang/buck is from the newer US-IIIi based machines. I think it's
because the latter has integrated on-chip memory controls, and both
1st and 2nd-level cache. The load-to-use memory latency is much lower
and this allows a much cheaper implementation to come out roughly
equal on the benchmarks with the 8MB cache US-III.

Memory expansion capability on the 1500/2500 is good, and I believe
the memory is also a faster kind than on 1000/2000.

The only place where 1000/2000 wins is if you really want to use UPA
graphics cards. They are the last of that breed (unfortunately).

Cheers

Chris
--
Chris Morgan
"Post posting of policy changes by the boss will result in
real rule revisions that are irreversible"

- anonymous correspondent