[H-GEN] Compaq Proliant

[David Jericho]

[ Humbug *General* list - semi-serious discussions about Humbug and     ]
[ Unix-related topics. Posts from non-subscribed addresses will vanish. ]

On Sat, Jul 26, 2003 at 01:27:01AM -0400, Robert Brockway wrote:
> [1] Actually I have a feeling Tony has mentioned before that he's attended
> comnputer auctions, but others may find this interesting.

Always a good idea to go with someone in the know your first time.

> > The 4 Gb one takes about 15 seconds to spin up, according to the bios.
> 
> Ouch.  I wonder if they are getting a bit old.

Well, depends on what bus they're sitting on. I've got 4GB disks
running U2W (80mbytes/s) in a 150MHz machine, and 4GB disks running on 
SCSI 2 (20mbytes/s) in a 175MHz machine. The U2W drives without a doubt 
are faster.

Spin up time is hardly an indication anyway. Most SCSI controllers
will spin a single disk up at a time to prevent overloading any
backplanes or power supplies that reside in the system. 

> It is possible to speak in general terms though.  A 2x200Mhz box isn't
> going to outpace a more modern 500Mhz box.  The newer box will likely have
> a faster IO subsystem, etc.  In any case, even if the job being done is

Very true. You not only don't have bus contention from two processors
trying to access the memory and pci bus at the same time, but given
the development cycles of the Pentium lines, a slower bus. 

>From memory, there was a 500MHz P3 with a 133MHz FSB, whereas the Pentium 
200s had only 66MHz. FSB makes a HUGE amount of difference in usability 
of a box.

I know I'll open a can of worms saying this, but when you compare a P4
with a 533MHz FSB against a Athlon with a 266MHz FSB (at the same
clock rate), the P4 walks all over the Athlon in any task that won't
fit in cache. RISC vendors have been screaming about higher
throughput FSBs for years, and it's only recently that Intel (and now
AMD) have started listening.[1]

> faster single cpu system.  This is because most of the times that you want
> grunt it's to run a single large app.

Mozilla is a great example.

> Bottom line is, unless the box is a server which expects to be running
> multiple cpu intensive apps often you're better off getting one fast cpu
> instead of 2 cpus half the speed.

I'd refine that further. You'd be better off buying two X speed
processors for multiple intensive tasks, than one X speed processor.

I'll often chose a 800MHz processor over a 400MHz processor anyday,
regardless of task at hand.

[1] The RISC vs CISC processor[2] design path has been interesting to
watch. It's only now MIPS and SPARC are starting to cotton onto the
fact that people do infact want faster clockrates, and throughput
isn't everything.

[2] I know the lines of RISC vs CISC are quite blurred, but how
do lump MIPS/SPARC/PA-RISC[3] into a different box from Intel and
AMD?

-- 
David Jericho
Systems Administrator, Plugged In Software

--
* This is list (humbug) general handled by majordomo at lists.humbug.org.au .
* Postings to this list are only accepted from subscribed addresses of
* lists 'general' or 'general-post'.  See http://www.humbug.org.au/