Re: [gentoo-sparc] what ultrasparc compare likely such Pentium box on desktop performance

[Keith M Wesolowski]

On Thu, Dec 02, 2004 at 12:21:29AM +0800, Zhang Weiwu wrote:

> Okay, so I am wrong: is it true that U5 and U10 are worse than U2 when
> equiped with 300~400 CPU?

Usual disclaimers apply when discussing performance: YMMV, performance
is highly workload-dependent, it's possible to contruct pathological
workloads that prove almost any ridiculous contention, etc, etc.

The best CPUs available for U5/10 (440) will probably offer you
performance similar to some of the slower available for the U2 (200,
250).

The U5/10 are PC systems and they are full of bottlenecks.  The memory
subsystem is a joke.  Most of the CPUs, even the higher-clocked ones,
have inadequate cache.  The U5 has no UPA slot so graphics will always
be slow.  The PCI in these systems is slower than a modern PC's and
much slower than the U2's 64-bit SBUS.  The standard disk system in
the U5/10 is IDE - to make matters worse the chipset is a junk ALi one
with severe bugs that sometimes make DMA unusable while the U2 has a
fast/wide SCSI system which is certainly not very modern but
infinitely better.

A U5/10 is comparable in both performance and quality with an
eMachines celeron system of similar clock and memory capacity,
provided you can use IDE DMA on both.  If the U5/10 cannot use DMA, it
will be slower still.

A U2 is comparable with a higher-clocked P3 system of similar memory
capacity and number of CPUs.  There are addons available for either
type of system that could give it an advantage (for example, an U2
with FC disks would gain further substantial advantages over an
IDE-equipped PC, but an Ultra2 SCSI-equipped PC would have an
advantage over an U2 using the standard onboard SCSI subsystem).  All
U2s are SMP-capable, while most P3 systems are not.  Tasks which are
memory-bandwidth intensive will fare even better on the U2, which has
about 4 times the bandwidth of common P2/P3 PC boards (the U2 memory
system is 576 bits wide and runs at 100 MHz - similar in performance
to modern Athlon and Opteron systems).

> Another question I had for a long time: does SMP really makes sense on
> desktop, browsing and word processing? I don't expect dual CPU average
> performance 200% of single CPU, I don't even expect 120% performance, I
> thought not needed for multi-process for desktops?

If you never run an application that uses threads, or you never run
more than one application at once, SMP will give you only a very tiny
benefit (<5%).

Such workloads are fairly unusual; as a gentoo user you will at
minimum occasionally compile software, which almost always can be made
parallel by the use of make's -j option.  You might also consider
that, for example, running openoffice and gimp at the same time also
means X is running, and all 3 may well have work to do simultaneously.
How well you exploit your system's CPU(s) depends on the very
nitty-gritty details of the work you do.  In general, SMP definitely
does provide a noticeable performance boost in common situations.

It would be nice to lay this issue to rest.  The Ultra 5/10 and Blade
1x0 are junk.  They are similar in both design and performance to
low-end PCs available at the times they were released.  Other Ultra
systems are significantly better and, although their CPUs are not
clocked especially fast and their SCSI controllers are somewhat dated
now, most offer both SMP and design characteristics which are only now
becoming available in high-end PC workstations, such as switched, wide
system interfaces.  Finally, as always, understand your workload
before buying.

-- 
Keith M Wesolowski
"Site launched. Many things not yet working." --Hector Urtubia

--
gentoo-sparc@g.o mailing list