| Thursday,
October 11, 2001
We have done preliminary performance testing
on the new Quicksilver dual G4/800 and single G4/867 running
Mac OS 10.1, and thought we share the results with you. Like
the 867 MHz Quicksilver, the 800 MHz machine has 256K of on
chip L2 cache running at full processor speed, and 2 MB of
L3 cache running at 1/4 the processor speed. In the case of
the 800 MHz machine everything is doubled. (L3 cache sits
between the processor and main memory.)
OS X is supposed to be built from the ground
up to take advantage of dual processors. Once you are running
a Carbon or Cocoa
application, we are told, you should see a significant performance
improvement. From our testing that has indeed been the case
with parts of some programs. The programs we used for testing
are heavily weighted towards applications developed by Apple.
We felt that these programs would be the most up-to-date applications
in terms of taking advantage of the G4 processor and also
dual processor configurations.
As you'll see from the results below, performance
of the dual processor machine is really mixed. Converting
a QuickTime movie to a DV Stream showed an 80% speed improvement
for the dual processor Power Mac. However when encoding the
same file using a Sorenson codec, the dual processor machine
was slightly slower than the single processor 867 MHz. So
we are puzzled why we are getting good results in one instance
but poor results in the other.
Many of the other test, also showed no performance
improvement.
However you can see the potential in a dual
processor setup by viewing the AltiVec fractal results. If
you adjust for processor speed, the dual processor machine
was almost twice as fast. This test completely saturates the
processors with data, and you can see how this could be an
advantage to people that do the type of work on their computers
that calls for huge amounts of continuous data crunching.
We have contacted Apple for input on helping
to explain why we are getting these varying results. Of particular
interest is why we are seeing no improvement in the iTunes,
QuickTime (Sorenson codec) & iMovie Effects, when running
under a dual processor environment.
We would also welcome reader input about these
curious results and suggestions for other applications we
could use to stress dual processor setups, Mac OS X, and the
G4.
Take a look at the results below. We have included
our best guess about what is going-on to generate these differences.
If you have a better explanation, or a different opinion,
please let us know. Either e-mail
us, or use this link
to post your ideas on our bulletin board.
All machines were configured similarly and running
OS 10.1
Note: This is our initial look at the performance
of the new Quicksilver Towers. We will be running more tests
and hope to have a full review of all the new Power Macs in
the next few weeks.
Additional resources:
"Real World" Tests
The tests below are from our suite of real world application
tests. These tests feature a diverse selection of applications
commonly used by the Mac community. The test suite was designed
to render an accurate and well rounded picture of a machine's
performance. All of the tests below were timed with a stopwatch.
The times are then converted to percentages relative to the
Quicksilver 867 MHz machine which is set to 100%. For all
scores, higher numbers are better.
QuickTime Tests
Although a Sorenson encode responds
well the Velocity Engine function of the G4, it does not
appear able to take advantage of dual processors
This is the kind of performance we would expect from a dual
processor setup. There is some overhead, and so in most cases
you do not see a doubling of performance
The drive performance is probably more of a factor here but
not totally. It seems that having an extra processor is of
no advantage when carrying out this type of function
Only modest improvement in performance for the dual processor
machine. We noticed that neither processor was fully utilized
when carrying out this operation. Indeed, only about 1/4 of
the processing power of each processor was being used. It
is hard to believe that this was because of the inability
of the hard drive to keep the processor fed with data. The
file used for this test was 320 MB in size.
Dual processors show no advantage. Bottleneck
may be the SuperDrive performance
Dual processors have no effect, speed of processor
does
The Search & Replace function also does
not seem to be able to utilize dual processors
Ok, this is what is suppose to happen with
a Carbon or Cocoa application when running under OS X
This
test is completely processor intensive and both processors
in the dual machine were continuously flooded with data
This
test opens and closes 1,000 windows as fast as possible.
No benefit to having a dual processor here
These results mirror the performance differences
we observed between the two machines when running under OS
9.2
This again was about the same under 9.2
In this test several applications are opened
and closed repeatedly using AppleScript. The better performance
of the dual processor machine may be due to better drive performance
rather than the fact that the machine has two processors
|
