🚀 go-pugleaf

On 29 Mar 1996, Ted Oliver wrote:
> But seriously. Flaws in you analysis include, but are hardly limited to:
> 
> 1) using a single measurement

Well, the tests were just for floating point performace.  But I 
agree, there should be other tests to adequately measure all areas
of system performace.  However, the fp results do speak for themselves.


> 2) not indicating other system information (RAM, HD, whether the Power100
> had L2 cache, etc.)

Yes, the presence of a cache should have an effect.  How much, I don't
know.


> 3) comparing linux with a GUI OS, which inherently incorporates more
> computing overhead. A better comparison would be under Windows. Even
> better, comparing both systems on linux later this spring/early summer
> when it is available for Nubus Mac's.

The fp tests should in no way test the GUI aspects of MacOS.  The
fp test program was just that:  nothing GUI about it.  It's
a pure test of the CPU performance for floating point computation,
and the GUI overhead should be minimal, if nonexistent.

Unfortunately, there has already been a test between the PowerPC 604
for the MacOS and WinNT.  Macworld, in a previous issue, did a 
comparison between WinNT 3.51 on a Motorola PowerStack 604
@100MHz vs. PowerMac 9500/132.  The PowerStack posted better low
level system benchmarks than the PowerMac.  Macworld then went on
to try to explain the discrepancy.  I remember that I/O on the
PMac 9500 dragged down performance.


> 4) more recent versions of MacOS (e.g., 7.5.1) have a much improved
> mathlib over System 7.5. Maybe the original poster has 7.5.1, but it isn't
> indicated as in the above text.

The tester indicated that the Mac side of the fp test program was 
compiled in CodeWarrior 8.  CodeWarrior 8 supposedly has a vastly
improved (speedwise) math library, so I'm not sure this would explain
the difference in speed.  The original tester may also want to run
the tests using Motorola's compiler, which is also highly optimized.


> 5) MacOS is not fully native, linux is native on x86. The MacOS is not the
> best OS to show how fast a PPC CPU is. Try comparing your linux results
> with Win3.1 results on the same machine.

Again, the fp tests should not depend on the parts of the MacOS which
are still emulated.  The math library is PowerPC native, and I feel
that the tests fairly measured what the tester wanted to benchmark.


Regards,

Walter Hsueh
hsuehw@ecf.utorontot.ca