Re: Random Fast Eddie thoughts

[andre.horstmann@unifr.ch (Andre Horstmann)]

In article <59pr4n$3ls@cwis-20.wayne.edu>,
adrissmn@gopher.science.wayne.edu (Avi Drissman) wrote:

> So I decided to test the speeds of FE and GuS. I used a program called
> "CPU Speed" 1.2 by a Robert Claney. On my real GS, it's fairly accurate,
> and the speed figures vary in the hundredths place.
> 
> On FE, the speed reported was wildly varying, bouncing all over from about
> 1.6mhz to 3.5mhz. In GuS, the speed hovered in the 19mhz range,
> occaisionally dipping to 16mhz. It must be emphasized that these
> measurements _are not real_, because emulators really can only be tested
> under load, but they serve as an interesting comparison between emulators.

Fast Eddie is running very constantly at 3.5MHz on your Mac. The variation
you get from the benchmark program is due to the way these programs
measure the speed. Most of them count cycles between two interrupts. If
now Fast Eddie has to process a Mac event during this calculation, the
measurement will be incorrect. Normally you won't notice this (very short)
slow down, but since the counting time is very short the final calculation
is strongly affected.

No doubt, Gus seems to be faster than Eddie and so it's very likely that
the counting loop is not interrupted by a Mac event.

JFYI,
Andre

-- 
mailto:andre.horstmann@unifr.ch - http://www.swix.ch/clan/shadow/ (11.4.96)

The worst about computers: they never admit being responsible for an error.

              *** Fast Eddie Laboratories, Switzerland ***