Such things have been around for years. I remember Compute! magazine
had one. I tried it on the various computers I had, and the Aquarius
beat the heck out of all my 8-bits, and beat the heck out of all the
8-bits that Compute! magazine listed.
It was a calculation-based benchmark, and as you probably know, the
Aquarius BASIC uses fewer bytes for a floating point number and
therefore has less accuracy than most BASICs. (The TI 99/4A and the
Tomy Tutor are very accurate, as good as your calculator, I bet.
They also store floating point in a very strange "radix 100" format,
which is probably better for printing than it is for calculating).
I have also seen "print speed" based benchmarks. The CoCo wins that
one. It's hardly fair, though. The CoCo screen is only 512 bytes,
while, say, the Commodore 64's is 2000 (1000 for video matrix and
1000 for color nybs).
BYTE magazine reported "dhrystones" when discussing speed. Give a
good search for that (note the unusual spelling), on google.com.
Me? When testing the timing of my emulators (and they have to match
fairly well, for the speaker clicker frequency to sound right), I run
a few tests like these (this is data for the NTSC VZ200):
--- In tomytutor@y..., "David S. Brain" <dsbrain@n...> wrote:
> Here's a project for someone (emucompboy?), come up with some good
> benchmarks for testing the relative speeds of BASIC on various
> computers. Aren't you all curious to see how various systems stack