Some initial benchmark results

[Maury Markowitz]

I'm keeping a running list here:

 

https://github.com/maurymarkowitz/Atari-BASIC-benchmarks/blob/main/RESULTS.md

 

Just got off the phone with Jim Galbreath, author of the original 1981 Byte Sieve. I was trying to track down the version of BASIC used in the original listing, as it doesn't match anything I've seen (and its missing some line numbers). Its at the bottom the page here. He said his son wrote that version, but he's in the car driving to California so I don't expect an answer until at least Sunday.

 

Some interesting points:

 

1) can anyone explain the strange result for MS BASIC running broucke? It's significantly slower than any other variety, even Atari, but nothing in the source seems to suggest a reason, it's just loops and simple math, which MS is generally better at across the board

 

2) All of the math-pack versions put paid to the argument that binary math is faster than BCD. As you can see, XE and Turbo both run at least on the same performance level, somewhat faster on the rugg8 tests and somewhat slower (although only slightly) in the maths test in scruss

 

3) another curiosity is Turbo's performance of the for loop relative to XE in fast mode. This would seem to suggest that XE's math is slower? But the relative difference in the math test suggests this is not the source of the difference.

Edited June 11, 2021 by Maury Markowitz