HP Forums

Full Version: Casio Calculator Performance Index Report
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
I just got the Casio fx-9860GIII and I put in one of the two standard benchmarks to test it.

The benchmark is located at

https://www.thimet.de/CalcCollection/Cal...mance.html

The Casio scores an index of about 4350. That is a very respectable score, according to the table on the site!

I wrote the benchmark in Python and since the Casio has no internal timer, I timed the calculator with a stopwatch, then adjusted the number of loops so the runtime is 5 seconds. The calculation of the index on the site is only correct for 1 loop. For multiple loops, the formula is

Index = 34 / T * Loops

so on the Casio, since I was able to run 640 loops in 5 seconds, the index is

Index = 34/5*640

which is an index of 4352.0

Here's the Python program I used:
Code:
from math import *
loops=640
for i in range(0,loops):
  r0=10
  while r0>0:
    x=r0
    x+=1
    x-=4.567E-4
    x+=70
    x-=69
    x*=7
    x/=11
    r0-=1
  x=log(x)
  x=sin(x)
  x=sqrt(x)
  x=sqrt(x)
print(x)
print("Index =",34/5*loops)
For reference, here are two other calculator performance benchmarks. The Casio fx-9860GIII is included in the Summation benchmark. It is fairly speedy for a relatively inexpensive model (U.S. pricing), as long as you don't use Casio BASIC. It predecessor, the Casio fx-9860GII, scores really high in the N-Queens benchmark when programmed in assembly.

Calculator Speed Benchmark using the N-Queens Problem

Summation based benchmark for calculators
(09-09-2023 01:00 PM)Steve Simpkin Wrote: [ -> ]For reference, here are two other calculator performance benchmarks. The Casio fx-9860GIII is included in the Summation benchmark. It is fairly speedy for a relatively inexpensive model (U.S. pricing), as long as you don't use Casio BASIC. It predecessor, the Casio fx-9860GII, scores really high in the N-Queens benchmark when programmed in assembly.

Calculator Speed Benchmark using the N-Queens Problem

Summation based benchmark for calculators

On the Savage benchmark (Byte 1985 #11 Inside the IBM PCs, pages 67-88)
the Casio gets 2500.000000001178 in about 4.5 seconds! This is quite accurate and very fast compared to the computers tested in the article.
Of course, the Casio doesn't run a standard BASIC but it does run MicroPython which I'm now using to compare all current and future calculators.

One great thing about current calculators (HP, TI, Casio, etc.) is that we can run the benchmarks in a common language (MicroPython). This gives a much more accurate comparison of the hardware than using each calculator's native interpreter or machine language. In those cases, you aren't really comparing the hardware directly. It's true that each calculator has its own implementation of MicroPython but the similarities are much greater than comparing TI-Basic to Casio Basic to HPPL, or whatever. I used to evaluate hardware from various vendors (everything from mainframes to programmable calculators) for a US government agency. Many times people would give me a system to evaluate and not specify what language to use. They couldn't understand the importance of standardizing the benchmarking language. I tried to explain that benchmarking is a science rather than an art but they just didn't seem to get it. Oh well, I got paid enormous amounts of money for my work even if very few people knew exactly what I did!

Code:
from math import *
a=1
for i in range(1,2500):
  a=tan(atan(exp(log(sqrt(a*a)))))+1
print(a)
Reference URL's