Yesterday I created a simple script to benchmark CPU's instruction execution speed with Python. I should not have use numpy since this library degraded the MIPS result a lot.
Another run for fun small Python program to benchmark your CPU's instruction speed (MIPS) is here. Just run (no dependency at all!) and enjoy the result.
https://brew.bsd.cafe/maulanahirzan/Python-Useful-Scripts/src/branch/main/MIPS-Benchmark
@maulanahirzan Seems to run faster writing output to a file than to the terminal (in my test).
screen:
=> Result : 1220176.98 IPS / 1.22 MIPS
(~ 5 sconds)
file:
=> Result : 8626522.56 IPS / 8.63 MIPS
(< 1 second)
@jspath55 I see. thanks for the feedback. I will test it immediately. 😊
EDIT: found the culprit in the run function. Thanks to you I found what code that degrades the MIPS result. The result still good with normal stdout now. I will update the git immediately. feel free to test it again.😉
@maulanahirzan More consistent results now, a range of +/- 20% overhead on my tests?
$ python3 ./MIPSBenchmark_2.py | grep MIPS\$
=> Result : 48304223.15 IPS / 48.30 MIPS
=> Result : 50137258.69 IPS / 50.14 MIPS
=> Result : 42309927.51 IPS / 42.31 MIPS
=> Result : 49413017.65 IPS / 49.41 MIPS
=> Result : 45110298.50 IPS / 45.11 MIPS
=> Result : 43321850.73 IPS / 43.32 MIPS
=> Result : 43241075.31 IPS / 43.24 MIPS
=> Result : 43026256.75 IPS / 43.03 MIPS
=> Result : 43497130.33 IPS / 43.50 MIPS
That's cool - I recently upgraded my #GoPiGo3 #ROS_2 #Robot to #RaspberryPi5
50.62 MIPS on Pi5 running 64-bit #PiOSBookworm with #Python 3.11.2:
@maulanahirzan I had not heard anyone talking MIPS in a long time.
Wiki: "Meaningless Indicator of Processor Speed" and Tom's Hardware:
A. M. H.
vermaden
Jim Spath
cyclical_obsessive