Python 3.8 time.time() Precision on AMD Ryzen 7 3700X 8 cores 16 threads CPU on Ubuntu 20.04.2 LTS

https://stackoverflow.com/questions/1938048/high-precision-clock-in-python

 Python 3.8 time.time() Precision on AMD Ryzen 7 3700X 8 cores 16 threads CPU on Ubuntu 20.04.2 LTS

Test 1 time.time()

import time

lis = [time.time() for _ in range(1000)]

for index, x  in enumerate(lis):

print(str(index).zfill(3), str('%.15f'%x)) # milli micro nano pico femto

From below result, within the same micro second, the nano second differs by 200-300ns.

A rough description is that Python's time.time() can make timestamps in 200-300ns.

000 1627032470.455012798309326

001 1627032470.455013036727905

002 1627032470.455013036727905

003 1627032470.455013036727905

004 1627032470.455013275146484

005 1627032470.455013275146484

006 1627032470.455013275146484

007 1627032470.455013513565063

008 1627032470.455013513565063

Test 2 time.time_ns()

import time

lis = [time.time_ns() for _ in range(1000)]

for index, x  in enumerate(lis):

print(str(index).zfill(3), x) # milli micro nano not pico femto

Stackoverflows says:

"For Linux and Mac precision is +- 1 microsecond or 0.001 milliseconds"

"Windows uses +- 16 milliseconds precision due to clock implementation problems due to process interrupts"

My test says that it is 100ns not 1us. 10 times higher precision.

second milli micro nano

000 1627032860 145 901 780

001 1627032860 145 902 021

002 1627032860 145 902 131

003 1627032860 145 902 231

004 1627032860 145 902 311

005 1627032860 145 902 431

006 1627032860 145 902 522

007 1627032860 145 902 592

008 1627032860 145 902 682

009 1627032860 145 902 792

010 1627032860 145 902 882

011 1627032860 145 902 962

012 1627032860 145 903 042

So, precision on Ubuntu is 100ns, meaning for every 2 consecutive execution the interval is 100ns.