Don’t use time.time()

Note#75

notes
Published

2023-04-17

When measuring time intervals in a Python program, it is recommended to avoid using time.time() and choose time.monotonic() or time.monotonic_ns() instead.

Problems with time.time()

  • Returns the current time in seconds since the epoch (usually January 1, 1970, 00:00:00 UTC).
  • Commonly used to measure the time elapsed between two points in a program.
  • Not immune to system clock adjustments, like Daylight Saving Time changes or manual clock updates.

Benefits of time.monotonic() and time.monotonic_ns()

time.monotonic()

  • Returns a monotonic clock value, which is a clock that always increases.
  • Not affected by system clock adjustments.
  • Useful for measuring the time elapsed between two points in a program when you need a clock that won’t be influenced by external factors.
  • The returned value is not an absolute timestamp, and its meaning is specific to the program it’s used in.

time.monotonic_ns()

  • Similar to time.monotonic(), but it returns the value in nanoseconds.
  • Provides higher resolution for time measurements.
  • Useful for measuring small time intervals with high accuracy.

When working with very small time intervals or requiring high-precision measurements, consider using time.monotonic_ns() instead of time.monotonic() to avoid the precision loss caused by the float type

Example: Reformatting time.monotonic_ns() output for human readability

You can easily convert the output of time.monotonic_ns() to seconds for better readability:

import time

start_ns = time.monotonic_ns()
# Your code here
end_ns = time.monotonic_ns()

elapsed_seconds = (end_ns - start_ns) / 1e9
print(f"Elapsed time: {elapsed_seconds:.5f} seconds")

By using time.monotonic() or time.monotonic_ns() instead of time.time(), you can avoid potential inaccuracies due to system clock adjustments and achieve more reliable time measurements.