Wasting cpu cycles with python

Question

I am trying to create a simple app that wastes cpu cycles for multi-core research. The one I created takes up 100% core usage. I want it to be around 30% 60% 70%, which adjustments should I make in order to achieve this? Thanks in advance.

Current version:

a=999999999
while True:
   a=a/2

it is not the answer, but maybe it should: don't use Python for that — MariusSiuram, Dec 09 '16 at 11:41
`import time; while 1: time.sleep(n)` for a suitable size of `n`. — TigerhawkT3, Dec 09 '16 at 12:16
@mozcelikors you might want to refer this link http://xmodulo.com/run-program-process-specific-cpu-cores-linux.html — Prabhakar Lad, Dec 09 '16 at 12:23
@TigerhawkT3 that seems a valid answer on itself. Taking into account that `n` is "seconds" and can be a floating point number. — MariusSiuram, Dec 09 '16 at 15:22
@TigerhawkT3 if u want to answer the question, I can accept the answer. It worked great. Just a challange of finding the right 'n'. So, thanks. — mozcelikors, Dec 10 '16 at 14:53

score 1 · Accepted Answer · edited May 23 '17 at 12:01

Starting at a large number isn't necessary, as dividing a number by 2 will quickly end up as 0/2 over and over again anyway. Besides, you don't have to actually do anything in a loop to consume CPU cycles - the mere action of looping is enough. This is why any infinite loop, even something as simple as while 1: pass, will eat up an entire CPU core until killed. To avoid taking up an entire core, use time.sleep to pause execution of the thread for a certain period of time. This function takes a single argument representing the time in seconds for the thread to sleep. It accepts a floating-point number.

import time
while 1:
    time.sleep(0.0001)

Simply run an instance of this script (with an appropriate sleep time for the workload you'd like to put on your particular system) for each core you'd like to test.

Note that some operating systems may not support sleep times of less than one millisecond, causing shorter sleep times to come through as zero, making them incompatible with this strategy. See Python: high precision time.sleep and How accurate is python's time.sleep()? for more.

Wasting cpu cycles with python

1 Answers1