Quantifying the cost of context switch ====================================== Note that this is not my own code but rather imported from the archived sources here: * http://cs.rochester.edu/u/cli/research/switch.htm The original authors and the corresponding paper can be found here: * http://dl.acm.org/citation.cfm?id=1281702 Measuring the indirect cost of context switch is a challenging problem. In this paper, we show our results of experimentally quantifying the indirect cost of context switch using a synthetic workload. Specifically, we measure the impact of program data size and access stride on context switch cost. We also demonstrate the potential impact of OS background interrupt handling on the measurement accuracy. Such impact can be alleviated by using a multi-processor system on which one processor is employed for context switch measurement while the other runs OS background tasks. 0. What are in the directory: measureSwitch.c -- two processes communication via a pipe measureSingle.c -- single process simulating two processes communications util.c -- utility program for context switch cost measurement util.h -- utility header file 1. Compile: make -- compile the programs on a Unix platform 2. Usage: ./measureSingle [-n ArraySize] [-s StrideSize] ./measureSwitch [-n ArraySize] [-s StrideSize] 2.1 By default, ArraySize and StrideSize are both zeros. Both ArraySize and StrideSize need to be given in byte. 2.2 Remember to give the same parameters to measureSingle and measureSwitch. 3. Output: measureSingle outputs time1 time1 = Overhead of traversing through array + pipe overhead measureSwitch outputs time2 time2 = time1 + context switch overhead total cost of context switch = time2 - time1 (microseconds) indirect cost of context switch = total cost - direct cost 4. Note: 4.1 When both parameters are zeros, the result (time2 - time1) is the direct cost of context switch. 4.2 We separate measureSingle and measureSwitch into two programs in order to avoid interference they may cause to each other.