Name Last Update
..
plots Loading commit data...
asm32.txt Loading commit data...
plot.py Loading commit data...
readme.md Loading commit data...
x86_64.txt Loading commit data...

Context Switch Measurements

The goal of these 'quick and dirty' measurements is to get a feeling for the expected minimum runtime of different operations involved in cotext switching/ stackfull coroutines.

We compare a jump_buf + stack pointer manipulation method, boost.context (represented the c wrapper around their assembly to avoid importing whole boost), a custom fiber call written by us in assembly and finally some baseline for method call costs.

We plot the results for both an x86_64 and arm32 system to decide what implementation we choose for pls, as this operation will be performed for every single task creation.