Related research papers: |
Usually we observe a 10%-100% overhead compared with running g++ without MC enabled. The speed depends on the SM and the size of the input in terms of # of AST nodes. If "per-variable state" is used then it also depends on the number of variables in the input. The metal paper talks a little more about variables that affect running time.We haven't observed swapping yet so we don't care too much about the memory usage. I expect that it should be approximately linear in the input size, but don't quote me on it.
Return to the main OSQ
page.