I just applied a couple of patches. You can now use TEST_PROG_ARGS to pass flags to parrot when running tests. And the flags are now more interesting: -g disabled the computed goto core if you're on a machine that supports it in the first place, and -P now works for enabling the prederef core even if you don't build a shared library first. See docs/running.pod for details.
The purpose behind this was to make it easier to do comparative timings. Here they are for my machine: lifetest runops mopstest generations/sec core ops/sec (millions) =============== ====== ================== 371.7 normal 16.8 438.7 prederef 21.3 452.2 cgoto 51.0 445.6 jit 345.0 Oddly, computed goto beat out the x86 jit on lifetest by a nose. Those timings are the median of 3 runs of each of make lifetest TEST_PROG_ARGS=-g make lifetest TEST_PROG_ARGS=-P make lifetest TEST_PROG_ARGS= make lifetest TEST_PROG_ARGS=-j make mopstest TEST_PROG_ARGS=-g make mopstest TEST_PROG_ARGS=-P make mopstest TEST_PROG_ARGS= make mopstest TEST_PROG_ARGS=-j I saw extremely low variance on mopstest and the two slower lifetest cores, and 10 gen/sec or so for the two fastest lifetest runops cores.
