Closed ghost closed 9 years ago
This is quite a bit of code. And much of it is suboptimal. But this adds a basic work-stealing thread-pool that runs the tasks.
Before: test chain_10_use_die ... bench: 202637 ns/iter (+/- 110323) test chain_10_wait ... bench: 84634 ns/iter (+/- 9213) test chain_1_000_use_die ... bench: 7872877 ns/iter (+/- 1098996) test chain_1_000_wait ... bench: 7806022 ns/iter (+/- 1312216) test fibb_depth_6 ... bench: 2969794 ns/iter (+/- 1723503) test start_die ... bench: 445 ns/iter (+/- 7) After: test bench_fibb_steal ... bench: 405139 ns/iter (+/- 90794) test chain_10_use_die ... bench: 939562 ns/iter (+/- 294590) test chain_10_wait ... bench: 120315 ns/iter (+/- 34264) test chain_1_000_use_die ... bench: 1138655 ns/iter (+/- 409262) test chain_1_000_wait ... bench: 596570 ns/iter (+/- 59771) test fanout_1_000 ... bench: 883532 ns/iter (+/- 350949) test repeat_100_x_100 ... bench: 2088028 ns/iter (+/- 230252) test repeat_1_000 ... bench: 578743 ns/iter (+/- 96315) test repeat_1_000_x_1_000 ... bench: 189189143 ns/iter (+/- 6672870) test start_die ... bench: 946777 ns/iter (+/- 143467)
Sorry, I've completely lost track of the changes. Please don't wait for me if you are blocked.
This is quite a bit of code. And much of it is suboptimal. But this adds a basic work-stealing thread-pool that runs the tasks.