Messages in this thread | | | Date | Wed, 19 Jan 2000 12:25:29 -0500 (EST) | From | Mark Hahn <> | Subject | Re: Interesting analysis of linux kernel threading by IBM |
| |
> > http://www-4.ibm.com/software/developer/library/java2/index.html > > I read it. It was a excellent paper. I'm amazed by just rearranging the > data in task_struct it reduces the time in the schedular up to 30% on
I read it and was underwhelmed. anyone who follows linux-kernel would know that the scheduler is optimized for normal loads. that is, those with only a handfull of runnable threads. people have rearranged task_struct before, but that was a while ago, and fast development implies high code entropy. the paper _was_ organized nicely ;)
> Pentium II. I have the patch included below. Anyone working on a > many-to-many model for threads for linux?
god, I hope not. user-level threading is Just Plain Wrong if you can possibly avoid it. the only venture in that direction I can imagine would be Mach-like scheduler activations. though I'm sure it would take a pretty major problem with the existing scheduler to motivate Linus &co to take that step...
regards, mark hahn. -- operator may differ from spokesperson. hahn@coffee.mcmaster.ca http://java.mcmaster.ca/~hahn
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |