Messages in this thread | | | From | Con Kolivas <> | Subject | Re: [PATCH] O12.2int for interactivity | Date | Sat, 16 Aug 2003 12:29:03 +1000 |
| |
On Sat, 16 Aug 2003 04:17, Timothy Miller wrote: > >All "nice" 0 tasks get the same size timeslice. If their dynamic priority > > is different (the PRI column in top) they still get the same timeslice. > > Why isn't dynamic priority just an extension of static priority? Why do > you modify only the ordering while leaving the timeslice alone?
Because master engineer Molnar has determined that's the correct way.
> So, tell me if I infer this correctly: If you have a nice 5 and a nice > 7, but the nice 5 is a cpu hog, while the nice 7 is interactive, then > the interactivity scheduler can modify their dynamic priorities so that > the nice 7 is being run before the nice 5. However, despite that, the > nice 7 still gets a shorter timeslice than tha nice 5. > > Have you tried altering this?
Yes, not good with fluctuating timeslices all over the place makes for more bounce in the algorithm, and the big problem - the cpu intensive applications get demoted to smaller timeslices and they are the tasks that benefit the most from larger timeslices (for effective cpu cache usage).
Con
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |