Messages in this thread | | | Date | Thu, 5 Aug 1999 11:30:45 -0400 (EDT) | From | "Robert G. Brown" <> | Subject | Re: Measured overhead of timer interrupts |
| |
On Tue, 20 Jul 1999, Pavel Machek wrote:
> Hi! > > > > On Mon, Jul 19 1999, Artur Skawina wrote: > > > > what i would be interested in seeing is: the time it takes to > > > > run a cpu bound app (eg raytracing am image) with HZ=100 and HZ=1024. > > > > That would give a more realistic approximation of the overhead that > > > > increasing HZ adds. > > > > > > Then do the bench, nobody prevents you from doing so. > > > > Should HZ=1024 be safe on x86? > > Yes. Irda used to be broken w.r.t. HZ but is fixed now. Oh and you'll > need to recompile things like top. But with exception of pstools > HZ=1024 i386 system works just well.
Sorry for the late hit (I've been on vacation) but, would it be worthwhile to add HZ to a file in /proc, e.g. /proc/sys/kernel/hz? One could then patch and recompile pstools (and possibly things like games that might care) to read the value. Indeed, one could even permit the value to be set via the proc point and let people retune from coarse to ultra-fine as CPU speeds and the like continue to increase.
Just an opinion, but a solution that is properly integrated and is scalable in the future seems preferable to just altering the #define.
rgb
Robert G. Brown http://www.phy.duke.edu/~rgb/ Duke University Dept. of Physics, Box 90305 Durham, N.C. 27708-0305 Phone: 1-919-660-2567 Fax: 919-660-2525 email:rgb@phy.duke.edu
- 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/
| |