Re: Can and should the kernel HZ value be changed?
Jamie Lokier (lkd@tantalophile.demon.co.uk)
Sat, 9 Jan 1999 09:44:29 +0000
On Wed, Jan 06, 1999 at 08:58:25PM +0100, MOLNAR Ingo wrote:
> On Tue, 5 Jan 1999, Egil Kvaleberg wrote:
> > [...] The is pretty vital for any real-timeish application, and
> > nicely visualized and obvious (as well as gratifying for ones sanity) when
> > running X.
>
> so, _why_ does it make a difference for X, and can you quantify it with
> some reproducible testcase? It's not the number of context switches for
> CPU-bound processes (this is what HZ really means), thats ludicrous.
> (mouse events get to X just as fast as keyboard events) I do not say you
> dont see a speedup, i'd just like to fix the real reason ...
>
> my best guess is that menus and other widgets use low-value [mandatory]
> select() timeouts, which translate into a 10-20msec delay. Changing HZ to
> something more finegrained appears to the advanced user as a speedup. The
> solution to this problem is not to change HZ, but rather to (maybe)
> implement arbitrary precision timers. (which implmentation btw. already
> exists)
Someone mentioned that rebuilding the kernel was significantly faster on
their machine, with HZ == 400 instead of 100.
More rapid bottom-half handling is the only reason I can think of for
that. Apart from different I/O characteristics, I'd expect the
CPU-bound part of the compile to run _slower_ because of the increase
in cache flushing.
-- Jamie
-
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/