Messages in this thread | | | Date | Mon, 17 Nov 1997 11:45:23 +0100 (MET) | From | Rik van Riel <> | Subject | Re: [PATCH] vhand-2.1.63 |
| |
On Sat, 15 Nov 1997, Joe Fouche wrote: > > It looks to me like the code in kswapd to wake vhand up is being called too > often. After commenting out the following in vmscan.c (swap_tick): > > /* > if (jiffies > next_scan_jiffies) { > if (!vhand_awake) { > wake_up(&vhand_wait); > need_resched = 1; > } > } > */ > > vhand no longer uses an appreciable amount of CPU. Performance still seems > improved. Comments?
Hmm, how about kswapd CPU usage... Anyways, since performance went up, I think I'm going to try this myself. Or otherwise we could implement a less agressive tuning method, a maximum delay of HZ/10, or whatever...
Time to tune the tuning,
Rik.
---------- Send Linux memory-management wishes to me: I'm currently looking for something to hack...
| |