Skip to content

Commit

Permalink
powerpc/numa: Disable NEWIDLE balancing at node level
Browse files Browse the repository at this point in the history
On big POWER7 boxes we see large amounts of CPU time in system
processes like workqueue and watchdog kernel threads.

We currently rebalance the entire machine each time a task goes
idle and this is very expensive on large machines. Disable newidle
balancing at the node level and rely on the scheduler tick to
rebalance across nodes.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
  • Loading branch information
Anton Blanchard authored and Benjamin Herrenschmidt committed Sep 20, 2011
1 parent d4761ad commit 7bebcf0
Showing 1 changed file with 1 addition and 1 deletion.
2 changes: 1 addition & 1 deletion arch/powerpc/include/asm/topology.h
Original file line number Diff line number Diff line change
Expand Up @@ -75,7 +75,7 @@ static inline int pcibus_to_node(struct pci_bus *bus)
.forkexec_idx = 0, \
\
.flags = 1*SD_LOAD_BALANCE \
| 1*SD_BALANCE_NEWIDLE \
| 0*SD_BALANCE_NEWIDLE \
| 1*SD_BALANCE_EXEC \
| 1*SD_BALANCE_FORK \
| 0*SD_BALANCE_WAKE \
Expand Down

0 comments on commit 7bebcf0

Please sign in to comment.