Files
linux/arch/powerpc/kernel
Anton Blanchard f5f0307f42 powerpc: Improve scheduling of system call entry instructions
After looking at our system call path, Mary Brown suggested that we
should put all mfspr SRR* instructions before any mtspr SRR*.

To test this I used a very simple null syscall (actually getppid)
testcase at http://ozlabs.org/~anton/junkcode/null_syscall.c

I tested with the following changes against the pseries_defconfig:

CONFIG_VIRT_CPU_ACCOUNTING=n
CONFIG_AUDIT=n

to remove the overhead of virtual CPU accounting and syscall
auditing.

POWER6:
baseline:       mean = 757.2 cycles       sd = 2.108
modified:       mean = 759.1 cycles       sd = 2.020

POWER7:
baseline:       mean = 411.4 cycles       sd = 0.138
modified:       mean = 404.1 cycles       sd = 0.109

So we have 1.77% improvement on POWER7 which looks significant. The
POWER6 suggest a 0.25% slowdown, but the results are within 1
standard deviation and may be in the noise.

Signed-off-by: Anton Blanchard <anton@samba.org>
Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
2011-05-19 14:30:42 +10:00
..
2011-03-31 11:26:23 -03:00
2011-03-31 11:26:23 -03:00
2011-03-31 11:26:23 -03:00
2011-04-27 13:02:02 +10:00
2011-03-30 10:44:00 +11:00
2011-03-31 11:26:23 -03:00
2011-03-31 11:26:23 -03:00
2011-04-27 14:18:44 +10:00
2011-03-31 11:26:23 -03:00
2011-05-04 15:02:15 +10:00
2011-03-31 11:26:23 -03:00
2010-10-05 17:27:54 -07:00
2011-03-31 11:26:23 -03:00
2010-12-09 15:35:32 +11:00
2011-03-31 11:26:23 -03:00
2011-03-31 11:26:23 -03:00
2011-04-27 14:18:44 +10:00
2010-09-22 09:33:50 -07:00
2010-07-19 10:56:54 +02:00
2011-03-31 11:26:23 -03:00
2010-03-12 15:52:32 -08:00