lkml.org 
[lkml]   [2016]   [May]   [17]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
/
From
Subject[PATCH v3 0/6] powerpc use pv-qpsinlock instead of spinlock
Date
change fome v1:
separate into 6 pathes from one patch
some minor code changes.

benchmark test results are below.
run 3 tests on pseries IBM,8408-E8E with 32cpus, 64GB memory
perf bench futex hash
perf bench futex lock-pi
perf record -advRT || perf bench sched messaging -g 1000 || perf report

summary:
_____test________________spinlcok______________pv-qspinlcok_____
|futex hash | 556370 ops | 629634 ops |
|futex lock-pi | 362 ops | 367 ops |
|workloads | 23% | 13% |

details:

spinlock:
test 1)
# Running futex/hash benchmark...
Run summary [PID 9962]: 32 threads, each operating on 1024 [private] futexes for 10 secs.
Averaged 556370 operations/sec (+- 0.61%), total secs = 10

test 2)
# Running futex/lock-pi benchmark...
Run summary [PID 9962]: 32 threads doing pi lock/unlock pairing for 10 secs.
Averaged 362 operations/sec (+- 0.00%), total secs = 10

test 3)
perf bench sched messaging -g 1000
perf record -avdRT and perf report
# Samples: 2M of event 'cycles:ppp'
# Event count (approx.): 2045582241213
#
# Overhead Command Shared Object Symbol
# ........ ............... ................... ........................................................
#
22.96% sched-messaging [kernel.kallsyms] [k] _raw_spin_lock_irqsave
19.76% sched-messaging [kernel.kallsyms] [k] __spin_yield
2.09% sched-messaging [kernel.kallsyms] [k] __slab_free
2.07% sched-messaging [kernel.kallsyms] [k] unix_stream_read_generic


pv-qspinlock:

test 1)
# Running futex/hash benchmark...
Run summary [PID 3219]: 32 threads, each operating on 1024 [private] futexes for 10 secs.
Averaged 629634 operations/sec (+- 0.38%), total secs = 10

test 2)
# Running futex/lock-pi benchmark...
Run summary [PID 3219]: 32 threads doing pi lock/unlock pairing for 10 secs.
Averaged 367 operations/sec (+- 0.00%), total secs = 10

test 3)
perf bench sched messaging -g 1000
perf record -avdRT and perf report
# Samples: 1M of event 'cycles:ppp'
# Event count (approx.): 1250040606393
#
# Overhead Command Shared Object Symbol
# ........ ............... .............................. ........................................................
#
9.87% sched-messaging [kernel.vmlinux] [k] __pv_queued_spin_lock_slowpath
3.66% sched-messaging [kernel.vmlinux] [k] __pv_queued_spin_unlock
3.37% sched-messaging [kernel.vmlinux] [k] __slab_free
3.06% sched-messaging [kernel.vmlinux] [k] unix_stream_read_generic


Pan Xinhui (6):
qspinlock: powerpc support qspinlock
powerpc: pseries/Kconfig: qspinlock build config
powerpc: lib/locks.c: cpu yield/wake helper function
pv-qspinlock: powerpc support pv-qspinlock
pv-qspinlock: use cmpxchg_release in __pv_queued_spin_unlock
powerpc: pseries: pv-qspinlock build config/make

arch/powerpc/include/asm/qspinlock.h | 39 +++++++++++++++++++
arch/powerpc/include/asm/qspinlock_paravirt.h | 38 +++++++++++++++++++
.../powerpc/include/asm/qspinlock_paravirt_types.h | 13 +++++++
arch/powerpc/include/asm/spinlock.h | 31 +++++++++------
arch/powerpc/include/asm/spinlock_types.h | 4 ++
arch/powerpc/kernel/Makefile | 1 +
arch/powerpc/kernel/paravirt.c | 44 ++++++++++++++++++++++
arch/powerpc/lib/locks.c | 36 ++++++++++++++++++
arch/powerpc/platforms/pseries/Kconfig | 9 +++++
arch/powerpc/platforms/pseries/setup.c | 5 +++
kernel/locking/qspinlock_paravirt.h | 2 +-
11 files changed, 209 insertions(+), 13 deletions(-)
create mode 100644 arch/powerpc/include/asm/qspinlock.h
create mode 100644 arch/powerpc/include/asm/qspinlock_paravirt.h
create mode 100644 arch/powerpc/include/asm/qspinlock_paravirt_types.h
create mode 100644 arch/powerpc/kernel/paravirt.c

--
1.9.1

\
 
 \ /
  Last update: 2016-05-17 10:41    [W:0.067 / U:0.496 seconds]
©2003-2020 Jasper Spaans|hosted at Digital Ocean and TransIP|Read the blog|Advertise on this site