From: Sebastian Andrzej Siewior <bige...@linutronix.de> It popped up in perf testing that the worker consumes some amount of CPU. It boils down to the increment of `ops` which causes cache line bouncing between the individual threads.
This patch aligns the struct by 256 bytes to ensure that not a cache line is shared among CPUs. 128 byte is the x86 worst case and grep says that L1_CACHE_SHIFT is set to 8 on s390. Signed-off-by: Sebastian Andrzej Siewior <bige...@linutronix.de> Cc: Davidlohr Bueso <dbu...@suse.de> Cc: Peter Zijlstra <pet...@infradead.org> Link: http://lkml.kernel.org/r/20161016190803.3392-1-bige...@linutronix.de Signed-off-by: Arnaldo Carvalho de Melo <a...@redhat.com> --- tools/perf/bench/futex-hash.c | 5 ++++- 1 file changed, 4 insertions(+), 1 deletion(-) diff --git a/tools/perf/bench/futex-hash.c b/tools/perf/bench/futex-hash.c index 8024cd5febd2..d9e5e80bb4d0 100644 --- a/tools/perf/bench/futex-hash.c +++ b/tools/perf/bench/futex-hash.c @@ -39,12 +39,15 @@ static unsigned int threads_starting; static struct stats throughput_stats; static pthread_cond_t thread_parent, thread_worker; +#define SMP_CACHE_BYTES 256 +#define __cacheline_aligned __attribute__ ((aligned (SMP_CACHE_BYTES))) + struct worker { int tid; u_int32_t *futex; pthread_t thread; unsigned long ops; -}; +} __cacheline_aligned; static const struct option options[] = { OPT_UINTEGER('t', "threads", &nthreads, "Specify amount of threads"), -- 2.7.4
