On 12/04/2017 11:25 AM, Gavriliuk, Anton (HPS Ukraine) wrote: > Ok, now test for 10 minutes with the same I/O latencies . So there is no startup overhead. > > dl560g10spmem01:~ # perf record /usr/local/bin/fio --filename=/dev/dax0.0 --rw=randrw --refill_buffers --norandommap --randrepeat=0 --ioengine=mmap --bssplit=4k/4:8k/7:16k/7:32k/15:64k/65:128k/1:256k/1 --rwmixread=5 --iodepth=1 --numjobs=16 --runtime=600 --group_reporting --name=4-rand-rw-3xx --size=290g --numa_cpu_nodes=0 --time_based=1 > 4-rand-rw-3xx: (g=0): rw=randrw, bs=(R) 4096B-256KiB, (W) 4096B-256KiB, (T) 4096B-256KiB, ioengine=mmap, iodepth=1 > ... > fio-3.2-64-gaf7fb > Starting 16 processes > > read: IOPS=36.7k, BW=1865MiB/s (1956MB/s)(1093GiB/600002msec) > clat (nsec): min=120, max=16029k, avg=6897.44, stdev=5503.68 > lat (nsec): min=144, max=16029k, avg=6932.35, stdev=5504.08 > write: IOPS=697k, BW=34.6GiB/s (37.2GB/s)(20.3TiB/600002msec) > clat (nsec): min=111, max=72027k, avg=12546.82, stdev=9845.85 > lat (nsec): min=132, max=72027k, avg=12592.74, stdev=9851.98 > > Samples: 38M of event 'cycles:ppp', Event count (approx.): 182321387032131 > Overhead Command Shared Object Symbol > 77.20% fio fio [.] clock_thread_fn > 8.32% fio libc-2.22.so [.] __memcpy_avx_unaligned > 6.07% fio fio [.] fill_random_buf > 4.82% fio [kernel.kallsyms] [k] page_remove_rmap > 3.02% fio [kernel.kallsyms] [k] native_write_msr_safe > 0.10% fio fio [.] get_io_u That seems so odd, it makes no sense. --no-children added to perf report? I think this is skewed because the code is so hot for startup. -- Jens Axboe -- To unsubscribe from this list: send the line "unsubscribe fio" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
