On 9 August 2016 at 00:02, j.Lucas <j_lucas@xxxxx> wrote: > > From output: > Run time = 600001 ms > queue depth = 1 > performed IO = 25740 > > So correct me if i am wrong, but 600001 / 25740 = 23.31 ms > However, fio reports an average latency of 38.24 ms. > Should those numbers be consistent? Let me repost some excerpts of your fio output: > fio-2.1.10 [...] > read : io=7541.2MB, bw=12870KB/s, iops=25740, runt=600001msec [...] > lat (usec): min=28, max=12057, avg=38.24, stdev=24.21 [...] > issued : total=r=15444203/w=0/d=0, short=r=0/w=0/d=0 25740 represents I/O operations *per second* (as opposed to total I/Os) and as such is already an average. Further, the latency was reported in microseconds whereas the runtime in milliseconds. However what you pasted says the total I/Os issued were 15444203 so: 600001/15444203*1000 = 38.84 (us) which is only slightly more than fio's reported 38.24 us. The higher number could be due to overhead inbetween each submitted I/O and differences between where we start and stop counting time. PS: Your fio version is quite old - you may want to upgrade... -- Sitsofe | http://sucs.org/~sits/ -- 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
