Re: fio bug (?)

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

 



On 11/24/22 05:38, Mateusz Masiarz wrote:
> Hello, I'm trying to benchmark pendrives with fio. I'm running:
> fio --name=test --ioengine=posixaio --direct=1 --rw=randwrite --bs=4k
> --size=4g --numjobs=1 --iodepth=1 --runtime=60 --group_reporting
> --time_based

Try ioengine=psync and you can drop iodepth=1 as it is useless. That is
equivalent to aio with iodepth=1.
You can also try ioengine=libaio (with iodepth >= 1) on Linux.

> the io file creates normally, but then the job starts. Firstly, the
> write speed is not shown. Secondly, after the job runs for one minute,
> the completion percentage drops to 50 and the eta shows a few thousand
> days and the job doesn't end. After about 5 minutes of running the job
> ends, but write speed is supposedly only a few bytes per second, which
> (I think) is wrong, because crystaldisk mark shows around a few
> megabytes per second with a similar random write test. I'm running
> fio-3.33 and I tested this on MacBook M1, raspberry pi, some old
> kubuntu system and on Windows, and every result was similar.
> By the way, is it normal for "Laying down IO file" to take a while?

-- 
Damien Le Moal
Western Digital Research




[Index of Archives]     [Linux Kernel]     [Linux SCSI]     [Linux IDE]     [Linux USB Devel]     [Video for Linux]     [Linux Audio Users]     [Yosemite News]     [Linux SCSI]

  Powered by Linux