Dmitry Sidorov

Can you please check with [2.0.0](https://github.com/JonMagon/KDiskMark/releases/tag/2.0.0) and [1.6.2](https://github.com/JonMagon/KDiskMark/releases/tag/1.6.2)? ``` mkdir build && cd build cmake -D CMAKE_BUILD_TYPE=Release .. cmake --build . ```

Then it may be caused by separation loops inside kdiskmark. Could you please try also [1.6.0](https://github.com/JonMagon/KDiskMark/releases/tag/1.6.0)?

Then it can be really not kdiskmark issue.

What is the output of the commands below? ```bash echo 3 | sudo tee /proc/sys/vm/drop_caches fio --ioengine=libaio --direct=1 --randrepeat=0 --refill_buffers --end_fsync=1 --filename=test-fio.tmp --name=readjob --size=128M --bs=1m --rw=read --iodepth=8 --loops=5 ``` ```bash...

I would guess that this is the real reading performance. What if test the speed with dd? Copy the all output. ```bash dd if=/dev/zero of=test-fio.tmp bs=1M count=128 echo 3 |...

No idea, something wrong with fio.

By the way, @tim77 haven't you changed partitions since the results was correct? Maybe alignment is wrong...

I think you should open an issue in the [fio repository](https://github.com/axboe/fio), I can't resolve it.

I understand perfectly, but I can't fix it myself. It's one thing if it never worked right, it's quite another thing if it suddenly broke down, and with only one...

I plan to create this in the near future.