Hi David, Thanks for your interest in the patch.
On Sun, Jul 27, 2014 at 10:20 PM, David Horner <[email protected]> wrote: > You mention > 100 % improvement. > > Can you help me to run some (micro) benchmarks that I could contribute to > V1? Sure, no problem, generally I just run tests over the loop block device directly, since I want to expose performance improvement. But I did run test bench(mkfs, copy big file and compare, ...) over filesystem mounted on the loop device too, and looks it is stable enough. So your benchmarks are either on loop block or filesystem, both are fine. > If you can direct me to relevant threads and docs that would be greatly > appreciated. Follows my test which was observed with 100% improvement( throughout is doubled, actually it is more than doubled): - losetup -f /dev/nullb0 #so /dev/loop0 will be backed with /dev/nullb0 #if it is the 1st loop device - run below fio script[1] in single job to test direct randread on /dev/loop0 - it is just for single job improvement, with more than 1 jobs, the throughout should be increased more with blk-mq conversion I will send out V1 this week, and more test cases may be posted too. [1], fio test script [global] direct=1 size=32G bsrange=4k-4k timeout=20 numjobs=1 ioengine=libaio iodepth=64 filename=/dev/loop0 group_reporting=1 [f] rw=randread Thanks, -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected] More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
