Re: CFQ is worse than other IO schedulers in some cases

From: Mike Galbraith
Date: Wed Feb 18 2009 - 03:05:51 EST


On Wed, 2009-02-18 at 14:00 +0800, Shan Wei wrote:

> In sysbench(version:sysbench-0.4.10), I confirmed followings.
> - CFQ's performance is worse than other IO schedulers when only multiple
> threads test.
> (There is no difference under single thread test.)
> - It is worse than other IO scheduler when
> I used read mode. (No regression in write mode).
> - There is no difference among other IO schedulers. (e.g noop deadline)
>
>
> The Test Result(sysbench):
> UNIT:Mb/sec
> __________________________________________________
> | IO | thread number |
> | scheduler |-----------------------------------|
> | | 1 | 3 | 5 | 7 | 9 |
> +------------|------|-------|------|------|------|
> |cfq | 77.8 | 32.4 | 43.3 | 55.8 | 58.5 |
> |noop | 78.2 | 79.0 | 78.2 | 77.2 | 77.0 |
> |anticipatory| 78.2 | 78.6 | 78.4 | 77.8 | 78.1 |
> |deadline | 76.9 | 78.4 | 77.0 | 78.4 | 77.9 |
> +------------------------------------------------+
ï
My Q6600 box agrees that cfq produces less throughput doing this test,
but throughput here is ~flat. Disk is external SATA ST3500820AS.
_________________________________________________
| IO | thread number |
| scheduler |----------------------------------|
| | 1 | 3 | 5 | 7 | 9 |
+------------|------|------|------|------|------|
|cfq | 84.4 | 89.1 | 91.3 | 88.8 | 88.8 |
|noop |102.9 | 99.3 | 99.4 | 99.7 | 98.7 |
|anticipatory|100.5 |100.1 | 99.8 | 99.7 | 99.6 |
|deadline | 97.9 | 98.7 | 99.5 | 99.5 | 99.3 |
+-----------------------------------------------+

> Steps to reproduce(sysbench):
>
> (1)#echo cfq > /sys/block/sda/queue/scheduler
>
> (2)#sysbench --test=fileio --num-threads=1 --file-total-size=10G --file-test-mode=seqrd prepare
>
> (3)#sysbench --test=fileio --num-threads=1 --file-total-size=10G --file-test-mode=seqrd run
> [snip]
> Operations performed: 655360 Read, 0 Write, 0 Other = 655360 Total
> Read 10Gb Written 0b Total transferred 10Gb (77.835Mb/sec)
> 4981.44 Requests/sec executed ~~~~~~~~~~~
> (4)#sysbench --test=fileio --num-threads=1 --file-total-size=10G --file-test-mode=seqrd cleanup
>
> (5)#sysbench --test=fileio --num-threads=5 --file-total-size=10G --file-test-mode=seqrd prepare
> (6)#sysbench --test=fileio --num-threads=5 --file-total-size=10G --file-test-mode=seqrd run
> [snip]
> Operations performed: 655360 Read, 0 Write, 0 Other = 655360 Total
> Read 10Gb Written 0b Total transferred 10Gb (43.396Mb/sec)
> 2777.35 Requests/sec executed ~~~~~~~~~~~~
> (7)#sysbench --test=fileio --num-threads=5 --file-total-size=10G --file-test-mode=seqrd cleanup
>
> when doing step 2 or 5, sysbench creats 128 files, and 80M each one.
> when doing step 4 or 7, sysbench deletes the files.
> when doing step 3 or 6, thread reads these files continuously and
> reads file-block-size(default:16Kbyte) at once, just like :
>
> t_0 t_0 t_0 t_0 t_0 t_0 t_0
> ^ ^ ^ ^ ^ ^ ^
> ---|-----|-----|-----|-----|-----|-----|--------
> file | 16k | 16k | 16k | 16k | 16k | 16k | 16k | ...
> ------------------------------------------------
> (num-threads=1)
>
> (t_0 stand for the first thread)
>
> t_0 t_1 t_2 t_3 t_4 t_0 t_1
> ^ ^ ^ ^ ^ ^ ^
> ---|-----|-----|-----|-----|-----|-----|--------
> file | 16k | 16k | 16k | 16k | 16k | 16k | 16k | ...
> ------------------------------------------------
> (num-threads=5)
>
> (the executed threads are decide by the thread scheduler)
>
>
> The Hardware Infos:
> Arch :x86_64
> CPU :4cpu; GenuineIntel 3325.087 MHz
> MEMORY :4044128kB
>
> ----
> Shan Wei
>
> --
> To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
> the body of a message to majordomo@xxxxxxxxxxxxxxx
> More majordomo info at http://vger.kernel.org/majordomo-info.html
> Please read the FAQ at http://www.tux.org/lkml/

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/