Re: [Qemu-devel] [PATCH v1 00/17] dataplane: optimization and multi virtqueue support

[Ming Lei]

On Wed, Aug 6, 2014 at 7:28 PM, Ming Lei <address@hidden> wrote:
> On Wed, Aug 6, 2014 at 6:09 PM, Kevin Wolf <address@hidden> wrote:

>
> I use the /dev/nullb0 block device to test, which is available in linux kernel
> 3.13+, and follows the difference, which looks not very big(< 10%):
>
> And I added two parameter to your img-bench patch:
>
>       -c CNT  # which is passed to 'data.n'
>       -b           #enable bypass coroutine introduced in this patchset
>
> Another difference is that dataplane uses its own thread, and this
> bench takes main_loop.
>
> ming@:~/git/qemu$ sudo ~/bin/perf stat -e
> L1-dcache-loads,L1-dcache-load-misses,cpu-cycles,instructions,branch-instructions,branch-misses,branch-loads,branch-load-misses,dTLB-loads,dTLB-load-misses
> ./qemu-img bench -f raw -t off -n -c 10000000  -b /dev/nullb0
> read time: 58024ms
>
>  Performance counter stats for './qemu-img bench -f raw -t off -n -c
> 10000000 -b /dev/nullb0':
>
>     34,874,462,357      L1-dcache-loads
>               [40.00%]
>        714,018,039      L1-dcache-load-misses     #    2.05% of all
> L1-dcache hits   [40.00%]
>    133,897,794,677      cpu-cycles                [40.05%]
>    116,714,230,004      instructions              #    0.87  insns per
> cycle         [50.02%]
>     22,689,223,546      branch-instructions
>               [50.01%]
>        391,673,952      branch-misses             #    1.73% of all
> branches         [50.00%]
>     22,726,856,215      branch-loads
>               [50.01%]
>     18,570,766,783      branch-load-misses
>               [49.98%]
>     34,944,839,907      dTLB-loads
>               [39.99%]
>         24,405,944      dTLB-load-misses          #    0.07% of all
> dTLB cache hits  [39.99%]
>
>       58.040785989 seconds time elapsed
>
>
> ming@:~/git/qemu$ sudo ~/bin/perf stat -e
> L1-dcache-loads,L1-dcache-load-misses,cpu-cycles,instructions,branch-instructions,branch-misses,branch-loads,branch-load-misses,dTLB-loads,dTLB-load-misses
> ./qemu-img bench -f raw -t off -n -c 10000000  /dev/nullb0
> read time: 63369ms

BTW, Stefan's coroutine resize patch is applied in both the
tests(qem-img bench).

Thanks,