Blogbench RAID benchmarks

Francois Tigeot ftigeot at wolfpond.org
Mon Jul 11 12:03:50 PDT 2011


On Mon, Jul 11, 2011 at 07:39:11PM +0200, Francois Tigeot wrote:
> On Mon, Jul 11, 2011 at 10:19:37AM -0700, Matthew Dillon wrote:
> >     Try this:  Change arcmsr.c's cam_sim_alloc() call from this:
> > 			1,
> >     To this:
> > 			ARCMSR_MAX_OUTSTANDING_CMD,
> 
> Done, I'll start a new set of benchmarks with RAID0 configurations.

I hate to say it, but this modification didn't change anything
performance-wise.

On the fastest disk configuration (RAID0, 4 disks), the results are:

writes:           600
reads :        178372

The whole text output is attached.

-- 
Francois Tigeot
dragonfly:/root#blogbench -d /mnt/blogbench

Frequency = 10 secs
Scratch dir = [/mnt/blogbench]
Direct I/O: disabled
Spawning 3 writers...
Spawning 1 rewriters...
Spawning 5 commenters...
Spawning 100 readers...
Benchmarking for 30 iterations.
The test will run during 5 minutes.

  Nb blogs   R articles    W articles    R pictures    W pictures    R comments    W comments
       108       610959          5825        413264          5689        364571         17971
       155       726247          2873        482978          2405        451912         10633
       196       719003          3104        489948          2793        480023         11058
       246       724979          3453        488839          2678        470941         12072
       268       697794          1593        471824          1570        448765          6295
       303       658438          2279        449681          2198        429947          7225
       337       597568          2111        411796          2274        365589          5521
       359       494463          1344        341211          1192        302163          2808
       373       365185          1063        249377           726        217173          2703
       383       244130           622        168766           574        142801          2423
       392       233537           902        159389           554        138846          2663
       404       222294           721        150963           617        133133          2523
       413       208573           646        142463           557        119386          2298
       423       184486           855        128677           491        110635          2470
       436       193158          1010        131412           633        113013          2843
       445       177683           713        120144           611        105730          2807
       457       161034           691        110099           597         95486          2461
       468       182947           961        121084           578        109612          2448
       481       155982           761        104085           731         93549          2694
       488       153539           708        101753           599         96616          2619
       498       143743           691         96624           556         85150          2335
       510       143902           642         97564           739         86367          2557
       523       135071           840         91890           779         81569          2933
       534       133432           606         90209           666         78763          2544
       543       129323           671         88359           699         79309          2608
       555       135680           932         91768           523         84829          2484
       571       134262           920         90410           819         81608          2760
       580       127395           651         87013           749         75147          2497
       589       123183           584         83289           667         78551          2241
       600       122696           680         83964           576         73340          2336

Final score for writes:           600
Final score for reads :        178372




More information about the Kernel mailing list