ahd(4) poor performance

[Volodymyr Kostyrko]

Justin T. Gibbs wrote:
> Access the raw device to bypass the filesystem and VM subsystem.
> Start N dds where N is the queue depth you've configured for
> the device.

   I've made almost the same test on /dev/da0s1b (the swap partition).

   Script:

dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
dd if=/dev/zero of=/dev/da0s1b count=20480&
wait

 > camcontrol tags 0:0 -v
(pass0:ahd0:0:0:0): dev_openings  8
(pass0:ahd0:0:0:0): dev_active    8
(pass0:ahd0:0:0:0): devq_openings 8
(pass0:ahd0:0:0:0): devq_queued   0
(pass0:ahd0:0:0:0): held          0
(pass0:ahd0:0:0:0): mintags       2
(pass0:ahd0:0:0:0): maxtags       255


     2 users    Load  0.31  0.49  0.49                  23 чер 21:13

Mem:KB    REAL            VIRTUAL                     VN PAGER  SWAP PAGER
         Tot   Share      Tot    Share    Free         in  out     in  out
Act  106680   13672   316112    27816 1624096 count
All  424236   19976 73870680    41352         pages
                                                                  Interrupts
Proc:r  p  d  s  w    Csw  Trp  Sys  Int  Sof  Flt        cow    8190 total
            9 80      1541    9  553  555    2      144940 wire 
4: sio0
                                                     83608 act 
14: ata
  0.0%Sys   0.0%Intr  0.0%User  0.0%Nice 99.9%Idl   197908 inact     3 
16: uhc
|    |    |    |    |    |    |    |    |    |         12 cache 
19: uhc
                                                   1624084 free 
20: em1
                                                           daefr 
23: ehc
Namei         Name-cache    Dir-cache                     prcfr     3 
32: em0
     Calls     hits    %     hits    %                     react   196 
66: ahd
         4        4  100                                   pdwak 
67: ahd
                                           zfod            pdpgs  1997 
cpu0: time
Disks   da0   da1 pass0 pass1             ozfod           intrn  1997 
cpu2: time
KB/t   0.50  0.00  0.00  0.00             %slo-z   114464 buf    1997 
cpu1: time
tps     195     0     0     0             tfree        51 dirty  1997 
cpu3: time
MB/s   0.10  0.00  0.00  0.00                      100000 desiredvnodes
% busy  100     0     0     0                        7750 numvnodes
                                                      6193 freevnodes
   Script output is:

20480+0 records in
20480+0 records out
10485760 bytes transferred in 363.641053 secs (28835 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 483.640098 secs (21681 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 483.650721 secs (21680 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 604.895852 secs (17335 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 618.500802 secs (16954 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 618.539979 secs (16952 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 618.545158 secs (16952 bytes/sec)
20480+0 records in
20480+0 records out
10485760 bytes transferred in 618.553915 secs (16952 bytes/sec)

   Hmm... My ATA drive does the same almost three times faster...

-- 
[WBR], Arcade.