kuroの覚え書き

96の個人的覚え書き

8xのRAIDカードを1xに刺すとダメダメになるのか検証

Haswellなxeonの中古サーバを試験用に保持しているのだが、PCIeスロットが1x/1x/16x/4xと結構しょぼい。
4xにN8103-150というRAIDカードが刺さっており、こちらから2.5inch 450GB SAS HDDが3枚RAID5構成で接続されている。
しかしこのカードがここにいるとせっかくのPCIe3.0 16xスロットに2スロットなグラボがさせない。
かといってRAIDカードをやめてオンボードからSATAディスクに接続するとなるとこの4xスロットの並びにSATA用のminiSASコネクタが垂直に生えていて、結局大きめなグラボはぶつかってしまうため、搭載できない。
ということでRAIDカードを1xスロットに刺したら使い物にならないのか?をテストしてみることにした。

OSはCentOS7で、CPUはxeon1320v3、メモリは8GB搭載している。

まずは定位置の4xスロットで起動。
hdparmで読み出し、ddで書き込みをテストするという簡易測定を実施。

[kuro@E5800-T110f-E ~]$ sudo hdparm -t /dev/sda1
[sudo] kuro のパスワード:

/dev/sda1:
 Timing buffered disk reads: 598 MB in  3.01 seconds = 198.72 MB/sec

 Timing buffered disk reads: 618 MB in  3.00 seconds = 205.97 MB/sec

 Timing buffered disk reads: 622 MB in  3.01 seconds = 206.76 MB/sec

 Timing buffered disk reads: 612 MB in  3.00 seconds = 203.91 MB/sec

 Timing buffered disk reads: 554 MB in  3.01 seconds = 184.29 MB/sec

[kuro@E5800-T110f-E ~]$ sudo time dd if=/dev/zero of=/tmp/hdparm_write.tmp ibs=1M obs=1M count=1024
1024+0 レコード入力
1024+0 レコード出力
1073741824 バイト (1.1 GB) コピーされました、 0.363617 秒、 3.0 GB/秒
0.07user 0.29system 0:00.39elapsed 91%CPU (0avgtext+0avgdata 2972maxresident)k
152inputs+2097152outputs (1major+789minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、  0.360279 秒、 3.0 GB/秒
0.07user 0.42system 0:00.49elapsed 99%CPU (0avgtext+0avgdata 2968maxresident)k
0inputs+2097152outputs (0major+789minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、  0.360702 秒、 3.0 GB/秒
0.07user 0.43system 0:00.51elapsed 99%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+791minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、  0.364864 秒、 2.9 GB/秒
0.06user 0.41system 0:00.48elapsed 99%CPU (0avgtext+0avgdata 2968maxresident)k
0inputs+2097152outputs (0major+789minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、  1.07666秒、 997 MB/秒
0.05user 0.44system 0:01.36elapsed 36%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+791minor)pagefaults 0swaps

1073741824  バイト (1.1 GB) コピーされました、  0.365009 秒、 2.9 GB/秒
0.07user 0.41system 0:00.48elapsed 99%CPU (0avgtext+0avgdata 2976maxresident)k
0inputs+2097152outputs (0major+792minor)pagefaults 0swaps

これくらい
一方、1xスロットに移動して起動すると、

[kuro@E5800-T110f-E ~]$ sudo hdparm -t /dev/sda1
[sudo] kuro のパスワード:

/dev/sda1:
 Timing buffered disk reads: 620 MB in  3.00 seconds = 206.60 MB/sec

 Timing buffered disk reads: 622 MB in  3.00 seconds = 207.02 MB/sec

 Timing buffered disk reads: 620 MB in  3.00 seconds = 206.58 MB/sec

 Timing buffered disk reads: 634 MB in  3.00 seconds = 211.00 MB/sec

 Timing buffered disk reads: 620 MB in  3.00 seconds = 206.53 MB/sec

[kkuro@E5800-T110f-E ~]$ sudo hdparm -t /dev/sda1

/dev/sda1:
 Timing buffered disk reads: 632 MB in  3.01 seconds = 210.27 MB/sec

[kuro@E5800-T110f-E ~]$ sudo time dd if=/dev/zero of=/tmp/hdparm_write.tmp ibs=1M obs=1M count=1024
1024+0 レコード入力
1024+0 レコード出力
1073741824 バイト (1.1 GB) コピーされました、 0.458587 秒、 2.3 GB/秒
0.08user 0.38system 0:00.52elapsed 87%CPU (0avgtext+0avgdata 2976maxresident)k
152inputs+2097152outputs (1major+791minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、 0.352292 秒、 3.0 GB/秒
0.06user 0.44system 0:00.51elapsed 99%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+790minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、 0.352826 秒、 3.0 GB/秒
0.05user 0.43system 0:00.49elapsed 99%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+791minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、 0.353325 秒、 3.0 GB/秒
0.06user 0.42system 0:00.48elapsed 100%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+791minor)pagefaults 0swaps

1073741824 バイト (1.1 GB) コピーされました、 0.353314 秒、 3.0 GB/秒
0.06user 0.42system 0:00.49elapsed 99%CPU (0avgtext+0avgdata 2972maxresident)k
0inputs+2097152outputs (0major+791minor)pagefaults 0swaps

こんな感じで、1GBの単独ファイルの読み書きでは全然差がないことが解る。ランダムアクセスしたり大量のファイルを書き込んだりすると差が出るのかいなあ。

そもそもカードスペックは
PCI-Express Specification Revision 2.0
x8 PCI Express bus operating at 5 Gb/s or 2.5 Gb/s serial transfer rate
となっているのだが、1xスロットでもそれくらいは出るってことかね?
そもそもすでに4xに挿している時点で本来の規格よりも低いのだが。

もうちょっとちゃんとしたベンチマークソフトで比較したほうがいいかもな。

fioというのを使ってみる。
fioを使ってストレージの性能を計測してみた - Qiita
こちらを参考に。
ソフト自体はすでにインストール済みだ。

[kuro@E5800-T110f-E ~]$ fio -filename=/tmp/test2g -direct=1 -rw=write -bs=4k -size=2G -numjobs=64 -runtime=10 -group_reporting -name=file1
file1: (g=0): rw=write, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
...
fio-3.7
Starting 64 processes
file1: Laying out IO file (1 file / 2048MiB)
Jobs: 64 (f=64): [W(64)][30.0%][r=0KiB/s,w=369MiB/s][r=0,w=94.6k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][40.0%][r=0KiB/s,w=369MiB/s][r=0,w=94.4k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][50.0%][r=0KiB/s,w=370MiB/s][r=0,w=94.6k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][60.0%][r=0KiB/s,w=369MiB/s][r=0,w=94.6k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][70.0%][r=0KiB/s,w=370MiB/s][r=0,w=94.6k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][80.0%][r=0KiB/s,w=369MiB/s][r=0,w=94.6k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][90.0%][r=0KiB/s,w=366MiB/s][r=0,w=93.7k IOPS][eta 00m:0Jobs: 64 (f=64): [W(64)][100.0%][r=0KiB/s,w=370MiB/s][r=0,w=94.7k IOPS][eta 00m:00s]
file1: (groupid=0, jobs=64): err= 0: pid=2151: Mon Aug 16 19:59:39 2021
  write: IOPS=94.3k, BW=368MiB/s (386MB/s)(3684MiB/10001msec)
    clat (usec): min=61, max=11002, avg=676.60, stdev=93.70
     lat (usec): min=61, max=11002, avg=676.86, stdev=93.70
    clat percentiles (usec):
     |  1.00th=[  660],  5.00th=[  668], 10.00th=[  668], 20.00th=[  668],
     | 30.00th=[  676], 40.00th=[  676], 50.00th=[  676], 60.00th=[  676],
     | 70.00th=[  676], 80.00th=[  676], 90.00th=[  685], 95.00th=[  685],
     | 99.00th=[  693], 99.50th=[  693], 99.90th=[ 1221], 99.95th=[ 1467],
     | 99.99th=[ 2606]
   bw (  KiB/s): min= 5632, max= 5960, per=1.56%, avg=5891.27, stdev=55.58, samples=1230
   iops        : min= 1408, max= 1490, avg=1472.82, stdev=13.89, samples=1230
  lat (usec)   : 100=0.01%, 250=0.01%, 500=0.05%, 750=99.62%, 1000=0.12%
  lat (msec)   : 2=0.18%, 4=0.01%, 10=0.01%, 20=0.01%
  cpu          : usr=0.49%, sys=2.53%, ctx=978109, majf=0, minf=2200
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=0,942994,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
  WRITE: bw=368MiB/s (386MB/s), 368MiB/s-368MiB/s (386MB/s-386MB/s), io=3684MiB (3863MB), run=10001-10001msec

Disk stats (read/write):
    dm-0: ios=0/930312, merge=0/0, ticks=0/613785, in_queue=617670, util=99.15%, aggrios=0/942996, aggrmerge=0/0, aggrticks=0/622132, aggrin_queue=622984, aggrutil=98.67%
  sda: ios=0/942996, merge=0/0, ticks=0/622132, in_queue=622984, util=98.67%

[kuro@E5800-T110f-E ~]$ fio -filename=/tmp/test2g -direct=1 -rw=read -bs=4k -size=2G -numjobs=64 -runtime=10 -group_reporting -name=file1
file1: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
...
fio-3.7
Starting 64 processes
Jobs: 64 (f=64): [R(64)][100.0%][r=370MiB/s,w=0KiB/s][r=94.8k,w=0 IOPS][eta 00m:00s]
file1: (groupid=0, jobs=64): err= 0: pid=2326: Mon Aug 16 20:08:43 2021
   read: IOPS=101k, BW=393MiB/s (412MB/s)(3932MiB/10013msec)
    clat (nsec): min=964, max=24291k, avg=634621.44, stdev=231967.90
     lat (nsec): min=984, max=24291k, avg=634806.78, stdev=231996.96
    clat percentiles (nsec):
     |  1.00th=[   1080],  5.00th=[   1160], 10.00th=[ 651264],
     | 20.00th=[ 659456], 30.00th=[ 667648], 40.00th=[ 667648],
     | 50.00th=[ 675840], 60.00th=[ 675840], 70.00th=[ 684032],
     | 80.00th=[ 684032], 90.00th=[ 692224], 95.00th=[ 692224],
     | 99.00th=[ 700416], 99.50th=[ 708608], 99.90th=[1253376],
     | 99.95th=[1351680], 99.99th=[7372800]
   bw (  KiB/s): min= 5592, max=28552, per=1.55%, avg=6244.59, stdev=2411.70, samples=1275
   iops        : min= 1398, max= 7138, avg=1561.15, stdev=602.93, samples=1275
  lat (nsec)   : 1000=0.03%
  lat (usec)   : 2=6.21%, 4=0.01%, 10=0.01%, 20=0.01%, 50=0.01%
  lat (usec)   : 100=0.01%, 250=0.02%, 500=0.07%, 750=93.34%, 1000=0.15%
  lat (msec)   : 2=0.14%, 4=0.01%, 10=0.01%, 20=0.01%, 50=0.01%
  cpu          : usr=0.45%, sys=2.37%, ctx=944329, majf=0, minf=2418
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=1006564,0,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
   READ: bw=393MiB/s (412MB/s), 393MiB/s-393MiB/s (412MB/s-412MB/s), io=3932MiB (4123MB), run=10013-10013msec

Disk stats (read/write):
    dm-0: ios=932627/2, merge=0/0, ticks=617264/0, in_queue=619520, util=99.19%, aggrios=943587/2, aggrmerge=0/0, aggrticks=624350/0, aggrin_queue=624896, aggrutil=98.60%
  sda: ios=943587/2, merge=0/0, ticks=624350/0, in_queue=624896, util=98.60%

[kuro@E5800-T110f-E ~]$ fio -filename=/tmp/test2g -direct=1 -rw=read -bs=4k -size=2G -numjobs=64 -runtime=10 -group_reporting -name=file1
file1: (g=0): rw=read, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
...
fio-3.7
Starting 64 processes
Jobs: 64 (f=64): [R(64)][100.0%][r=370MiB/s,w=0KiB/s][r=94.8k,w=0 IOPS][eta 00m:00s]
file1: (groupid=0, jobs=64): err= 0: pid=2326: Mon Aug 16 20:08:43 2021
   read: IOPS=101k, BW=393MiB/s (412MB/s)(3932MiB/10013msec)
    clat (nsec): min=964, max=24291k, avg=634621.44, stdev=231967.90
     lat (nsec): min=984, max=24291k, avg=634806.78, stdev=231996.96
    clat percentiles (nsec):
     |  1.00th=[   1080],  5.00th=[   1160], 10.00th=[ 651264],
     | 20.00th=[ 659456], 30.00th=[ 667648], 40.00th=[ 667648],
     | 50.00th=[ 675840], 60.00th=[ 675840], 70.00th=[ 684032],
     | 80.00th=[ 684032], 90.00th=[ 692224], 95.00th=[ 692224],
     | 99.00th=[ 700416], 99.50th=[ 708608], 99.90th=[1253376],
     | 99.95th=[1351680], 99.99th=[7372800]
   bw (  KiB/s): min= 5592, max=28552, per=1.55%, avg=6244.59, stdev=2411.70, samples=1275
   iops        : min= 1398, max= 7138, avg=1561.15, stdev=602.93, samples=1275
  lat (nsec)   : 1000=0.03%
  lat (usec)   : 2=6.21%, 4=0.01%, 10=0.01%, 20=0.01%, 50=0.01%
  lat (usec)   : 100=0.01%, 250=0.02%, 500=0.07%, 750=93.34%, 1000=0.15%
  lat (msec)   : 2=0.14%, 4=0.01%, 10=0.01%, 20=0.01%, 50=0.01%
  cpu          : usr=0.45%, sys=2.37%, ctx=944329, majf=0, minf=2418
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=1006564,0,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
   READ: bw=393MiB/s (412MB/s), 393MiB/s-393MiB/s (412MB/s-412MB/s), io=3932MiB (4123MB), run=10013-10013msec

Disk stats (read/write):
    dm-0: ios=932627/2, merge=0/0, ticks=617264/0, in_queue=619520, util=99.19%, aggrios=943587/2, aggrmerge=0/0, aggrticks=624350/0, aggrin_queue=624896, aggrutil=98.60%
  sda: ios=943587/2, merge=0/0, ticks=624350/0, in_queue=624896, util=98.60%

[kuro@E5800-T110f-E ~]$ fio -filename=/tmp/test2g -direct=1 -rw=randwrite -bs=4k -size=2G -numjobs=64 -runtime=10 -group_reporting -name=file1
file1: (g=0): rw=randwrite, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
...
fio-3.7
Starting 64 processes
Jobs: 64 (f=64): [w(64)][100.0%][r=0KiB/s,w=1425KiB/s][r=0,w=356 IOPS][eta 00m:00s]
file1: (groupid=0, jobs=64): err= 0: pid=2413: Mon Aug 16 20:09:57 2021
  write: IOPS=803, BW=3213KiB/s (3290kB/s)(31.9MiB/10175msec)
    clat (usec): min=112, max=1163.6k, avg=79013.12, stdev=90407.13
     lat (usec): min=113, max=1163.6k, avg=79013.47, stdev=90407.06
    clat percentiles (usec):
     |  1.00th=[    519],  5.00th=[    865], 10.00th=[   1037],
     | 20.00th=[   1319], 30.00th=[   1598], 40.00th=[   2008],
     | 50.00th=[   2999], 60.00th=[ 149947], 70.00th=[ 166724],
     | 80.00th=[ 179307], 90.00th=[ 193987], 95.00th=[ 206570],
     | 99.00th=[ 227541], 99.50th=[ 240124], 99.90th=[ 291505],
     | 99.95th=[ 549454], 99.99th=[1166017]
   bw (  KiB/s): min=    8, max=  672, per=1.58%, avg=50.71, stdev=124.96, samples=1278
   iops        : min=    2, max=  168, avg=12.64, stdev=31.25, samples=1278
  lat (usec)   : 250=0.15%, 500=0.65%, 750=2.29%, 1000=5.66%
  lat (msec)   : 2=30.98%, 4=14.11%, 10=2.23%, 20=0.06%, 50=0.23%
  lat (msec)   : 100=0.34%, 250=43.09%, 500=0.15%, 750=0.04%, 1000=0.01%
  cpu          : usr=0.00%, sys=0.07%, ctx=15864, majf=0, minf=2288
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=0,8173,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
  WRITE: bw=3213KiB/s (3290kB/s), 3213KiB/s-3213KiB/s (3290kB/s-3290kB/s), io=31.9MiB (33.5MB), run=10175-10175msec

Disk stats (read/write):
    dm-0: ios=0/8173, merge=0/0, ticks=0/634467, in_queue=638149, util=98.41%, aggrios=0/8173, aggrmerge=0/0, aggrticks=0/638522, aggrin_queue=638522, aggrutil=98.23%
  sda: ios=0/8173, merge=0/0, ticks=0/638522, in_queue=638522, util=98.23%

[kuro@E5800-T110f-E ~]$ fio -filename=/tmp/test2g -direct=1 -rw=randread -bs=4k -size=2G -numjobs=64 -runtime=10 -group_reporting -name=file1
file1: (g=0): rw=randread, bs=(R) 4096B-4096B, (W) 4096B-4096B, (T) 4096B-4096B, ioengine=psync, iodepth=1
...
fio-3.7
Starting 64 processes
Jobs: 64 (f=64): [r(64)][100.0%][r=128MiB/s,w=0KiB/s][r=32.8k,w=0 IOPS][eta 00m:00s]
file1: (groupid=0, jobs=64): err= 0: pid=2490: Mon Aug 16 20:10:22 2021
   read: IOPS=10.8k, BW=42.0MiB/s (44.1MB/s)(425MiB/10113msec)
    clat (nsec): min=1070, max=1343.7M, avg=5903426.77, stdev=34468847.98
     lat (nsec): min=1094, max=1343.7M, avg=5903467.66, stdev=34468861.48
    clat percentiles (nsec):
     |  1.00th=[     1112],  5.00th=[     1112], 10.00th=[     1128],
     | 20.00th=[     1144], 30.00th=[     1144], 40.00th=[     1160],
     | 50.00th=[     1160], 60.00th=[     1176], 70.00th=[     1192],
     | 80.00th=[     1224], 90.00th=[    57600], 95.00th=[ 21626880],
     | 99.00th=[166723584], 99.50th=[233832448], 99.90th=[438304768],
     | 99.95th=[549453824], 99.99th=[859832320]
   bw (  KiB/s): min=    7, max= 6496, per=1.53%, avg=659.90, stdev=1078.10, samples=1235
   iops        : min=    1, max= 1624, avg=164.94, stdev=269.53, samples=1235
  lat (usec)   : 2=85.50%, 4=2.85%, 10=0.15%, 20=0.01%, 50=0.34%
  lat (usec)   : 100=2.70%, 250=0.13%, 500=0.05%, 750=0.03%, 1000=0.03%
  lat (msec)   : 2=0.13%, 4=0.43%, 10=1.30%, 20=1.24%, 50=1.75%
  lat (msec)   : 100=1.43%, 250=1.51%, 500=0.36%, 750=0.06%, 1000=0.01%
  cpu          : usr=0.01%, sys=0.04%, ctx=12651, majf=0, minf=2533
  IO depths    : 1=100.0%, 2=0.0%, 4=0.0%, 8=0.0%, 16=0.0%, 32=0.0%, >=64=0.0%
     submit    : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     complete  : 0=0.0%, 4=100.0%, 8=0.0%, 16=0.0%, 32=0.0%, 64=0.0%, >=64=0.0%
     issued rwts: total=108831,0,0,0 short=0,0,0,0 dropped=0,0,0,0
     latency   : target=0, window=0, percentile=100.00%, depth=1

Run status group 0 (all jobs):
   READ: bw=42.0MiB/s (44.1MB/s), 42.0MiB/s-42.0MiB/s (44.1MB/s-44.1MB/s), io=425MiB (446MB), run=10113-10113msec

Disk stats (read/write):
    dm-0: ios=12281/0, merge=0/0, ticks=625279/0, in_queue=631423, util=98.91%, aggrios=12509/0, aggrmerge=0/0, aggrticks=642278/0, aggrin_queue=642274, aggrutil=98.76%
  sda: ios=12509/0, merge=0/0, ticks=642278/0, in_queue=642274, util=98.76%


1x スロットにさした場合。
2GB / block size 4k
シーケンシャルライト
write: IOPS=94.3k, BW=368MiB/s (386MB/s)(3684MiB/10001msec)レイテンシ=750usec
シーケンシャルリード
read: IOPS=101k, BW=393MiB/s (412MB/s)(3932MiB/10013msec)レイテンシ=750usec
ランダムライト
write: IOPS=803, BW=3213KiB/s (3290kB/s)(31.9MiB/10175msec)レイテンシ=250msec
ランダムリード
read: IOPS=10.8k, BW=42.0MiB/s (44.1MB/s)(425MiB/10113msec)レイテンシ=2usec

4x スロットにさした場合。
シーケンシャルライト
write: IOPS=146k, BW=569MiB/s (596MB/s)(5688MiB/10001msec)レイテンシ=500usec
シーケンシャルリード
read: IOPS=95.5k, BW=373MiB/s (391MB/s)(3732MiB/10001msec)レイテンシ=750usec
ランダムライト
write: IOPS=782, BW=3132KiB/s (3207kB/s)(31.2MiB/10213msec)レイテンシ=250msec
ランダムリード
read: IOPS=8749, BW=34.2MiB/s (35.8MB/s)(347MiB/10155msec)レイテンシ=2usec

やっぱり1xと4xでそんなに大きくは変わらない気がする。こりゃ1xに挿しておけばOKだな。