下のIOPS=33.9kはxfsでやると3354と表示されました。何度やっても10倍くらいの差になります。

ファイルシステムでそこまで変わるとは思えないので、フォーマットのやり方に問題があるのかもしれませんがこちらのとおりやってます。
https://qiita.com/shiru/items/c765accff6d3c5f8ea34

ただ、マウント時にxfsだと下のオプションが使えないのでそのままmount /dev/md0とやっていますが、これが問題でしょうか？

XFS
mount /dev/md0 /media/ephemeral0

EXT4
mount -o rw,relatime,errors=continue,user_xattr,acl,barrier=1,data=ordered /dev/md0 /media/ephemeral0

fio -name=random-write -ioengine=libaio -rw=randrw -rwmixread=0 -bs=16k -numjobs=16 -iodepth=16 -size=100m -direct=1 -directory=/media/ephemeral0 -group_reporting
random-write: (g=0): rw=randrw, bs=(R) 16.0KiB-16.0KiB, (W) 16.0KiB-16.0KiB, (T) 16.0KiB-16.0KiB, ioengine=libaio, iodepth=16
...
fio-3.3
Starting 16 processes
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
random-write: Laying out IO file (1 file / 100MiB)
Jobs: 1 (f=1): [w(1),_(15)][100.0%][r=0KiB/s,w=502MiB/s][r=0,w=32.1k IOPS][eta 00m:00s]
random-write: (groupid=0, jobs=16): err= 0: pid=6855: Thu Jan 4 08:26:29 2018
write: IOPS=33.9k, BW=529MiB/s (555MB/s)(1600MiB/3022msec)
slat (usec): min=3, max=27676, avg=253.30, stdev=1452.90
clat (usec): min=89, max=64093, avg=7102.46, stdev=5838.76
lat (usec): min=141, max=64104, avg=7356.88, stdev=6093.65
clat percentiles (usec):
| 1.00th=[ 157], 5.00th=[ 219], 10.00th=[ 318], 20.00th=[ 750],
| 30.00th=[ 3359], 40.00th=[ 5276], 50.00th=[ 6456], 60.00th=[ 7898],
| 70.00th=[ 9372], 80.00th=[11469], 90.00th=[15008], 95.00th=[17433],
| 99.00th=[23200], 99.50th=[27132], 99.90th=[43779], 99.95th=[47973],
| 99.99th=[54789]
bw ( KiB/s): min=22176, max=52544, per=6.54%, avg=35470.31, stdev=4276.22, samples=81
iops : min= 1386, max= 3284, avg=2216.56, stdev=267.30, samples=81
lat (usec) : 100=0.01%, 250=6.73%, 500=9.79%, 750=3.48%, 1000=1.71%
lat (msec) : 2=3.19%, 4=7.53%, 10=41.29%, 20=24.05%, 50=2.18%
lat (msec) : 100=0.04%
cpu : usr=1.68%, sys=5.76%, ctx=20890, majf=0, minf=514
IO depths : 1=0.1%, 2=0.1%, 4=0.1%, 8=0.1%, 16=99.8%, 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=0.1%, 32=0.0%, 64=0.0%, >=64=0.0%
issued rwt: total=0,102400,0, short=0,0,0, dropped=0,0,0
latency : target=0, window=0, percentile=100.00%, depth=16

Run status group 0 (all jobs):
WRITE: bw=529MiB/s (555MB/s), 529MiB/s-529MiB/s (555MB/s-555MB/s), io=1600MiB (1678MB), run=3022-3022msec

Disk stats (read/write):
md0: ios=0/96376, merge=0/0, ticks=0/0, in_queue=0, util=0.00%, aggrios=0/49418, aggrmerge=0/1797, aggrticks=0/229064, aggrin_queue=229245, aggrutil=82.68%
xvdb: ios=0/50290, merge=0/926, ticks=0/307177, in_queue=307422, util=81.91%
xvdc: ios=0/48546, merge=0/2669, ticks=0/150951, in_queue=151069, util=82.68%