Difference between revisions of "drive speed tests"

From Noah.org
Jump to navigationJump to search
Line 10: Line 10:
 
!write speed in MB/s !! read speed in MB/s !! drive name !! model number !! USB ID !! Notes
 
!write speed in MB/s !! read speed in MB/s !! drive name !! model number !! USB ID !! Notes
 
|-
 
|-
| 4.34 || 22.63 || SanDisk Cruzer Fit 16GB || 0781:5571 || This USB flash drive is a little nub hardly bigger than the USB plug.
+
| 4.34 || 22.63 || SanDisk Cruzer Fit 16GB ||  || 0781:5571 || This USB flash drive is a little nub hardly bigger than the USB plug.
 
|-
 
|-
 
| 6.36 || 19.88 || Pie Digital, Inc. PieKey 4GB ||  || 22a6:ffff || This was from a Flash drive I worked on -- I set the PID! Pretty decent specs.
 
| 6.36 || 19.88 || Pie Digital, Inc. PieKey 4GB ||  || 22a6:ffff || This was from a Flash drive I worked on -- I set the PID! Pretty decent specs.

Revision as of 16:59, 31 July 2012


drive speed test results -- trivial benchmarks

Raw read and write speed may not the best benchmark, but it's a start. For USB Flash drives this is usually the benchmark that I care about the most since I usually use USB Flash drives for moving large files around.

write speed in MB/s read speed in MB/s drive name model number USB ID Notes
4.34 22.63 SanDisk Cruzer Fit 16GB 0781:5571 This USB flash drive is a little nub hardly bigger than the USB plug.
6.36 19.88 Pie Digital, Inc. PieKey 4GB 22a6:ffff This was from a Flash drive I worked on -- I set the PID! Pretty decent specs.
4.01 18.10 SanDisk Cruzer 16GB SDCZ36-016G 0781:5530
3.51 23.76 Transcend Information OCZ Diesel 4GB 1307:0165
10.27 17.50 OEM 2GB 13fe:1f00 Impressive write speed for a cheap drive.
6.64 15.58 Cisco Linksys Connect Easy Setup Key 1GB NK2 13b1:002e
44.83 78.78 Seagate Momentus 7200.3 ST9160411ASG Serial Number: 5TG0K2CB Firmware Revision: DE17
28.14 120.80 Transcend 2.5" SATA-2 SSD TS64GSSD25S-M Serial Number: 002538030036 Firmware Revision: V090216
70.98 125.97 OCZ VERTEX-LE Serial Number: f0409001b, Firmware Revision: 1.05
73.17 166.16 apocalypse.dreamhost.com Run on a shared hosting service. Drive specs unknown. Quite fast for a shared host.
115.82 534.99 Amazon EC2 184.72.242.0 Amazon EC2 West Coast server. Write speeds vary drastically. Average of 18 runs. Look at that read speed!
131.15 269.65 Amazon EC2 184.72.18.0 Amazon EC2 West Coast server. Write speeds vary drastically. Average of 9 runs.
4.32 18.67 Lexar Media 4GB 05dc:a768
4.01 15.45 Alcor Micro Corp. 2GB 058f:6387 This was from a free Jaguar car keyfob schwag. It was kind of crappy.

Do not mount with sync option

The first test run shows that write performance is over 12 times slower when the drive is mounted with the sync option versus the second test without the sync option. The other options have little effect. These tests were run on a USB flash drive.

first test

/dev/sdb1 on /media/usb0 type vfat (rw,noexec,nodev,sync,noatime,nodiratime)

Running test three times...
test 1: write:    0.34 MB/s, read:    18.67 MB/s
test 2: write:    0.34 MB/s, read:    18.61 MB/s
test 3: write:    0.34 MB/s, read:    18.29 MB/s

second test

/dev/sdb1 on /media/usb0 type vfat (rw)

Running test three times...
test 1: write:    4.32 MB/s, read:    18.51 MB/s
test 2: write:    4.32 MB/s, read:    18.51 MB/s
test 3: write:    4.31 MB/s, read:    18.46 MB/s

trivial benchmark tool

You can use `dd` to measure sustained drive throughput if you do it right. You want to bypass any cache and buffering in the I/O path. This is not hard, but it is uncommon enough that it can be difficult to find documentation on the correct options to use.

It is unclear if this will also bypass RAID controller cache. It almost certainly will not bypass the cache on the drive itself. You would need to activate additional option via `hdparm` if your drive offers them. I considered this a little too invasive and I wasn't sure if the options would be the same for every drive. Plus I figured that the drive controller cache is not really an optional component of the drive, so there is no reason to take it out of the equation; although, it does make it more difficult to interpret read-speed tests.

the most trivial drive benchmark tool

alias test-drive-speed='dd if=/dev/zero of=test_data.bin oflag=dsync conv=fdatasync  bs=8388608 count=16 2>&1 | grep "bytes" | cut -f1,6 -d" " | awk '\''{printf ("write-speed: %7.2f MB/s, ", $1 / $2 / (1024*1024))}'\'' && dd if=test_data.bin iflag=direct conv=fdatasync of=/dev/null bs=8388608 count=16 2>&1 | grep "copied" | cut -f1,6 -d" " | awk '\''{printf ("read-speed:  %7.2f MB/s\n", $1 / $2 / (1024*1024))}'\'''

slightly less trivial benchmark tool

#!/bin/bash
#
# test-drive-speed
#
# DESCRIPTION
#
#     This performs a very simple drive speed test. It measures sustained
#     write and read speed to a 128 MB file.
#
# AUTHOR
#
#     Noah Spurrier <noah@noah.org>
#
# LICENSE
#
#     This license is OSI and FSF approved as GPL-compatible.
#     This license identical to the ISC License and is registered with and
#     approved by the Open Source Initiative. For more information vist:
#         http://opensource.org/licenses/isc-license.txt
#
#     Copyright (c) 2010, Noah Spurrier <noah@noah.org>
#
#     Permission to use, copy, modify, and/or distribute this software for any
#     purpose with or without fee is hereby granted, provided that the above
#     copyright notice and this permission notice appear in all copies.
#
#     THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
#     WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
#     MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
#     ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
#     WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
#     ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
#     OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
#
# VERSION
#
#     Version 1
#

echo "Uptime: $(uptime)"
PWD=$(pwd)
echo "The current directory is ${PWD}."
CURRENT_DEV=$(df -hT $(pwd) \
  | sed -n -e 's/^\(\/dev\/[a-zA-Z]\+[0-9]\+\).*/\1/p')
if [ "${CURRENT_DEV}" = "" ]; then
  echo "Cannot find device associated with this directory."
else
  echo "The current directory is stored on ${CURRENT_DEV}"
  if echo ${CURRENT_DEV} | grep -q /dev/md; then
    echo "This device is part of a sofware RAID."
    echo "The software RAID is configured as follows:"
    echo -en "\t"
    egrep 'md[0-9]+' /proc/mdstat
    CURRENT_DEV=$(egrep 'md[0-9]+' /proc/mdstat \
      |sed -n -e 's/.*[[:space:]]\([a-z]*[0-9]*\)\[.\]$/\1/p')
    CURRENT_DEV="/dev/${CURRENT_DEV}"
    echo -n "The last drive in the array is "
  fi
  hdparm -I ${CURRENT_DEV} | sed -n -e '/^$/d' \
    -e '0,/Standards:/p' | sed -n -e '$!p'
fi
echo "Running test three times..."
for n in 1 2 3; do
  sync
  dd if=/dev/zero of=junk.bin oflag=dsync conv=fdatasync \
    bs=8388608 count=16 2>&1|grep "copied"|cut -f1,6 -d" " \
    |awk '{printf("write: %7.2f MB/s, ",$1/$2/(1024*1024))}'
  sync
  dd if=junk.bin iflag=direct conv=fdatasync of=/dev/null \
    bs=8388608 count=16 2>&1|grep "copied"|cut -f1,6 -d" " \
    |awk '{printf("read:  %7.2f MB/s\n",$1/$2/(1024*1024))}'
  rm junk.bin
done
# vim:set sr et ts=4 sw=4 ft=sh: // See Vim, :help 'modeline'

Online guides

http://www.harddrivebenchmark.net/