LinuxQuestions.org
Visit Jeremy's Blog.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - General
User Name
Password
Linux - General This Linux forum is for general Linux questions and discussion.
If it is Linux Related and doesn't seem to fit in any other forum then this is the place.

Notices


Reply
  Search this Thread
Old 08-27-2014, 05:24 PM   #16
Firerat
Senior Member
 
Registered: Oct 2008
Distribution: Debian sid
Posts: 2,683

Rep: Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783

Are you copying across mountpoints?

Crude example

Code:
cp /mnt/part1/files /mnt/part2/dest
Are you doing similar in windows?
i.e. C: to D: ?
 
Old 08-27-2014, 06:08 PM   #17
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
What is strange, when I test with something like iops I get quite good results or maybe I'm wrong:
Code:
root@alienware:/partitions/test# ./iops-2011-02-11.py --num_threads 16 --time 2 /dev/sda2
/dev/sda2, 104.82 GB, 32 threads:
 512   B blocks: 42995.3 IO/s,  21.0 MiB/s (176.1 Mbit/s)
   1 KiB blocks: 36062.9 IO/s,  35.2 MiB/s (295.4 Mbit/s)
   2 KiB blocks: 24353.3 IO/s,  47.6 MiB/s (399.0 Mbit/s)
   4 KiB blocks: 16112.1 IO/s,  62.9 MiB/s (528.0 Mbit/s)
   8 KiB blocks: 16141.5 IO/s, 126.1 MiB/s (  1.1 Gbit/s)
  16 KiB blocks: 7793.3 IO/s, 121.8 MiB/s (  1.0 Gbit/s)
  32 KiB blocks: 6054.8 IO/s, 189.2 MiB/s (  1.6 Gbit/s)
  64 KiB blocks: 3112.3 IO/s, 194.5 MiB/s (  1.6 Gbit/s)
 128 KiB blocks: 1830.0 IO/s, 228.7 MiB/s (  1.9 Gbit/s)
 256 KiB blocks: 1261.0 IO/s, 315.2 MiB/s (  2.6 Gbit/s)
 512 KiB blocks:  797.3 IO/s, 398.7 MiB/s (  3.3 Gbit/s)
   1 MiB blocks:  443.5 IO/s, 443.5 MiB/s (  3.7 Gbit/s)
   2 MiB blocks:  249.9 IO/s, 499.7 MiB/s (  4.2 Gbit/s)
   4 MiB blocks:  131.8 IO/s, 527.1 MiB/s (  4.4 Gbit/s)
   8 MiB blocks:   66.0 IO/s, 528.1 MiB/s (  4.4 Gbit/s)
  16 MiB blocks:   33.2 IO/s, 531.2 MiB/s (  4.5 Gbit/s)
  32 MiB blocks:   16.6 IO/s, 532.1 MiB/s (  4.5 Gbit/s)
More benchmarks:
Code:
root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null count=1024k
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.544665 s, 121 MB/s
Code:
root@alienware:/partitions/test# stat -f .
  File: "."
    ID: ad25be0593acbd55 Namelen: 255     Type: reiserfs
Block size: 4096       Fundamental block size: 4096
Blocks: Total: 25591026   Free: 18164439   Available: 18164439
Inodes: Total: 0          Free: 0

Last edited by firedrake; 08-27-2014 at 07:15 PM.
 
Old 08-27-2014, 06:24 PM   #18
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
Yes, I'm copying across mountpoints.
When I'm copying big file like avi the speed is about 30-40 MB/s.
When I'm copying game with small files the speed after couple of seconds drops to 4-2 MB/s.
Copying 1.4GB avi takes about 40s

In Windows 7 while copying across mountpoints (c: -> d speed is around 400 MB/s
Copying 13GB (game with small files) takes about 30-40s

Maybe I should try new kernel like 3.14, 3.15, 3.16 - what do you think ?

My hardware - hwinfo --block:
Code:
root@alienware:/partitions/test# hwinfo --block
27: IDE 00.0: 10600 Disk                                        
  [Created at block.245]
  Unique ID: 3OOL.uqqxB1M5kv0
  Parent ID: w7Y8.daA1xw3T5mA
  SysFS ID: /class/block/sda
  SysFS BusID: 0:0:0:0
  SysFS Device Link: /devices/pci0000:00/0000:00:1f.2/ata1/host0/target0:0:0/0:0:0:0
  Hardware Class: disk
  Model: "Samsung SSD 840"
  Vendor: "Samsung"
  Device: "SSD 840"
  Revision: "EXT0"
  Serial ID: "S1D9NEADA01399E"
  Driver: "ahci", "sd"
  Driver Modules: "ahci"
  Device File: /dev/sda
  Device Files: /dev/sda, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0, /dev/disk/by-id/wwn-0x500253885009c12b
  Device Number: block 8:0-8:15
  BIOS id: 0x80
  Geometry (Logical): CHS 121601/255/63
  Size: 1953525168 sectors a 512 bytes
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #19 (SATA controller)

28: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: bdUI.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda1
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda1
  Device Files: /dev/sda1, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part1, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part1, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part1, /dev/disk/by-uuid/6048097B480950E8, /dev/disk/by-label/boot, /dev/disk/by-id/wwn-0x500253885009c12b-part1
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

29: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: 2pkM.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda2
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda2
  Device Files: /dev/sda2, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part2, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part2, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part2, /dev/disk/by-uuid/9241f109-d873-44e5-a875-37785dfcc57e, /dev/disk/by-id/wwn-0x500253885009c12b-part2, /dev/root
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

30: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: W__Q.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda3
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda3
  Device Files: /dev/sda3, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part3, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part3, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part3, /dev/disk/by-uuid/10B48536B4851F7A, /dev/disk/by-label/OS, /dev/disk/by-id/wwn-0x500253885009c12b-part3
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

31: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: z9FV.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda4
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda4
  Device Files: /dev/sda4, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part4, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part4, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part4, /dev/disk/by-id/wwn-0x500253885009c12b-part4
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

32: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: QLVZ.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda5
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda5
  Device Files: /dev/sda5, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part5, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part5, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part5, /dev/disk/by-uuid/a5f629db-96a6-40e6-a5dd-5b6df8c014fc, /dev/disk/by-id/wwn-0x500253885009c12b-part5
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

33: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: tWld.SE1wIdpsiiC
  Parent ID: 3OOL.uqqxB1M5kv0
  SysFS ID: /class/block/sda/sda6
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sda6
  Device Files: /dev/sda6, /dev/disk/by-id/ata-Samsung_SSD_840_EVO_1TB_S1D9NEADA01399E-part6, /dev/disk/by-id/scsi-SATA_Samsung_SSD_840S1D9NEADA01399E-part6, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part6, /dev/disk/by-uuid/9241f109-d873-44e5-a875-37785dfcc57e, /dev/disk/by-id/wwn-0x500253885009c12b-part6
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #27 (Disk)

34: IDE 100.0: 10600 Disk
  [Created at block.245]
  Unique ID: WZeP.MAM11Dom9L0
  Parent ID: w7Y8.daA1xw3T5mA
  SysFS ID: /class/block/sdb
  SysFS BusID: 1:0:0:0
  SysFS Device Link: /devices/pci0000:00/0000:00:1f.2/ata2/host1/target1:0:0/1:0:0:0
  Hardware Class: disk
  Model: "ST9500423AS"
  Device: "ST9500423AS"
  Revision: "0006"
  Serial ID: "6WR2S1VZ"
  Driver: "ahci", "sd"
  Driver Modules: "ahci"
  Device File: /dev/sdb
  Device Files: /dev/sdb, /dev/disk/by-id/ata-ST9500423AS_6WR2S1VZ, /dev/disk/by-id/scsi-SATA_ST9500423AS_6WR2S1VZ, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0, /dev/disk/by-id/wwn-0x5000c50060c49fd0
  Device Number: block 8:16-8:31
  BIOS id: 0x81
  Geometry (Logical): CHS 60801/255/63
  Size: 976773168 sectors a 512 bytes
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #19 (SATA controller)

35: None 00.0: 11300 Partition
  [Created at block.414]
  Unique ID: h4pj.SE1wIdpsiiC
  Parent ID: WZeP.MAM11Dom9L0
  SysFS ID: /class/block/sdb/sdb1
  Hardware Class: partition
  Model: "Partition"
  Device File: /dev/sdb1
  Device Files: /dev/sdb1, /dev/disk/by-id/ata-ST9500423AS_6WR2S1VZ-part1, /dev/disk/by-id/scsi-SATA_ST9500423AS_6WR2S1VZ-part1, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0-part1, /dev/disk/by-uuid/1c50e827-c467-4b42-8a3f-47ab6c42b452, /dev/disk/by-id/wwn-0x5000c50060c49fd0-part1
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #34 (Disk)

36: SCSI 200.0: 10602 CD-ROM (DVD)
  [Created at block.249]
  Unique ID: KD9E.SK_WnxEsWB4
  Parent ID: w7Y8.daA1xw3T5mA
  SysFS ID: /class/block/sr0
  SysFS BusID: 2:0:0:0
  SysFS Device Link: /devices/pci0000:00/0000:00:1f.2/ata3/host2/target2:0:0/2:0:0:0
  Hardware Class: cdrom
  Model: "HL-DT-ST DVD+-RW GA31N"
  Vendor: "HL-DT-ST"
  Device: "DVD+-RW GA31N"
  Revision: "A201"
  Driver: "ahci", "sr"
  Driver Modules: "ahci"
  Device File: /dev/sr0 (/dev/sg2)
  Device Files: /dev/sr0, /dev/scd0, /dev/disk/by-path/pci-0000:00:1f.2-scsi-0:0:0:0, /dev/cdrom5, /dev/cdrom, /dev/cdr5, /dev/cdr, /dev/cdwriter5, /dev/cdwriter, /dev/cdrw5, /dev/cdrw, /dev/writer, /dev/dvd5, /dev/dvd, /dev/dvdrw5, /dev/dvdrw, /dev/dvdwriter5, /dev/dvdwriter
  Device Number: block 11:0 (char 21:2)
  Features: CD-R, CD-RW, DVD, DVD-R, DVD-RW, DVD-R DL, DVD+R, DVD+RW, DVD+R DL, DVD-RAM, MRW, MRW-W
  Drive status: no medium
  Config Status: cfg=new, avail=yes, need=no, active=unknown
  Attached to: #19 (SATA controller)
  Drive Speed: 24

Last edited by firedrake; 08-27-2014 at 06:27 PM.
 
Old 08-27-2014, 06:34 PM   #19
suicidaleggroll
LQ Guru
 
Registered: Nov 2010
Location: Colorado
Distribution: OpenSUSE, CentOS
Posts: 5,573

Rep: Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143
Quote:
Originally Posted by firedrake View Post
In Windows 7 while copying across mountpoints (c: -> d speed is around 400 MB/s
I'm assuming both of those are on the SSD? If one of those partitions is on the HDD then something is going screwy with Windows.
 
Old 08-27-2014, 07:22 PM   #20
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
More tests which bring me to the new question - how to speed up "cp" command:
Code:
root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.57122 s, 115 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=4096 count=1000000
16065+0 records in
16065+0 records out
65802240 bytes (66 MB) copied, 0.549953 s, 120 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.54542 s, 121 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.545071 s, 121 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.550401 s, 120 MB/s

root@alienware:/partitions/test# dd iflag=direct if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 2.95059 s, 22.3 MB/s

root@alienware:/partitions/test# dd iflag=direct if=/dev/sda1 of=/dev/null bs=1024k count=1000000
62+1 records in
62+1 records out
65802240 bytes (66 MB) copied, 0.53551 s, 123 MB/s

root@alienware:/partitions/test# dd if=linux-3.16.1.tar.xz of=/dev/null
157202+1 records in
157202+1 records out
80487456 bytes (80 MB) copied, 0.144715 s, 556 MB/s

root@alienware:/partitions/test# dd iflag=direct if=linux-3.16.1.tar.xz of=/dev/null
157202+1 records in
157202+1 records out
80487456 bytes (80 MB) copied, 7.35157 s, 10.9 MB/s

root@alienware:/partitions/test# dd iflag=direct if=linux-3.16.1.tar.xz bs=1024k of=/dev/null
76+1 records in
76+1 records out
80487456 bytes (80 MB) copied, 1.14175 s, 70.5 MB/s
 
Old 08-27-2014, 07:24 PM   #21
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by suicidaleggroll View Post
I'm assuming both of those are on the SSD? If one of those partitions is on the HDD then something is going screwy with Windows.
All of the partitions are on the SSD.
I can't believe that copying in linux can be that slow according to Windows.
Now when I copy 1.4GB avi I have 40MB/s accros mountpoints. Copying on the same partition gives me 20MB/s.
I think I should get higher speed.

Last edited by firedrake; 08-27-2014 at 07:29 PM.
 
Old 08-27-2014, 08:24 PM   #22
suicidaleggroll
LQ Guru
 
Registered: Nov 2010
Location: Colorado
Distribution: OpenSUSE, CentOS
Posts: 5,573

Rep: Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143
Quote:
Originally Posted by firedrake
benchmarking with dd gives me speed of around 5 MB/s
Quote:
Originally Posted by firedrake View Post
More tests which bring me to the new question - how to speed up "cp" command:
Code:
root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.57122 s, 115 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=4096 count=1000000
16065+0 records in
16065+0 records out
65802240 bytes (66 MB) copied, 0.549953 s, 120 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.54542 s, 121 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.545071 s, 121 MB/s

root@alienware:/partitions/test# dd if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 0.550401 s, 120 MB/s

root@alienware:/partitions/test# dd iflag=direct if=/dev/sda1 of=/dev/null bs=512 count=1000000
128520+0 records in
128520+0 records out
65802240 bytes (66 MB) copied, 2.95059 s, 22.3 MB/s

root@alienware:/partitions/test# dd iflag=direct if=/dev/sda1 of=/dev/null bs=1024k count=1000000
62+1 records in
62+1 records out
65802240 bytes (66 MB) copied, 0.53551 s, 123 MB/s

root@alienware:/partitions/test# dd if=linux-3.16.1.tar.xz of=/dev/null
157202+1 records in
157202+1 records out
80487456 bytes (80 MB) copied, 0.144715 s, 556 MB/s

root@alienware:/partitions/test# dd iflag=direct if=linux-3.16.1.tar.xz of=/dev/null
157202+1 records in
157202+1 records out
80487456 bytes (80 MB) copied, 7.35157 s, 10.9 MB/s

root@alienware:/partitions/test# dd iflag=direct if=linux-3.16.1.tar.xz bs=1024k of=/dev/null
76+1 records in
76+1 records out
80487456 bytes (80 MB) copied, 1.14175 s, 70.5 MB/s
I'm confused - in your original post you said a dd dump gave you 5 MB/s, now you're showing ~120? What changed?

When I set up a new system I always do some basic benchmarks. Namely:
Code:
dd count=50000 bs=1M if=/dev/zero of=/50gb.bin
dd bs=1M if=/50gb.bin of=/dev/null
hdparm -t /dev/sda
For each drive/mount point on the system. For systems with more than ~25 GB of RAM I increase this to 100+ GB to avoid caching issues.

I always get within 80-90% of the rated speed for all drives in the system. That means 400+ MB/s for modern SSDs and 80+ MB/s for HDDs. RAID systems often give me over 1 GB/s depending on configuration and RAID brand/specs.

Last edited by suicidaleggroll; 08-27-2014 at 08:27 PM.
 
Old 08-28-2014, 02:39 AM   #23
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by suicidaleggroll View Post
I'm confused - in your original post you said a dd dump gave you 5 MB/s, now you're showing ~120? What changed?.....
Yes, I'm confused as well. I would like to have the same speed as you - around 400MB/s.

About the tests - the first one was
Code:
dd if=/dev/zero of=test bs=64k count=2k
That gives me around 5 MB/s

Yesterday I found different benchmark so I tried that one, for example:
Code:
dd if=linux-3.16.1.tar.xz of=/dev/null
This one gave me 556 MB/s - reading one big file and writing to NULL

I think I have problems with writing or buffers (cache), because the last one (/dev/null) discards all data written to it but reports that the write operation succeeded and provides no data to any process that reads from it, yielding EOF immediately.

I will try to do your test and check result when I get back home, but I think it's gona be the same result - slow.

I saw result of doing the same operation twice - copying the same file twice which should be a lot faster second time because it's already in cache.

suicidaleggroll could you do the same tests as mine and check what results you get.

I really want to find the solution or at least the cause!

Last edited by firedrake; 08-28-2014 at 02:44 AM.
 
Old 08-28-2014, 01:28 PM   #24
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by suicidaleggroll View Post
When I set up a new system I always do some basic benchmarks. Namely:
Code:
dd count=50000 bs=1M if=/dev/zero of=/50gb.bin
dd bs=1M if=/50gb.bin of=/dev/null
hdparm -t /dev/sda
I did your tests and I still get up to 5MB/s. I tried to setup different parameters and still nothing.

Is it possible that the kernel and drivers slow down so much ?
How can I test if kernel is slowing down - how can I find where is the problem ?
 
Old 08-28-2014, 01:48 PM   #25
Firerat
Senior Member
 
Registered: Oct 2008
Distribution: Debian sid
Posts: 2,683

Rep: Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783
I don't really know much about SSD drives, not really used them.

I seem to remember reading somewhere that partition 'boundaries' and the file system 'block size' *really* matter

The partition should be aligned 'as such' , and the Filesystem block size needs to match that of the 'device'

I believe this is the reason Windows will refuse to partition removable FlashDrives, since the wrong 'setup' degrades performance.

Wish I could remember where I read it,,
and it may of course not apply to 'modern' SSD drives

Also the reiserfs is something I have not used for years, I went ext2 -> ext3 -> resierfs -> ext4 .. and these days mostly btrfs ( but I do not use 'production systems' )
 
Old 08-28-2014, 02:29 PM   #26
firedrake
Member
 
Registered: May 2013
Posts: 59

Original Poster
Rep: Reputation: Disabled
Quote:
Originally Posted by Firerat View Post
I don't really know much about SSD drives, not really used them.

I seem to remember reading somewhere that partition 'boundaries' and the file system 'block size' *really* matter

The partition should be aligned 'as such' , and the Filesystem block size needs to match that of the 'device'

I believe this is the reason Windows will refuse to partition removable FlashDrives, since the wrong 'setup' degrades performance.

Wish I could remember where I read it,,
and it may of course not apply to 'modern' SSD drives

Also the reiserfs is something I have not used for years, I went ext2 -> ext3 -> resierfs -> ext4 .. and these days mostly btrfs ( but I do not use 'production systems' )
Yes you are right - partitioning is important.
Did you have problems with btrfs ?
What about data consistency ?
Did you loose some data because of power failure ?

I DID MORE TESTS:
I have on a different machine WD Caviar and ReiserFS and I get there 80-120MB/s. There is also different kernel:
Code:
Linux 3.10.17-smp #2 SMP Wed Oct 23 17:13:14 CDT 2013 i686 Intel(R) Core(TM) i5-4430 CPU @ 3.00GHz GenuineIntel GNU/Linux
The problems with slow hard drive started when I moved to the new (that time) kernel 3.12. Now I'm setting 3.16.1 - going line by line, option by option to set it correctly without
Code:
make oldconfig
When I finish I will let you know.

Meanwhile, if someone know the way to test and find the cause of slow read/writes, please let me know.

Last edited by firedrake; 08-28-2014 at 02:32 PM.
 
Old 08-28-2014, 03:55 PM   #27
Firerat
Senior Member
 
Registered: Oct 2008
Distribution: Debian sid
Posts: 2,683

Rep: Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783
Btrfs is fine for me

Btrfs on lvm2 is also fine..

I;ve had issues when over 80% 'full', but I was using sid (debian unstable) at the time.

Note, I have old hardware, 6 x 200gb sataII drives on an ab9pro mainboard.
'raid 5' btrfs .. too many bottlenecks for comparison.

Apparently btrfs is 'optimised' for ssd 'out of the box', but can't confirm how well it works

I have been considering an Evo ssd like yours, but can't really justify it until I get a new system.



Your read seems fine (500+mb/s), your slow write may be related to 'miss alined' partition, and fs block size resulting in two pages having to be juggled instead of one


A very poor explanation follows ..

with ssd, a write involves taking a 'page' copying altered ;page' to a 'free page' and marking the old page as 'free'

If miss aligned, and fs block size differs from the ssd page size, you end up doing too much work.

At least that is how I understood it when reading about ssd, partition, logical volumes and filesystems.
Each level has to be in perfect harmony with the drive's 'geometry'


could be nonsense, I read 'about' it a while back, I may have completely miss-understood it
and I have no experience to 'back it up'
 
Old 08-28-2014, 04:59 PM   #28
suicidaleggroll
LQ Guru
 
Registered: Nov 2010
Location: Colorado
Distribution: OpenSUSE, CentOS
Posts: 5,573

Rep: Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143Reputation: 2143
Well, I've used many SSDs in many systems, probably 30+ at this point, ranging from Fedora to Debian to OpenSUSE to Mint to CentOS, on laptops, desktops, workstations, and servers. I have never paid much attention at all to alignment while partitioning, and I always get within ~10% of the manufacturer's rated speed.

I always use ext2/3/4 on them though, never tried reiserfs or btrfs.

Maybe I've just been insanely lucky and they've all been perfectly aligned, but I doubt it. If poor alignment was able to cause THIS big of a performance hit, I'm pretty sure I would have run into it by now.

Last edited by suicidaleggroll; 08-28-2014 at 05:01 PM.
 
Old 08-28-2014, 05:34 PM   #29
Firerat
Senior Member
 
Registered: Oct 2008
Distribution: Debian sid
Posts: 2,683

Rep: Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783Reputation: 783
Yeah

I just skimed through this earlier

https://wiki.debian.org/SSDOptimization

Seems the alignment is now automagic, at least in debian (wheezy onwards?)
 
Old 08-28-2014, 06:23 PM   #30
TobiSGD
Moderator
 
Registered: Dec 2009
Location: Germany
Distribution: Whatever fits the task best
Posts: 17,148
Blog Entries: 2

Rep: Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886Reputation: 4886
Quote:
Originally Posted by suicidaleggroll View Post
Maybe I've just been insanely lucky and they've all been perfectly aligned, but I doubt it. If poor alignment was able to cause THIS big of a performance hit, I'm pretty sure I would have run into it by now.
You will only see degraded write performance, not read performance. Also, most modern partitioners (including fdisk, parted, GParted, ...) align partitions by default. In fact, I know only of cfdisk that still goes with the old MSDOS style partitioning (starting the first partition on sector 63).

Anyways, the fdisk output of the OP shows that the partitions are aligned.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
Helping diagnosing slow Xubuntu start on SSD communeofloon Linux - General 14 11-14-2013 09:46 PM
Slackware 13.37 64 bit and slow ssd speeds anonymous7777 Slackware 5 04-10-2012 01:14 PM
Wildfire 120GB SSD very slow in intel H67 nike.stars Linux - Hardware 1 08-12-2011 01:18 PM
Slow USB transfer rates (60kb p/s) External HDD --> Internal HDD drfrostbyte Linux - Hardware 4 03-28-2011 02:20 AM
Slow read/write operations on 3592 drive Joejr4u Linux - Hardware 2 03-17-2008 06:25 AM

LinuxQuestions.org > Forums > Linux Forums > Linux - General

All times are GMT -5. The time now is 12:47 AM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Open Source Consulting | Domain Registration