|
May 14, 2012, 05:14 PM
|
#1 | |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
 
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
[Benchmark] IOMeter Results Thread
Welcome to the IOMeter Results Thread!
There are quite a few different disk benchmarks out there and each has it's pros & cons. IOMeter is nice in that it lets you define different 'Access Specifications'. An Access Specification determines the size of transfer, the mix of sequential/random and read/write. IOMeter is also unique in that it'll run the same tests on Windows & Linux (sorry, no Mac support as far as I'm aware). I've put together a configuration for IOMeter that will let us run the same tests on all systems. These are also the same tests that most reviewers use. You can finally compare your benchmark numbers directly to Storage Review and other sites. Instructions:
IOMeter will cycle through each disk and run each test recording the results. The results will be for a Queue Depth of 4 as this represents 'heavy' usage for non-server workloads. (IOMeter will let you cycle each test at multiple Queue Depths but for 13 specifications and 6 queue depths we would be running 78 two-minute tests. (2.5 hours is a little long for a benchmark) Notes on Posting: In your post be sure to include your system information the test was run on. Example: <TPU Name> | <CPU @ GHz> | <# of Disks> | <Type of Disks> | <Bus Type> | <Description> | <Last Test MB/s> | <Last Test IOPs> | <Relevant notes> The MB/s and IOPs will come from the CSV file for most of the tests. I have an excel macro that will automatically pull the relevant information for me. But in your post, please include the results for the last test. This will be the "Workstation" test and is generally most indicative of our performance. It'll make the individual posts a bit more meaningful for everyone reading this thread, too. These numbers come from the "Results Display" tab and are the top two numbers ("Total I/Os per Second" = IOPs; "Total MBs per Second (Decimal)" = MB/s) Note on SSDs and Disk Size: If you leave Maximum Disk Size at 0 it'll use all free space on the drive. This results in quite a few write operations and may make later tests show degraded performance. Be sure to enter a value in "Maximum Disk Size". The number entered is the number of sectors to use. (Standard: 512 bytes / sector; Advanced Format: 4096 bytes / sector; Standard Drives: 20971520 sectors = 10GB) The Tests 4 KB Random Read/Write: Read – This workload is used to assess the back-end capabilities of a storage system. Random reads tend not to benefit from cache, so this test places the maximum stress on the disk drives within a storage system. Most of the I/O traffic generated by interactive applications like on-line databases and e-mail systems tend to be composed of random reads. The metric that matters the most for this workload is I/O’s per second.2 MB Sequential Read/Write: Similar to 512 KB but with larger blocks.Database (8K OLTP): An interactive on-line database application. Order entry and reservation systems are example of OLTP applications. Oracle and Microsoft SQL Server are examples of database applications that are used to create OLTP applications. OLTP applications are characterized by a number of users accessing a shared system in parallel. I/O’s are mostly random reads (70%).File Server: This workload is meant to mimic the I/O activity of a file system. Examples include a network-attached shared home or corporate directory.Web Server: This workload is meant to mimic the I/O activity of a web server, such as Apache.Workstation: This test most closely represents 'Normal Everyday Use'. It's mostly reads (67%) and mostly random (67%) using an 8k block.Results! 4 KB Random Read:
4 KB Random Write:
2 MB Sequential Read:
2 MB Sequential Write:
Database (8K OLTP):
File Server:
Web Server:
Workstation:
Last edited by theeldest; May 22, 2012 at 02:33 PM. |
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
| The Following User Says Thank You to theeldest For This Useful Post: |
|
May 14, 2012, 06:09 PM
|
#3 |
|
Better Than You
Join Date: Sep 2006
Location: Toledo,Ohio
Posts: 5,449 (2.24/day)
Thanks: 729
Thanked 1,572 Times in 994 Posts
|
|
|
|
|
|
May 14, 2012, 07:41 PM
|
#4 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
I forgot to mention that part. I've added a bit in the instructions. Can you please list your system info that the test was run on? <TPU Name> | <CPU @ GHz> | <# of Disks> | <Type of Disks> | <Bus Type> | <Description> | CSV | CSV | <Relevant notes> Thanks! Last edited by theeldest; May 14, 2012 at 08:18 PM. |
|
|
|
|
|
May 14, 2012, 08:06 PM
|
#5 |
|
Better Than You
Join Date: Sep 2006
Location: Toledo,Ohio
Posts: 5,449 (2.24/day)
Thanks: 729
Thanked 1,572 Times in 994 Posts
|
AthlonX2 | 2500K @ 3.3GHz | 1 | SSD | Corsair Force GT 180GB | CSV | CSV | SF2282 OS Drive
|
|
|
|
|
May 14, 2012, 10:50 PM
|
#6 |
 
Join Date: Jul 2006
Location: Cowtown, USSofA
Posts: 1,627 (0.65/day)
Thanks: 57
Thanked 275 Times in 216 Posts
|
What does error 1058 mean?
Arctucas | i7 950 @ 4.28 | 4 | SSD | SATA 2 | 4x 64GB ADATA S599 RAID 0 | CSV | CSV | OS Drive
__________________
Whiskey-Tango-Foxtrot. over. Last edited by Arctucas; May 14, 2012 at 11:44 PM. |
|
|
|
|
May 15, 2012, 12:05 AM
|
#7 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
May 15, 2012, 02:58 AM
|
#8 | |
|
Join Date: Jul 2006
Location: Cowtown, USSofA
Posts: 1,627 (0.65/day)
Thanks: 57
Thanked 275 Times in 216 Posts
|
Quote:
 Does the .csv file I uploaded look as if the benchmark completed properly?
__________________
Whiskey-Tango-Foxtrot. over. |
|
|
|
|
|
May 15, 2012, 01:27 PM
|
#9 |
 |
Small flaw in this strategy - your tests assume 512b sectors. Three of the test sets generate errors on 4k sector sized drives.
|
|
|
|
|
May 15, 2012, 01:53 PM
|
#10 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
May 15, 2012, 01:58 PM
|
#11 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
Pretty wicked results too! |
|
|
|
|
|
May 15, 2012, 02:58 PM
|
#12 |
|
|
Since you linked me here in my RAID6 laptop drive thread, I was gonna do your tests and then discovered that it doesn't like 4K sectors. I can make myself and my test rig available for assistance on fine-tuning your 4K needs.
Your first test (1 block) just needs to be set to 4K. As for the other two (web and ... uh... I can't remember, I'm at work and not looking at it) I think you just need to nuke the first three tests (512, 1K, 2K). But that makes it hard to align those results with the ones you already have, so... Hmm. |
|
|
|
|
May 15, 2012, 04:12 PM
|
#13 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
I removed the first test and changed the other two problem tests to align at 4k boundaries instead of sector boundaries. I tried this on the first test but performance was absolutely 100% different. It's not a very good test anyway. With the fileserver and webserver tests the alignment change didn't affect performance on my RAID set which is 512b sectors. If you don't get errors I'll just swap the file and we'll call it good. |
|
|
|
|
|
May 15, 2012, 05:59 PM
|
#14 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
It's somewhat surprising that they maintain such a lead over enterprise class hardware. This is a 12 disk RAID5 array. I'll have another 12 disk array coming shortly. (12 disk fibre channel in RAID 10) Last edited by theeldest; May 15, 2012 at 06:04 PM. |
|
|
|
|
|
May 15, 2012, 06:21 PM
|
#15 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
May 15, 2012, 08:49 PM
|
#16 | |
|
|
Quote:
|
|
|
|
|
|
May 18, 2012, 09:58 AM
|
#17 |
Join Date: Mar 2010
Location: Istanbul
Posts: 32 (0.03/day)
Thanks: 17
Thanked 3 Times in 3 Posts
|
kaktus1907 | Q9650 @ 4.35GHz | 1 | SSD | SATA II | 256GB Corsair Performance Pro | 152.5 MB | 18562 | Marvell 88SS9174 OS Drive
kaktus1907 | Q9650 @ 4.25GHz | 2 | 7.2k SATA | SATA II | 2x500GB WD 5001AALS-00L3B2 RAID0 | 4.32 MB | 528 | non-OS drive, RAID 0 - 128k stripe, Short Stroked 350GB kaktus1907 | Q9650 @ 4.25GHz | 1 | 5.4-7.2k SATA | SATA II | 2TB WD 20EARS-00MVWB0 | 0.74 MB | 90 | non-OS drive
__________________
 Man from Atlantis, MfA, kaktus1907 and zennino SIS 6326 > Ti 4200 > 9800XT > 9800GT > GTX 460 Celeron 366 > Celeron 1700 > Athlon XP 2500+ > E6300 > Q9650 Last edited by kaktus1907; May 18, 2012 at 01:11 PM. |
|
|
|
|
May 19, 2012, 01:12 PM
|
#18 |
|
Join Date: Mar 2010
Location: Istanbul
Posts: 32 (0.03/day)
Thanks: 17
Thanked 3 Times in 3 Posts
|
i think my previous results are wrong.. i did select all workers individually for drives while testing and it causes additional queues therefore total queues were (4 queues per 4 workers)16 and it causes very high results expect 2MB seq tests.. I'm sorry for trouble, here are legit results..
kaktus1907 | Q9650 @ 4.35GHz | 1 | SSD | SATA II | 256GB Corsair Performance Pro | 107.60 MB | 13,772.14 | Marvell 88SS9174 OS Drive kaktus1907 | Q9650 @ 4.25GHz | 2 | 7.2k SATA | SATA II | 2x500GB WD 5001AALS-00L3B2 RAID0 | 2.70 MB | 344.95 | non-OS drive, RAID 0 - 128k stripe, Short Stroked 350GB kaktus1907 | Q9650 @ 4.25GHz | 1 | 5.4-7.2k SATA | SATA II | 2TB WD 20EARS-00MVWB0 | 0.63 MB | 80.86 | non-OS drive
__________________
Man from Atlantis, MfA, kaktus1907 and zennino SIS 6326 > Ti 4200 > 9800XT > 9800GT > GTX 460 Celeron 366 > Celeron 1700 > Athlon XP 2500+ > E6300 > Q9650 |
|
|
|
|
May 21, 2012, 10:55 PM
|
#19 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
May 21, 2012, 10:57 PM
|
#20 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
May 21, 2012, 11:41 PM
|
#21 |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Oooh. And I got some new hardware to play with over the weekend. I'll take a bit longer to play with it but I'm hitting about 4k IOPS on the workstation test with mechanical drives.
|
|
|
|
|
May 22, 2012, 02:37 PM
|
#22 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
|
Jul 5, 2012, 04:19 AM
|
#23 | |
|
Join Date: Feb 2006
Location: Austin, TX
Posts: 611 (0.23/day)
Thanks: 38
Thanked 139 Times in 100 Posts
|
Quote:
|
|
|
|
|
 |
| Currently Active Users Viewing This Thread: 1 (0 members and 1 guests) | |
| Thread Tools | |
|
|
Similar Threads
|
||||
| Thread | Thread Starter | Forum | Replies | Last Post |
| MAXXMEM Benchmark Results Thread | Chicken Patty | Motherboards & Memory | 606 | Oct 6, 2012 08:08 AM |
| Mafia 2 Demo Benchmark Results Thread | EastCoasthandle | Games | 166 | Sep 4, 2010 07:54 AM |
| Lost Planet 2 Benchmark Results Thread | alexsubri | Games | 1 | Aug 19, 2010 11:41 PM |
| StarTales Benchmark Results thread | wolf | NVIDIA | 3 | May 6, 2009 11:48 PM |
| Devil May Cry Benchmark Results Thread | EastCoasthandle | Graphics Cards | 187 | Oct 20, 2008 11:53 AM |
System Specs
