-
1. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 10, 2010 9:46 AM (in response to kohns)Exporting AVI is a disk intensive benchmark. Run HDtach (a free utility) on both your OS disk and your RAID array and report the burst transfer rate, the average read transfer rate and CPU utilization.
-
2. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 10, 2010 12:07 PM (in response to Bill Gehrke)Thanks for the feedback.
My system crashed... and getting : unable to find NTLDR ... so I will do those tests after I recovered.
Soon I hope ...
-
3. Re: PPBM4 Benchmark : Discussing my results...
Harm Millaard Jan 10, 2010 1:25 PM (in response to kohns)Very observative that you noticed Graham's, Horatio's and your own AVI scores to be less than expected with your disk setups. You all have good disk setups, and one would expect better results. HDTach, as Bill mentioned, can be indicative of potential bottlenecks that keep you from top-notch performance. In addition it may be wise to inspect Process Explorer results to see what processes are running in the background that prevent better results. But first get your system up-and-running. Good luck.
-
4. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 11, 2010 4:29 AM (in response to Harm Millaard)Harm Millaard wrote:
Very observative that you noticed Graham's, Horatio's and your own AVI scores to be less than expected with your disk setups. You all have good disk setups, and one would expect better results. HDTach, as Bill mentioned, can be indicative of potential bottlenecks that keep you from top-notch performance. In addition it may be wise to inspect Process Explorer results to see what processes are running in the background that prevent better results. But first get your system up-and-running. Good luck.
I am almost there... finished the reinstall of windows/windows-updates/drivers and took a new image. I am now tweaking it again before I start installing Adobe and do those tests again...
It should be a little faster now because I now also replaced my 160GB WD 7200RPM with a 150GB WD Velociraptor 10000RPM. Can't wait to run my first benchmark again...
-
5. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 13, 2010 7:35 AM (in response to kohns)Ok my PC is "ready" for the first test... but it is not looking much better even with the new faster OS disk ... on the contrary ...
old run prior to crash new run Computer Manufacturer ASUS Asus Computer Model P6T SE P6T SE secs Total Benchmark Time 47 47.2 secs AVI Encoding Time 4.5 4.4 secs MPEG Elapsed Time 29.5 29.8 secs Rendering Time 13 13 CPU Manufacturer INTEL Intel CPU Model Core i7 960 45nm 3.20 Core i7 960 45nm 3.20 GHz CPU speed 3.2 3.2 Number of CPU chips 2 2 Total Number of Cores 4 4 GB RAM 12 12 APP Version PPBM4 DV 4.2.1 4.2.1 OSVersion Win 7 Pro 64 bit Win 7 Pro 64 bit OS Disk Interface SATA SATA GB OS Disk Capacity 160 150 OS Disk Speed 7200 10000 Project Disk Interface SATA SATA GB Project Disk Capacity 1000x6 R0 1000x6 R0 Project Disk Speed 7200 7200 Preview Disk Interface Same as Project Disk Same as Project Disk GB Preview Disk Capacity Same as Project Disk Same as Project Disk Preview Disk Speed Same as Project Disk Same as Project Disk Output Disk Interface Same qs Project Disk Same as Project Disk GB Output Disk Capacity Same as Project Disk Same as Project Disk Output Disk Speed Same as Project Disk Same as Project Disk Graphics Board ASUS EAH5770 Asus EAH5770 Comment Windows Pagefile on seperate disk 320GB x 2 Raid0 Windows Pagefile on seperate disk 320GB x 2 Raid0 Project Disk on Areca 1220 Cntrl Project Disk on Areca 1220 Cntrl Internet controller disabled and firewall as well (no virusscanner installed or running)
The processes running (tasklist /V) seconds before I hit "Start Queue" on AME...
HDTach results of my 3 disks few minutes after the benchmark ran
And below is a list of how I configured services to start/disable/...
Next to that I also disabled indexing on all drives and defragment all drives seconds before I ran the benchmark
Startup Type Status Name Automatic Started Workstation Automatic Started Windows Modules Installer Automatic Started Windows Management Instrumentation Automatic Started Windows Firewall Automatic Started Windows Event Log Automatic Started Windows Driver Foundation - User-mode Driver Framework Automatic Started Windows Audio Endpoint Builder Automatic Started Windows Audio Automatic Started User Profile Service Automatic Started TrueVector Internet Monitor Automatic Started TCP/IP NetBIOS Helper Automatic Started Task Scheduler Automatic Started System Event Notification Service Automatic Started Superfetch Automatic Started Shell Hardware Detection Automatic Started Server Automatic Started Security Accounts Manager Automatic Started RPC Endpoint Mapper Automatic Started Remote Procedure Call (RPC) Automatic Started Program Compatibility Assistant Service Automatic Started Print Spooler Automatic Started Power Automatic Started Plug and Play Automatic Started Network Store Interface Service Automatic Started Network Location Awareness Automatic Started IKE and AuthIP IPsec Keying Modules Automatic Started Group Policy Client Automatic Started DNS Client Automatic Started DHCP-client Automatic Started DCOM Server Process Launcher Automatic Started Cryptographic Services Automatic Started COM+ Event System Automatic Started Base Filtering Engine Automatic Started Windows Update Automatic Started Windows Defender Automatic Started Software Protection Automatic Started Security Center Manual Started UPnP Device Host Manual Started SSDP Discovery Manual Started SPP Notification Service Manual Started Network List Service Manual Started Network Connections Manual Started IPsec Policy Agent Manual Started Background Intelligent Transfer Service Manual Started Application Information Automatic Multimedia Class Scheduler Manual WMI Performance Adapter Manual Wired AutoConfig Manual Windows Time Manual Windows Presentation Foundation Font Cache 3.0.0.0 Manual Windows Installer Manual Windows Image Acquisition (WIA) Manual Windows Font Cache Service Manual Windows Event Collector Manual Windows Connect Now - Config Registrar Manual Windows Color System Manual Windows CardSpace Manual Windows Biometric Service Manual WebClient Manual Volume Shadow Copy Manual Virtual Disk Manual Thread Ordering Server Manual Telephony Manual Secure Socket Tunneling Protocol Service Manual Secondary Logon Manual Remote Access Connection Manager Manual Remote Access Auto Connection Manager Manual Quality Windows Audio Video Experience Manual Protected Storage Manual PnP-X IP Bus Enumerator Manual Performance Logs & Alerts Manual Performance Counter DLL Host Manual Microsoft iSCSI Initiator Service Manual Microsoft .NET Framework NGEN v2.0.50727_X86 Manual Microsoft .NET Framework NGEN v2.0.50727_X64 Manual KtmRm for Distributed Transaction Coordinator Manual Interactive Services Detection Manual HomeGroup Provider Manual HomeGroup Listener Manual Extensible Authentication Protocol Manual Distributed Transaction Coordinator Manual Disk Defragmenter Manual Credential Manager Manual COM+ System Application Manual CNG Key Isolation Manual BitLocker Drive Encryption Service Manual Application Management Manual Application Identity Manual Application Experience Manual ActiveX Installer (AxInstSV) Disabled WWAN AutoConfig Disabled WLAN AutoConfig Disabled WinHTTP Web Proxy Auto-Discovery Service Disabled Windows Search Disabled Windows Remote Management (WS-Management) Disabled Windows Media Player Network Sharing Service Disabled Windows Media Center Scheduler Service Disabled Windows Media Center Receiver Service Disabled Windows Error Reporting Service Disabled Windows Backup Disabled TPM Base Services Disabled Themes Disabled Tablet PC Input Service Disabled Storage Service Disabled SNMP Trap Disabled Smart Card Removal Policy Disabled Smart Card Disabled Routing and Remote Access Disabled Remote Registry Disabled Remote Procedure Call (RPC) Locator Disabled Remote Desktop Services UserMode Port Redirector Disabled Remote Desktop Services Disabled Remote Desktop Configuration Disabled Problem Reports and Solutions Control Panel Support Disabled Portable Device Enumerator Service Disabled PNRP Machine Name Publication Service Disabled Peer Networking Identity Manager Disabled Peer Networking Grouping Disabled Peer Name Resolution Protocol Disabled Parental Controls Disabled Offline Files Disabled Network Access Protection Agent Disabled Netlogon Disabled Net.Tcp Port Sharing Service Disabled Microsoft Software Shadow Copy Provider Disabled Media Center Extender Service Disabled Link-Layer Topology Discovery Mapper Disabled IP Helper Disabled Internet Connection Sharing (ICS) Disabled Intel(R) Matrix Storage Event Monitor Disabled Human Interface Device Access Disabled Health Key and Certificate Management Disabled Function Discovery Resource Publication Disabled Function Discovery Provider Host Disabled Fax Disabled Encrypting File System (EFS) Disabled Distributed Link Tracking Client Disabled Diagnostic System Host Disabled Diagnostic Service Host Disabled Diagnostic Policy Service Disabled Desktop Window Manager Session Manager Disabled Computer Browser Disabled Certificate Propagation Disabled BranchCache Disabled Bluetooth Support Service Disabled Block Level Backup Engine Service Disabled Application Layer Gateway Service Disabled AMD External Events Utility Disabled Adaptive Brightness -
6. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 14, 2010 1:30 PM (in response to kohns)While waiting on your thoughts on my above results I wanted to do some more investigation myself by comparing my disk setup and HDTach results with Harm his results.
I found this picture from you Harm in my other thread and wanted to compare them with my results (see above) and think you made some typo errors ...
First surprising thing is that your 2x1TB raid0 has a burst speed 4x higher then your 12x1TB raid30 !?!?
So I think those 2 numbers should be switched.
And if they are switched I also think that a burst speed of 1158 for your 2x1TB raid0 is impossible.
If I compare your numbers with my numbers we get this graphic:
Harm his Disk Specs My Disk Specs C Windows 1x150 WD Velociraptor C Windows 1x150 WD Velociraptor Burst speed avg read Burst spead avg read 244 106 253 104 D Pagefile 2x1TB Samsung Raid0 D Pagefile 2x320GB Raid0 Burst speed avg read Burst speed avg read 1158 177 347 132 E Project 12x1TB Samsung Raid30 E Project 6x1TB WD Black Raid0 Burst speed avg read Burst speed avg read 4272 853 966 690 -
7. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 14, 2010 2:57 PM (in response to kohns)Burst speed is quite dependant on how much cache the controller has, how much cache the disk drives have, the synthetic benchmark used to measure this data and the transfer rate of the controller to the system bus. I do not put much faith in burst rate except if it is lower than "normal" it may mean a bottleneck in that specific system.
-
8. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 14, 2010 3:41 PM (in response to Bill Gehrke)Bill Gehrke wrote:
Burst speed is quite dependant on how much cache the controller has, how much cache the disk drives have, the synthetic benchmark used to measure this data and the transfer rate of the controller to the system bus. I do not put much faith in burst rate except if it is lower than "normal" it may mean a bottleneck in that specific system.
A couple of things about above quote:
- You wrote yourself:
Bill Gehrke wrote:
Exporting AVI is a disk intensive benchmark. Run HDtach (a free utility) on both your OS disk and your RAID array and report the burst transfer rate, the average read transfer rate and CPU utilization.
So you are saying this is a disk issue. So I am (in the first place) comparing the disk setups.
Not that I only want to focus on that off course... on the contrary if out of this topic we find other intersting tweaks that would be great of course.
So let me just for a couple of seconds think about burst speed and what you wrote above... in point 2. and 3. - My Raid controller ARC-1220 has 256MB Cache Memory and supports 8*SATA II. I have inserted the controller into my x8 PCI-E slot (so using the max potential)
Harm his ARC-1680ix-12 controller has a default of 512MB Cache which is twice my amount of cache I agree.
His controller however also only supports PCI-E x8 so no speed gain on that.
So the only difference would be the double amount of cache.
But that is only for his 12 disc Raid30 volume.
This controller only support 12 disks internally so his 2x1TB raid0 must be configured (just like me) on the motherboards raid controller and we have the same motherboard, Asus P6T, so I can still not believe that he would have 1158 Burst speed on his 2x1TB RAID0 disks - This still does not explain why our Export AVI stats are so bad.
If you look at the charts on the http://ppbm4.com/Benchmark.html website there are people with much worse disk setups that have like 3 times better values for the Export AVI results.
I don't think my HDTach results are bad unless you consider Harm did not write a type error and his 2x1TB RAID0 disk burst speed is considered normal.
Thanks for thinking with me
- You wrote yourself:
-
9. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 14, 2010 3:50 PM (in response to kohns)Sorry, but Harm has an ugraded Areca with 2 GB of cache. With newer Areca's you can now even go to 4 GB
I have a second Areca and it has currently 512 MB of RAM. Shortly I will be upgrading to 2 GB and will report the differences.
-
10. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 14, 2010 3:52 PM (in response to Bill Gehrke)Bill Gehrke wrote:
Sorry, but Harm has an ugraded Areca with 2 GB of cache. With newer Areca's you can now even go to 4 GB
That could be but still that would only explain the HDTach results for his 12x1TB Raid30 volume... not for his 2x1TB in raid0.
And that still does not explain why myself and some others have so bad numbers for AVI Export since other users (in the spreadsheet) have 'worse' disk setups and 3 times better AVI Export results... which was the main reason I opened this topic.
And off course Harm is 'leader' in this... it is not my intention to measure myself with him... I only put his stats against mine to see if there was a problem with my disks HDTach results.
-
11. Re: PPBM4 Benchmark : Discussing my results...
Harm Millaard Jan 14, 2010 3:52 PM (in response to kohns)You are both correct and wrong.
Correct in that the raid0 burst results are strange, they appear too high, but that is what HDTach reported.
Wrong in exchanging the burst-rates for my raid0 and my raid30. There was no mix-up and yes, I was as surprised as you that the burst rate was that low on my raid30, but that is what HDTach gave me as results. On a side note, you will be hard pressed to find consistent and comparable test results when you ruin these tests a number of times. There will be meaningful differences between tests. So the message is, do not stare at the results too much but consider them as indicative of your systems performance.
Your average read speed with 6 x raid0 is about the maximum achievable (6 x 115 = 690 MB/s) with nearly empty disks and negligent overhead. That is a fantastic score.
BTW: I had a disk crash on my notebook, which took my attention with reformatting/reinstalling/etc. and that is why I did not respond to your other post. I hope you can understand...
-
12. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 14, 2010 4:12 PM (in response to Harm Millaard)Harm Millaard wrote:
You are both correct and wrong.
Correct in that the raid0 burst results are strange, they appear too high, but that is what HDTach reported.
Wrong in exchanging the burst-rates for my raid0 and my raid30. There was no mix-up and yes, I was as surprised as you that the burst rate was that low on my raid30, but that is what HDTach gave me as results. On a side note, you will be hard pressed to find consistent and comparable test results when you ruin these tests a number of times. There will be meaningful differences between tests. So the message is, do not stare at the results too much but consider them as indicative of your systems performance.
Your average read speed with 6 x raid0 is about the maximum achievable (6 x 115 = 690 MB/s) with nearly empty disks and negligent overhead. That is a fantastic score.
BTW: I had a disk crash on my notebook, which took my attention with reformatting/reinstalling/etc. and that is why I did not respond to your other post. I hope you can understand...
Thanks for clarification...
Strange that you have such strange HDTach results.
Like I wrote earlier there are other persons in the benchmark chart that have worse disk setups than mine. And you just confirmed there is nothing wrong with my disk setup and performance.
So that being out of the way... which other directions can we go from here to see where the bottleneck for these poor AVI Export ratings lies?
As I wrote before:
- This is a fresh install
- All disks are as good as empty and defragmented just before I run the benchmark
- A lot of Windows services are disabled and stopped (see lists above)
- No virusscanner installed and the firewall is turned off during the test
- ...
BTW: don't worry ... I don't mind you only were able to reply today... I am not expecting any reply withing the hour ;-) . To be honest I find the response times on this forum very correct. And I appreciate your and others help very much...
-
13. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 14, 2010 4:46 PM (in response to kohns)Welcome to the world of sometimes weird and sometimes wonderful results. While Harm was fighting his notebook disk problems I had a memory module and a fan fail in my Internet computer while I am experimenting bringing up a third computer (a new --for me i7-920). In the meantime I am trying to find out why I cannot repeat my data and now I am getting slightly slower results from PPBM4 with 4.2 Premiere than I did with 4.1 and in addition fighting eccentric behavior with conditional formatting in Excel and saving to the data to html (for the results page).
One thing that is missing from the PPBM4 information is just exactly how some of these people are getting their RAID--i.e. what controller are they using.
Just to show you how things can change here is a test I ran with 'Windows 7 RC and two different nVidia driver versions.
-
14. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 15, 2010 4:39 AM (in response to Bill Gehrke)That is strange Bill... what has a graphic driver to do with disk performance...
I don't think the graphic driver had any influence... I am more thinking about:
- the big time laps between you taking these 2 stats (more then a day between)
- how much did the content of your disk change: a lot of data changes occured, fragmentation , ...
- what other services ran during that time (there might be a big difference)
- did you take the test right after booting up or did you work for a while
- ... etc....
And if we look at your stats that is a difference of roughly 15% but which could be explained by above points.
That would still not explain why my AVI stats are > 300% slower than comparable or less performant setups.
You are right type and setup of the controller will play a big deal... but also how you formatted your disks ( NTFS cluster size etc. )
So you might want to add that to your chart to have better comparisons.
-
15. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 15, 2010 7:40 AM (in response to kohns)Well those video driver tests were done in the same setting. Absolutely the only thing that changed was the drivers. This was done when some rumors were circulated about new graphics drivers "hogging" the PCIe bus.
Edit: I just took a look at the date and time on HDtach. They were done on sucessive dates (July 1st and 2nd 2009) but since they were done on my editing computer it is only turned on when editing or benchmarking. If I remember correctly this was done with the Quadro FX1700 and before Areca had a Win 7 driver. That accounts for the higher than currently posted AVI encoding times. But it does show that the slower array performance does cause a slower encoding time.
P.S. When benchmarking I defragment generally at least daily if not more often
-
16. Re: PPBM4 Benchmark : Discussing my results...
kohns Jan 18, 2010 4:00 PM (in response to kohns)That does indeed make sense : the graphic card is sitting on the same pci-e bus as your raid controllers is.
But that would still not explain why my AVI stats are > 200%-300% slower than comparable or less performant hardware setups:
* You wrote AVI is mainly disk speed.
* Disk speed could be influenced by the graphic driver so that would/could explain 10-15% difference but ot all the rest.
Let us take for example Ozzi (Ann). She hits a score of 2.8 for AVI compared to my 4.5 with a lesser CPU and only 2 disks in Raid0.
I am pretty sure my disks are 2-3 times faster than hers. So where does that difference come from...
Is there anything I can tune further (running services, other windows tweaks, ... )
-
17. Re: PPBM4 Benchmark : Discussing my results...
Bill Gehrke Jan 18, 2010 6:22 PM (in response to kohns)EDIT!!!!
Sorry I slipped to a wrong line when I looked at your graphics card but try going back to an earlier driver.








