Enter The Matrix: Slice out and get the best part from your hard drives

[yasha]

BTW, I did pull the jumpers and my RAID 5 is indeed set to 64k strip size and 128k for Raid 0. I must say this is definitely the most responsive system I've had.

 

[bing]

Hello,

I have a mysterious problem.

It looks like this fellows graphs.
http://www.ocforums.com/showpost.php?p=4662441&postcount=110

The problem is that I have to enable volume-write-back-cache to significantly improve my overall speed. I made the entire array Raid 0.

I am using an Abit IP35 Pro with Intel Raid. The drives are two Seagate 7200.10 320GB drives.

My question is why do I have to enable the feature while other don't in order to have fast arrays. (I don't mean the people who set up a Matrix array. I know where their speed comes from. ) Is VWBC bad for the system at all?

Great thread you guys. Keep it up.

Wellcome to OcF Knight !

Thanks for the compliment, and you have an interesting question there !

Well, you've seen it your self that obviously WBC is increasing the STR/average transfer rate right ? Except for some trivial increase in CPU utilization, but I've tested it intensively in the past, it doesn't affect significantly on real-life usage nor at benchmarks.

Honestly, until today I still don't know what is the con by enabling WBC, I've traced carefully on the memory usage, and still no proof that it eats up more resources like memory usage.

But, I've been "guessing" that these WBC might have a bad effect on single core CPU, but alas so far there are not many samples reported yet to proof it.

Hi Bing,

I appreciate the welcome. I have certainly enjoyed your discussions here regarding Matrix Raid! The Raid 0 benchmark seems like it's not so stable...ie. large fluctuations from 290MB/s to 400 MB/s. I've looked at other Raid 0 benchmarks in this thread and many appear to be more 'flat'. My Raid 5 does seem ok. Here are the HD Tach results. Again, the Raid 0 is not smooth like others have seen. But if you think it's ok, then I have no argument against that

Oh, sorry, I guess I wasn't having enough sleep when I responded to your 1st post.

Ok, if you mean those jaggies curve, have you disabled those background tasks like anti virus, firewall, running applications if any, disk defragmenter and etc ?

If you see my benchmark results, all those were created with disabled background tasks like my AV, firewall, defragmenter, any running applications, and .. internet, yes, sometimes internet ccould affect the benchmark result especially with that pesky Windows Update enabled.

BTW, I did pull the jumpers and my RAID 5 is indeed set to 64k strip size and 128k for Raid 0. I must say this is definitely the most responsive system I've had.

Ok, this is not the 1st time I heard that "my rig feels so responsive", congratulation and glad you're enjoying it.

I know, most ICHxR Raid basher will say, Yeah right, how come your feeling can be an objective benchmark ? LOL, yes, have to admit that they have a valid point in that argument, nothing against them.

Its just I've been building many rigs for a very long time, these thing really makes significant different, anyone agree on me with this ?

Again, a "very subjective" matter, but don't ignore the statistic once they're become a significant numbers in the chart !

.

 

[Jod]

Most responsive, eh? W/ over 350 STR...that's sick...not right...I'm MAD envious.

What do the jumpers do on SATA drives anyway? Something to do w/ spread spectrum, 150/300, or NCQ? I suppose they'd be different from drive to drive. I'll have to take a look @ WD and Hitachi sites soon...unless any of you know off the top of your head?

Thanks!!

PS. The wacky lines in HD Tach don't mean much. If you really wanna test your stuff, copy some files...or maybe do some extraction tests. Nothing says lovin' like real world pron moving.

 

[fritzman]

bing... I agree entirely with the responsiveness feel. I think smooth better describes it for me though... everything is snappy alright, but it all has such a smooth feel to it.

Jod... the jumpers they were taling about take the Seagate perp drives from Sata150 to Sata300 is all. AFAIK it only relates to those particular drives, as I haven't ever seen mention of them with any others.

 

[KN|GHT]

Is it ok if I make to two Raid 0 array's instead of one Raid 0 and Raid 1 array? I mean is there a performance difference if I do it this way than using the entire two drives for Raid 0. My goal is to have two sections. One for programs and the other for storage.

NVM: Answered on page 20. I have to have tow different Raid styles.

 

[yasha]

If you see my benchmark results, all those were created with disabled background tasks like my AV, firewall, defragmenter, any running applications, and .. internet, yes, sometimes internet ccould affect the benchmark result especially with that pesky Windows Update enabled.


After disabling background tasks, the benches did come out better...



Ok, this is not the 1st time I heard that "my rig feels so responsive", congratulation and glad you're enjoying it.

I know, most ICHxR Raid basher will say, Yeah right, how come your feeling can be an objective benchmark ? LOL, yes, have to admit that they have a valid point in that argument, nothing against them.

Its just I've been building many rigs for a very long time, these thing really makes significant different, anyone agree on me with this ?

Again, a "very subjective" matter, but don't ignore the statistic once they're become a significant numbers in the chart !

.[/QUOTE]

 

[yasha]

Wellcome to OcF Knight !

Thanks for the compliment, and you have an interesting question there !

Well, you've seen it your self that obviously WBC is increasing the STR/average transfer rate right ? Except for some trivial increase in CPU utilization, but I've tested it intensively in the past, it doesn't affect significantly on real-life usage nor at benchmarks.

Honestly, until today I still don't know what is the con by enabling WBC, I've traced carefully on the memory usage, and still no proof that it eats up more resources like memory usage.

But, I've been "guessing" that these WBC might have a bad effect on single core CPU, but alas so far there are not many samples reported yet to proof it.




Oh, sorry, I guess I wasn't having enough sleep when I responded to your 1st post.

Ok, if you mean those jaggies curve, have you disabled those background tasks like anti virus, firewall, running applications if any, disk defragmenter and etc ?

If you see my benchmark results, all those were created with disabled background tasks like my AV, firewall, defragmenter, any running applications, and .. internet, yes, sometimes internet ccould affect the benchmark result especially with that pesky Windows Update enabled.




Ok, this is not the 1st time I heard that "my rig feels so responsive", congratulation and glad you're enjoying it.

I know, most ICHxR Raid basher will say, Yeah right, how come your feeling can be an objective benchmark ? LOL, yes, have to admit that they have a valid point in that argument, nothing against them.


Its just I've been building many rigs for a very long time, these thing really makes significant different, anyone agree on me with this ?

Again, a "very subjective" matter, but don't ignore the statistic once they're become a significant numbers in the chart !

.

Damn, thought I did the multiquote reply correctly

Here are the Raid 5 results (4xPerps)

 

[yasha]

Most responsive, eh? W/ over 350 STR...that's sick...not right...I'm MAD envious.

What do the jumpers do on SATA drives anyway? Something to do w/ spread spectrum, 150/300, or NCQ? I suppose they'd be different from drive to drive. I'll have to take a look @ WD and Hitachi sites soon...unless any of you know off the top of your head?

Thanks!!

PS. The wacky lines in HD Tach don't mean much. If you really wanna test your stuff, copy some files...or maybe do some extraction tests. Nothing says lovin' like real world pron moving.

You're right....After looking through some previous benches from other members, it does seem like 350 MB/s STR is indeed nice But of course, I couldn't find any others with 4 X ST325410AS to compare so it's not apples-to-apples. I think fritzman had a setup with 4 X Perps once?

I plan to copy some large files between drives, etc. to test the speed of the RAID arrays. But just from installing XP 5 times already, I can tell it's at least 2-3 times as fast as my previous non-RAID computer.

 

[MadMan007]

Which version drivers and software are people using with the P5K Premium (if you're NOT using Seagate drives)? I've read about issues with non-Seagate drives on that mobo for some reason, could be some bad ICH9Rs around or something else but to be safe I went with the F6 drivers and software on the CD - 4x Raptor Raid0/5. The install is so fast I may just try the newest ones for fun though unless anyone else can say yes or no for sure.

 

[bing]

You're right....After looking through some previous benches from other members, it does seem like 350 MB/s STR is indeed nice But of course, I couldn't find any others with 4 X ST325410AS to compare so it's not apples-to-apples. I think fritzman had a setup with 4 X Perps once?

I plan to copy some large files between drives, etc. to test the speed of the RAID arrays. But just from installing XP 5 times already, I can tell it's at least 2-3 times as fast as my previous non-RAID computer.

Want little boost on your own ego ?

Try PCmark and post your bench's score at the Orb and compare your rig there with other rig around the world that has same CPU, similiar mobo class, same RAM and same OC clock speed !

Of course beware of iRam owners, your's can't compare with that cause its solid state "RAM" drive.

Which version drivers and software are people using with the P5K Premium (if you're NOT using Seagate drives)? I've read about issues with non-Seagate drives on that mobo for some reason, could be some bad ICH9Rs around or something else but to be safe I went with the F6 drivers and software on the CD - 4x Raptor Raid0/5. The install is so fast I may just try the newest ones for fun though unless anyone else can say yes or no for sure.

MM, just download the latest one from Intel's site, they're constanly fixing lots of problem at each revision. If I'm not mistaken, from last two revs, they've fixed alot of issues on ICH9R.

 

[MadMan007]

Yes I have tried both. You may not be aware of it though but there are some threads on Asus forums regarding the P5K Prem having serious SATA RAID issues, especially with non-Seagate drives. Remember how you said 'Seagate has been with Intel...' or something like that to me? One of their techs advised to use the drivers and software off the C. I used those, and then decided to reinstall and use the newer ones. Both seem ok with Raptors so far.

 

[yasha]

Want little boost on your own ego ?

Try PCmark and post your bench's score at the Orb and compare your rig there with other rig around the world that has same CPU, similiar mobo class, same RAM and same OC clock speed !

Of course beware of iRam owners, your's can't compare with that cause its solid state "RAM" drive.

Hehe, no ego here...I'm far from being an expert on benches and system building. Just an ordinary Joe who decided to indulge a little on some components for my latest build

I'll check out the Orb...Thanks for the tip!

 

[bing]

Yes I have tried both. You may not be aware of it though but there are some threads on Asus forums regarding the P5K Prem having serious SATA RAID issues, especially with non-Seagate drives. Remember how you said 'Seagate has been with Intel...' or something like that to me? One of their techs advised to use the drivers and software off the C. I used those, and then decided to reinstall and use the newer ones. Both seem ok with Raptors so far.

LOL, yeah, I forgot that my self !

Yeah, I made that conclusion because when I checked & searched the word "NCQ" at Intel's site while ago regarding to Intel Matrix Storage Technology, the only meaningful resources is a white paper created by joint venture from Intel and Seagate "only".

That sparked my "speculation" cause I was questioning why the heck they didn't invite other major drive makers like WD, Hitachi etc, why exclusively with Seagate ?
That paper should be in more "generic tone" since those sata stuffs is originated from a consortium.

When two giant corporations working together to create something, even as silly as a white paper, IMO, it is not as innocent as it looks, definately it is not like hanging out with your neighbour making a toasted meat in the garden & drinking cold drinks.

Maybe its just my personal impulse in creating my own "conspiracy theory", so don't take it too seriously ! LOL


.

 

[MadMan007]

I love conspiracy theories But a whitepaper is a document outlining information and noting more imo. The SATA standards commitee still trumps whatever they might come up with, but the fact that there is a whitepaper from only two companies Maybe the other companies implement it differently and while it still may be correct according to SATA standards it may not work with the ICH9. That would be really bad because it would mean Intel is making their parts non-standard. It could also be that the other companies are doing something boarderline wrong and it only shows up sometimes.

Anyhow all seems fine so far, no problems installing, no slowdowns so I'm happy.

 

[fritzman]

If you see my benchmark results, all those were created with disabled background tasks like my AV, firewall, defragmenter, any running applications, and .. internet, yes, sometimes internet ccould affect the benchmark result especially with that pesky Windows Update enabled.

I maybe O/T here guys (just shoot me if it's too far off, but...) I seem to have a lot of stuff going on in the background... some I recognise and some I dont... Is there like a standard list of stuff that you "must have" left going just for benching? or does everyone simply take pot-luck?

Here's my list...



Appreciate any pointers...

 

[bing]

Fritz, though I never use Vista and still on XP-Pro, but I recognized several items that I think worth to disable when benching drive :

- Perfect Disk -> PDAgent.exe & PDEngine.exe
- Symantec -> Every items that has a word "Symantec" on it, Symantec's stuff is known to be a hogware.
- Acronis scheduler
- Nero Home

To disable those items above, better to use the "service" applet rather than terminate them directly at the task manager screen, its much safer that way.

I use a batch files to terminate properly list of running services, since it is more convinient and don't have to memorize which is which.

Here are the example of these two batch files of mine, one for terminate and another one for starting my Perfect Disk services.

Example of my PerfectDisk_STOP.CMD

NET STOP PDAgent
NET STOP PDEngine

Example of my PerfectDisk_START.CMD

NET START PDAgent
NET START PDEngine

I just put these two files at my desktop for quick & easy access.

Another examples from my other one, this for minimal services termination, not the full blown one :

SLIM.CMD

net stop "automatic updates"
net stop "DHCP Client"
net stop "DNS Client"
net stop "Fast User Switching Compatibility"
net stop "Help and Support"
net stop "HTTP SSL"
net stop "Print Spooler"
net stop "Talking Alarm Clock user logon monitor"
net stop "Secondary Logon"
net stop "Shell Hardware Detection"
net stop "Universal Plug and Play Device Host"
net stop "SSDP Discovery Service"
net stop "Task Scheduler"
net stop "Themes"
net stop "User Profile Hive Cleanup"
net stop "WebClient"
net stop "Windows Firewall/Internet Connection Sharing (ICS)"
net stop "Windows Time"
net stop "Windows Audio"
net stop "Wireless Zero Configuration"

Yours could be different since on different OS, tasks and different installed applications and etc, and this method saves me lot of boring repetition when terminating those services.

You can do trrial and error to get this kind of list for your own and make the batch for it.

Also I cut off my internet link by switching off my DSL when I did this mass services termination, just another good & safe practice.

Once I've done with it, I just re-boot the rig to have them all restored back to normal.

I use this site HERE as a hint for my XP on which services, I never change their start-up attributes, just looking there and compare for which one is safe enough to terminate to make my rig to use minimal resources. A trial & error is also recommended to find the best combination.

For your Vista I recommend to take a look at this HERE.

 

[fritzman]

As always bing... you come up with the goods.

Thanks mate

I will have a play with those, and I love the batch-file idea.

B

 

[v8440]

-


If I get enough vote of interest for these amazing BartPE and DriveXML tool usages, I'm thinking to write down a dedicated guide on how to prepare and use it in this special Intel Matrix Raid environment.

Consider this a vote-I had 2 drive matrix raid 0/1, and I just added 2 more drives. I'm restoring the disk image using acronis true image, and it's taking roughly 4 1/2 hours. I can't imagine why it would take so long, but I'm all for trying something else that will allow me to restore images before I'm dead of old age.

 

[bing]

Consider this a vote-I had 2 drive matrix raid 0/1, and I just added 2 more drives. I'm restoring the disk image using acronis true image, and it's taking roughly 4 1/2 hours. I can't imagine why it would take so long, but I'm all for trying something else that will allow me to restore images before I'm dead of old age.

4.5 hours ? How big is your Raid 0 volume and what was the free space left at that Raid 0 volume when you imaged it ?

Aggreed, that is definately too long ! Was that restoring the image saved in the Raid 1 into the Raid 0 ?

Maybe other Acronis owners could help here ?

About the DriveImageXML, please refer to HERE which is mine or HERE that was made by JDany, IMO, his new concept is better !

Or start reading this thread starting at post #860 in page 29 by JDany until page 30, watch the discussion between us, I believe it should be enough to start preparation into using DriveImageXML intensively.

One note though, in those discussion we're not discussing the details on how to prepare BartPE CD, but I'm pretty sure it is not that difficult, just go BartPE website to learn on how to embed a Raid driver, add DiskPartitioner inside it.

And also don't forget to read "nLite" on how to put your the raid driver into your Windows XP installation disc.

.

 

[v8440]

Let's see, the original raid 0 volume (with two hard disks) was 125 GB. After adding two more disks, I DOWNSIZED the raid 0 array to 75 GB to improve seek performance. The array was WAY under 50% full at the time I imaged it-maybe under 25% full.

The raid 1 was the remainder of the drives, and it was really really close to empty-like probably 5% or something. So, the raid 0 went to the new raid 0, and the raid 1 went to the new raid 10. It ended up taking so long I said screw it and went to bed. I woke up a few minutes ago, and it seems to have worked. I'll put it this way-there's certainly no time savings by doing it this way-the only thing I gained was the ability to start it and go do other things. That and not losing any progress on my stalker game. The rest I could have easily backed up by simple file transfer and come out the same time-wise by reinstalling windows and the apps. By the way, the compression used during image creation was the default level.