6_6_6

NCQ: Best Upgrade For a Power User!

Recommended Posts

I believe this warrants a new topic for those of you who does not want to dig through pages of posts. For referenence, here they are:

http://forums.storagereview.net/index.php?showtopic=26549

http://forums.storagereview.net/index.php?showtopic=26006

Simply put: PROPER WORKING NCQ is without a doubt, the single most important upgrade for a power user in almost a decade i believe. Now you can multitask.

I didn't need any more processing power than my good ole 3.0 GHz Northwood purchased 8 years ago. But nonetheless, upgraded to Q6600. Imagine my surprise to find that my new system with quad cores o/c'ed to match old clock (@3.0GHz), 8600GTs, half a tb of fastest SATAII NCQ drive did not seem even a bit more responsive than the one bought almost a decade ago.

I couldn't do a copy or virus scan operation without having my system in complete halt until the task completed. Took 3dmark 3 mins to start! Where the heck was multitasking? I use my pc heavily for multimedia, leisure, business and come on, let's face it... not everyone plays games whole days and nights... so latest and fastest CPU and GPU meant nothing for me. My system performed not any different than the one bought 8 years ago.

As it turns out, I just needed to get SATAII NCQ properly working... meaning that I needed to find a drive that has NCQ working properly as opposed to one that advertises it has NCQ. After reading czr's topic and tests, i started hunting for a 500GB 7200.11.. But i still could not find this drive where i am... and i was not willing to go for a 3-platter 750GB version (heat-noise). I wish i had... i wouldn't be wasting so many months.

HD_Speed Results (http://www.steelbytes.com/?mid=20)

SEAGATE ST3750330AS 7200.11 (750 GB, 7200 RPM, SATA-II)

1 instance: 110 MB/s

2 instance: 103 MB/s (0% and 50%, 256K blocks)

10 instances: 75-80 MB/s (at 10% intervals, 256K blocks)

SAMSUNG HD501LJ (500 GB, 7200 RPM, SATA-II)

1 instance: 98 MB/s

2 instance: 20 MB/s

I no longer need to wait 3 mins for a program to start up when i am copying some files :)

And best of all, I needed to do nothing. I just ghosted my samsung to seagate and rebooted... All worked right away.

If you are using a single drive, if your os partition and data is on a single drive, i wouldnt imagine something more useful than a proper NCQ implementation.

Thank you, thank you, thank you all... especially CZR... for bringing this to my attention. I can heartily say that this is the best upgrade for me in the last 10 years.

PS: My pc is the same, absolutely no changes... only the hard drive changed (From 500GB supposedly SATAII supporting Samsung to Seagate 7200.11).

Here are my specs from everest:

QuadCore Intel Core 2 Quad Q6600, 2400 MHz (9 x 267)

Gigabyte GA-X38-DQ6 (2 PCI, 3 PCI-E x1, 2 PCI-E x16, 4 DDR2 DIMM, Audio, Dual Gigabit LAN, IEEE-1394)

2048 MB (DDR2-800 DDR2 SDRAM)

NVIDIA GeForce 8600 GT (256 MB)

HP w2207 [22" LCD] (CND7293M40)

Intel� ICH9 SATA AHCI Controller

SAMSUNG HD501LJ (500 GB, 7200 RPM, SATA-II)

ACPI x64-based PC

Microsoft Windows Server 2008 Enterprise

O/c: 8 x 400MHz 3.2 GHz @1.35v, 1067 MHz DDR2 5-5-5-16 CR2

Share this post


Link to post
Share on other sites

Interesting! I'll have to consider NCQ more carefully before making my next drive purchase.

Does anyone know if the Western Digital WD6400AAKS 640GB drive handles NCQ well?

Share this post


Link to post
Share on other sites

Interesting: As kreativ I'm buying a couple of wd 640Gb wich probably I'll configure in raid. NCQ is usable also in raid configuration?

Raid features will be provided by my ICH9R southbridge: to use this features I need to set it in raid mode fom motherboard bios.

NCQ is spported only by ahci settng or also in raid seting?

Share this post


Link to post
Share on other sites

Yes, NCQ is usable in raid configuration. I have not tried RAID/AHCI option though. Mine is on AHCI.

I might be out of line and it is not relevant but... I dont understand why you would need mirroring on a pc. Usually, if there is a prob with a drive, there is a high probability that there would be a problem with the second one as well... Probably some PSU / electrical / voltage issue... I stress test for my drives for 2 days... I never had a drive fail on me after passing stress testing... But then again, I rotate my system drives every year. It is just wiser to plug in the 2nd drive and clone your entire contents to it every week or month and unplug it after backup so it doesnt remain connected to electricity. I believe RAID 1 is indeed redundant!

I have no idea for WD... but I already had a Samsung SATA-II with NCQ support... AND IT DID NOT WORK.

So when I read czr's test and post, I wanted to get the exact same drive with teh same model number... I did not want to take a chance seeing how already NCQ support meant nothing! or meant as androidi put so nicely: "we promise no data will be corrupted if you attempt to use NCQ but the support was hacked in there by a summer intern in 2 days." So it is wiser to stick with a drive known to have implemented it properly... or perhaps some nice folks will post results for the drive you are looking for.

I am so stupefied to have read so many reviews saying NCQ has no benefit for desktop users... So blatantly wronggggggggg. Before I had to wait for the disk operation to complete to be able to do something else on my pc... Almost the whole computer would lock (since hte OS partition is on the same disk on a single disk setup). Quad cores... bestest and fastest computer hardware meant nothing... absolutely nothing. Now I can really multitask.

czr, thanks again... and some more thanks to androidi for testing his samsungs.

Interesting: As kreativ I'm buying a couple of wd 640Gb wich probably I'll configure in raid. NCQ is usable also in raid configuration?

Raid features will be provided by my ICH9R southbridge: to use this features I need to set it in raid mode fom motherboard bios.

NCQ is spported only by ahci settng or also in raid seting?

Edited by 6_6_6

Share this post


Link to post
Share on other sites

On the downside...

This Seagate 750 GB is a whole lot noisier than my Samsung. I have a fanless system... and Seagate is audible during normal operation (3 platters) and seeks are very audible too. In comparison, I had no idea my Samsung was sitting there (i have my drives suspended by rubber to eliminate vibration). I will have to wrap this drive in sound insulation material which would probably shoot up the temperatures and cause premature failure. You cannot do acoustic management on it. Well, i am not intending to put up with its noise. SSD time soon.

Edited by 6_6_6

Share this post


Link to post
Share on other sites

Your idea about hard disk possibility of damage is surely true and I also had considered in the past that hdd damage are probably caused by envoiremental problems. Then if one disk fall in error it's probably that also ther disks of te computer result damageed.

But I also thought that external drives used for backup are much more exposed to danger of damages: they could fall, the cat or mother could make them fall, over a busy desk they could be forgoten and heated...

Then I thought that a second hard drive in mirror would be faster, cheaper and always upgraded, while a backup hard disk requires continius upgrades to make me sure I won't loose anything in the eventuality of damages to my storage disk. And I'm not so patient and precise...

Then I prefere a mirror sistem: it is cheap then I don't llose anithing doing it and it's surely more safe than the single storage dedicated disk...

I red I think in this topic about terrible performances of western digital disks when used in raid. I'd like to understand if it was a problem of only one ore some wd series or it's norma for every wd disk... In this case, It wuld be better to buy a 7200.11 disk.

I thank you for your answer about possibility to enable NCQ on ICH9R in raid configuration. Then when I set it to raid, NCQ is automatically enabled as it would be when set in AHCI mode..?

Thank ypu and sorry for my uncorrect english... I don't speak it from centuries..

Share this post


Link to post
Share on other sites
But I also thought that external drives used for backup are much more exposed to danger of damages: they could fall, the cat or mother could make them fall, over a busy desk they could be forgoten and heated...

Umm... you might be quite surprised how much a hard drive can take when not powered on. A friend of mine gave me a damaged disk. I tried repairing it but i could not succeed. The disk was just not visible even though it was spinning. I took it to the dealer... But dealer closed shop and i was trying to find main distributor for RMA. This disk stayed on the passenger seat of my car 3 months... It was thrown out of the car when being cleaned... It was stepped on, kicked out, thrown about... Then I put it to the truck of my car. It stayed a good 3 months more in the trunk of my car and the temperatures here are about 50-60C (122-140F) in the shade in summer. I reckon it must have been close to 80C in the truck of my car during the day when i was driving. Then suddenly, after 6 months it downed to me that it might be due to a master password... And voila... After 3 hours of master password recovery... The disk was back to life... And it still is working :)

Then I thought that a second hard drive in mirror would be faster, cheaper and always upgraded, while a backup hard disk requires continius upgrades to make me sure I won't loose anything in the eventuality of damages to my storage disk. And I'm not so patient and precise...

Then I prefere a mirror sistem: it is cheap then I don't llose anithing doing it and it's surely more safe than the single storage dedicated disk...

Well... whatever your reason might be... It just is not safe, you are assuming wrong. I would not trust my data when it is constantly connected to power. A backup disk is something you power on for an hour or so, then unplug and stash it away till the next time. It just takes me few clicks with ghost to clone my entire drive... It takes few hours when i take my system offline (you can do this online with shadow copies and all as well) and I do not have to deal with the additional heat/noise and power requirements of a second drive that is running all the time... And for the cost, it is the same... You use the same mirror drive as your backup drive. Just uplug it when done. That is my take on the issue. I have seen many friends whose all or most powered drives fried during a spike, psu failure or those infamous exploding AMDs with a failed fan... Yet, I have not lost a single byte of data in my 20 or so years of computer usage... Yeah... ghost... from binary research... back in the beginning of 1990s :) still the same ghost for me :)

I red I think in this topic about terrible performances of western digital disks when used in raid. I'd like to understand if it was a problem of only one ore some wd series or it's norma for

every wd disk... In this case, It wuld be better to buy a 7200.11 disk.

Can't comment, have not used RAID much. But i heard the same things about them in RAID. The drive mentioned above was WD Caviar 200GB by the way. WDs worked good for me.

Then when I set it to raid, NCQ is automatically enabled as it would be when set in AHCI mode..?

I am assuming so. I have not tried it though.

Share this post


Link to post
Share on other sites

Yur story is uncredible!! but unluck is always looking for you and if you have an hdd with important data and it falls, yu can be sure it will be lost. If you have to throw away a disk and you pass voer it with an abrams tank, it will be saved..

Anywere your idea is true and it helped me to decide what to do with my actual seagate 500Gb disck when it will be replaced with new disks in raid... I'll use external sata connector to plug it just for backup my storage!

Now the problem bout wd persist... how are them in raid? who helps me?

Share this post


Link to post
Share on other sites
On the downside...

This Seagate 750 GB is a whole lot noisier than my Samsung. I have a fanless system... and Seagate is audible during normal operation (3 platters) and seeks are very audible too. In comparison, I had no idea my Samsung was sitting there (i have my drives suspended by rubber to eliminate vibration). I will have to wrap this drive in sound insulation material which would probably shoot up the temperatures and cause premature failure. You cannot do acoustic management on it. Well, i am not intending to put up with its noise. SSD time soon.

A fanless quad-core system...impressive!

As you have discovered, Seagate doesn't implement AAM in their drives (some legal issue from years back). Seagate SATA drives come locked in performance mode and Seagate PATA drives in quiet mode. Which makes Seagate SATA drives not very good for silent systems. The WD 640GB has AAM (comes with it enabled) and is much quieter compared to both my Seagate 320GB SATA (no AAM) and WD 500GB SATA drives (AAM enabled). Whether it handles NCQ properly like the Seagate 7200.11 is yet to be determined...

Share this post


Link to post
Share on other sites

What do you say, could be a good idea a couple of that 640s for striping (and mirroring in a second time)?

I'd like to order them this evening (here in Italy it's midnight..) but I'm not able to decide...

What to do to disable AAM?

Share this post


Link to post
Share on other sites
A fanless quad-core system...impressive!

I am using fanless computers for a decade now. I do have a Zalman Reserator XT (fan removed, pump wrapped) this time... It cools my SilentMaxx 500W PSU, Q6600, 8600GT and HD. And Q6600 is o/c'ed to 3.2 GHz. Not a single problem for almost a year now on this setup and it runs 24/7 and temps are within specs.

Yes noise is an issue... and this Seagate... is just not it. I had AAM disabled on my samsung by the way and yet it was inaudible on a fanless setup.

Well... actually. Seagate noise is not an issue no more :) It runs quite cool though... was 35C without water... now with water and insulated/wrapped... at about 50C idle. It is inaudible :)

Share this post


Link to post
Share on other sites
What to do to disable AAM?

You can use manufacturer's disk utility... I used Hitachi's FTOOL to set/modify AAM. HD Tune Pro does that too i believe.

Share this post


Link to post
Share on other sites

Received my WD6400AAKS and tried the same test as 6_6_6 with HDSpeed. With 2 instances, transfer rate dropped to 34MB/s. So either WD6400AAKS does not have good NCQ implementation or it's my motherboard chipset (nVidia nForce4, SATA-IDE driver v6.92 with NCQ enabled in WinXP).

Share this post


Link to post
Share on other sites

Nice research work there.

Any idea how much of this applies to scsi drives?

I was running a disk intensive pgm the other day on one drive and found my entire disk subsystem was unresponsive (I have 3 seperate physical drives).

Share this post


Link to post
Share on other sites
Any idea how much of this applies to scsi drives?

I had two 36 GB Cheetah U320 15K3 's and ran them on a Gigabyte 8IKNXP-Ultra which had an onboard U320 controller.

1 instance at 80 mbps

2 instances at 25 mbps total

That is when i realized SCSI was not going to do it for me. Yes it was fast and responsive... but it would still block on disk operations.

Same results with an Adaptec U320 card as well (29320 or something it was i believe). Both drives posted similiar results.

These SCSI drives are old... but they were unused (3-4 hours total uptime on each). Could you please run hd_speed on your drive and post specs and results kittle?

I have no idea why this NCQ is so downplayed... but I am attributing it to falsely advertised drives that didnt perform for the testers so they couldnt get a taste of it. BIOS driver, OS, etc... issues... all bullshit. This all boils down to the HARD DISK.

Edited by 6_6_6

Share this post


Link to post
Share on other sites

Excellent information!

Am I correct that the only SATA drives we know of that perform proper NCQ right now are the Seagate 7200.11 models? I guess that's what I'm buying then.

Nick

Share this post


Link to post
Share on other sites

I'll run the benchmarks on my own HDs here soon.

but if SCSI doesnt support NCQ or if there are drivers issues with XP x64 -- then I'll be making some big hardware changes.

Share this post


Link to post
Share on other sites

ok. results of HD_Speed

disk 1: Maxtor Atlas 10k 3 18gb u320

1 instance: 41mb average

2 instances 18mb average

disk 2: Maxtor Atlas 10k V 74gb u320

1 instance: 84mb average

2 instances 15.8mb average

disk 3: Seagate ST3146707 10krpm 147gb u320

1 instance: 35.6mb average

2 instances 16.5mb average

howeve i can also 1 run instance on all 3 drives at the same time and get the same results as running 1 instance on each drive by itself.

drive 1 is for os files ONLY - purchased in 2005

drive 2 is for data (or was until it filled up) - purchased on 2005

drive 3 is my upgrade for christmas this year (2008).

controller is the onboard LSI logic controller for my Tyan K8WE motherboard.

Share this post


Link to post
Share on other sites

These are awful numbers. Do yourself a favor and get a 7200.11 and ditch these altogether.

You are unlikely to saturate the bus by running the tests parallel on each drive. That doesn't apply in our scenario anyway. I could get good numbers by running hds in parallel even if they did not support NCQ.

ok. results of HD_Speed

disk 1: Maxtor Atlas 10k 3 18gb u320

1 instance: 41mb average

2 instances 18mb average

disk 2: Maxtor Atlas 10k V 74gb u320

1 instance: 84mb average

2 instances 15.8mb average

disk 3: Seagate ST3146707 10krpm 147gb u320

1 instance: 35.6mb average

2 instances 16.5mb average

howeve i can also 1 run instance on all 3 drives at the same time and get the same results as running 1 instance on each drive by itself.

drive 1 is for os files ONLY - purchased in 2005

drive 2 is for data (or was until it filled up) - purchased on 2005

drive 3 is my upgrade for christmas this year (2008).

controller is the onboard LSI logic controller for my Tyan K8WE motherboard.

Share this post


Link to post
Share on other sites
Am I correct that the only SATA drives we know of that perform proper NCQ right now are the Seagate 7200.11 models? I guess that's what I'm buying then.

So far, all 7200.11s and Samsung F1 1TB.

The F1 1TB did much better with only two instances, averaging about 45MB/s for a total of 90MB/s.

The F1 750GB didn't do so good, with only two instances, averaging 10MB/s for a total of 20MB/s.

Oh, the single instance on the 1TB ran at an average of 114MB/s. The 750GB at an average of 98MB/s.

I'm using an Abit IP35 Pro motherboard with the ICH9R controller under Win XP in AHCI mode.

Share this post


Link to post
Share on other sites

WD6400AAKS Western Digital 640GB

HD Speed, 2 instances (0% and 50%), 256k blocks

With NCQ on, speed is maintained at 110MB/s but alternates between both instances every second. So when one instance is at 110MB/s, the other is at 0MB/s. So the average is still 55MB/s.

Is this normal for NCQ? Or should I expect to see a constant 110MB/s for each instance if NCQ is properly implemented?

Share this post


Link to post
Share on other sites

Yes, that is how it should be... NCQ alternates the read/writes so it does not block. Put the timing at 1 minute below (1:00) and let both instances run and sum the averages of both. You can also check the total bandwidth from TASK MANAGER > PERFORMANCE > RESOURCE MONITOR (on vista). It usually matches the hd_speed total if no other stuff is running.

This is how it is with NCQ on (please note that i already have emule writing my disk):

SATA_SEA_750GB_1.jpg

SATA_SEA_750GB_2.jpg

WD6400AAKS Western Digital 640GB

HD Speed, 2 instances (0% and 50%), 256k blocks

With NCQ on, speed is maintained at 110MB/s but alternates between both instances every second. So when one instance is at 110MB/s, the other is at 0MB/s. So the average is still 55MB/s.

Is this normal for NCQ? Or should I expect to see a constant 110MB/s for each instance if NCQ is properly implemented?

Share this post


Link to post
Share on other sites

I somehow cannot edit my post... Well, in this example:

Average for 1st instance for 1 min run is: 76 mbps

Average for 2nd instance for 1 min run is: 29 mbps

But as i noted, i have emule reading/writing to my disk at the moment... On an empty drive, the averages are close to each other (was 48 mbps for each instance before as far as i recall).

Therefore, NCQ must be working for you then. Post the results (i believe in your previous test, NCQ was not on).

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now