6_6_6

NCQ: Best Upgrade For a Power User!

Recommended Posts

The problem with your method is that you only verify checksums once a month, and there's still the possibility that you could suffer silent data corruption and don't notice it until it's too late. You could have overwritten your backups with silently corrupted data by the time you discover the error.

Anyone doing backups knows how to ROTATE them too. Otherwise, they are not doing it right.

Like i said, silent data corruption is not an issue. I haven't lost a single byte in my entire life, and i never had a md5sum mismatching in 20+ years. I don't give a rat's bottom if silent data corruption (!) corrupted a frame out of an 8 GB pr0n file once a year.

I'm sorry, but if you don't have time to waste on figuring out how to enable or disable the write cache, then it's obvious that you won't be able to do the tests correctly.

In that case, I would politely ask you to stop spreading misinformation and confusing people.

My information is accurate. My 7200.11 on my system at this very moment does not allow WRITE-CACHE settings changed. It is always on. I can neither do it from device manager nor from issuing SET FEATURES 82h command to it... from DOS or from linux... hdparm does not work either. This has been the case in the past.

All the recent drives were tested with 82h command value and none allowed that to be changed. Older drives happily complied. It was sufficient for me to deduce newer drives do not allow modification. If you wish, get a bunch of drives, drop to a shell and issue ATA commands yourself and reread the values and post for all of us to read... But more importantly, just to satisfy your own curiousity. I am sorry, i can't be much more of a help on this issue.

Share this post


Link to post
Share on other sites
Anyone doing backups knows how to ROTATE them too. Otherwise, they are not doing it right.

Like i said, silent data corruption is not an issue. I haven't lost a single byte in my entire life, and i never had a md5sum mismatching in 20+ years. I don't give a rat's bottom if silent data corruption (!) corrupted a frame out of an 8 GB pr0n file once a year.

It may not be an issue for you, but it's certainly an issue for many other people. It's been very clearly described in this paper:

http://indico.cern.ch/getFile.py/access?co...mp;confId=13797

My information is accurate. My 7200.11 on my system at this very moment does not allow WRITE-CACHE settings changed. It is always on. I can neither do it from device manager nor from issuing SET FEATURES 82h command to it... from DOS or from linux... hdparm does not work either. This has been the case in the past.

All the recent drives were tested with 82h command value and none allowed that to be changed. Older drives happily complied. It was sufficient for me to deduce newer drives do not allow modification. If you wish, get a bunch of drives, drop to a shell and issue ATA commands yourself and reread the values and post for all of us to read... But more importantly, just to satisfy your own curiousity. I am sorry, i can't be much more of a help on this issue.

I'm willing to bet that your drive on your system DOES allow its write cache settings to be changed. It's just a matter of knowing how to do it. Windows clearly knows how to do it (as evidenced by the benchmarks).

Disabling the write cache is very important in certain circumstances. It would be criminal for a manufacturer to ignore requests to disable the cache!

Take a look at this page from Microsoft: http://support.microsoft.com/kb/888794

To help preserve the integrity of the Active Directory database if a power loss or another failure were to occur, the Active Directory directory service performs unbuffered writes and tries to disable the disk write cache on volumes hosting the Active Directory database and log files... If forced unit access is not supported, you must disable the write cache on all volumes of the guest operating system that host the Active Directory database, the logs, and the checkpoint file... You must disable the write cache for all components that use Extensible Storage Engine (ESE) as their database format. These components include Active Directory, the File Replication service (FRS), Windows Internet Name Service (WINS), and Dynamic Host Configuration Protocol (DHCP)

As you can see, disabling the write cache may not be such an issue for a home user, but it surely is a huge and important issue for a server!

Regards.

Share this post


Link to post
Share on other sites

Nice day to everyone,

I would like to ask you a little question. Right now I own HDD Samsung HD502IJ 500GB version, and I am really dissapointed with a poor performance when encoding video, listening to music and trying to play some PC game. Obviously there is some problem with NCQ, because normally (1 thread) it is quite fast.

Well, my interest is to buy another HDD as a Christmas gift, but with working NCQ. My OS is Windows Vista x64 SP1 and the capacity should not be less than 500GB. I have been wondering about Seagate 7200.11 series, which should be working as I expect, but after reading this topic I am not that sure anymore.

Anynone could help me with this? Thanks in advance.

Share this post


Link to post
Share on other sites
I'm willing to bet that your drive on your system DOES allow its write cache settings to be changed. It's just a matter of knowing how to do it.

It is not as if you are doing inverse integral calculations. You are just ticking 2 boxes. Do not make it sound as if it requires any special knowledge.

When there is an issue with my drive, Windows is the last place i resort to for testing in order to eliminate items from the equation. Windows come after all goes trouble-free.

Disabling or not, write-cache setting is DEFAULT in all recent drives. If you have any data to the contrary, please start your own thread and i will happily contribute there. This is AHCI/NCQ topic and this has gone too long.

Nice day to everyone,

I would like to ask you a little question. Right now I own HDD Samsung HD502IJ 500GB version, and I am really dissapointed with a poor performance when encoding video, listening to music and trying to play some PC game. Obviously there is some problem with NCQ, because normally (1 thread) it is quite fast.

Well, my interest is to buy another HDD as a Christmas gift, but with working NCQ. My OS is Windows Vista x64 SP1 and the capacity should not be less than 500GB. I have been wondering about Seagate 7200.11 series, which should be working as I expect, but after reading this topic I am not that sure anymore.

Anynone could help me with this? Thanks in advance.

Go with the Seagate. As can be seen from the initial post, I had samsung too and Seagate made all the difference in the world for me.

Share this post


Link to post
Share on other sites
It is not as if you are doing inverse integral calculations. You are just ticking 2 boxes. Do not make it sound as if it requires any special knowledge.

Sorry, I didn't mean to imply that. What I tried to say (English is not my first language, sometimes I find it difficult to find the proper expression) was that Windows knows how to do it, it has the necessary "knowledge" to be able to disable and enable a drive's cache at will.

I don't really know if there's more to it than ATA command 82h. But I do know that Windows can control a drive's cache behavior in most cases (even with the newest drives like the Barracuda 7200.11) Perhaps the ATA specs of new drives have changed and linux's hdparm hasn't been updated to the new way of doing things.

I'll try to get some more info and report back. And I'll open a new thread, I agree that we have come way off topic here.

Regards.

Share this post


Link to post
Share on other sites
Sorry, I didn't mean to imply that. What I tried to say (English is not my first language, sometimes I find it difficult to find the proper expression) was that Windows knows how to do it, it has the necessary "knowledge" to be able to disable and enable a drive's cache at will.

It sends the same ATA command to the drive :)

I'll try to get some more info and report back. And I'll open a new thread

Yes, that would be better. I will try to contribute with my tests when i have time. Did you check to see if the disable cache command survives a hard boot on the drive? As far as I recall, some drives allowed that to be modified temporarily but discarded it when powered down. Windows might be issuing the same command to the drive on each boot.

Just for curiousity, you can start a new thread about checksumming in filesystems too. There are some folks here with deep fs knowledge. It would be good to hear what they think.

Share this post


Link to post
Share on other sites

Here are some test results that beg for an explanation. They are the iPEAK tests Tech Report uses in their hard drive tests.

The hardware is

Asus P5K-E BIOS v1102

Intel P35 , ICH9R

Samsung HD642JJ , FW: 1AA01112 S/N : S1AFJ1MQ801006

The test were done in Windows XP and 2003, drive is connected to the Intel ICH9R adapter.

BIOS is set to IDE mode, then to AHCI mode.

Driver for AHCI is Intels IMSM v8.5 (v8.6 gives the same results).

One can see that in one test IDE mode is FOUR times faster than AHCI.

ipeakxn2.png

w958.png

Share this post


Link to post
Share on other sites
Could you please test under Windows 2008 using Microsoft AHCI drivers and report the results?

No, because iPEAK does not work in 2008 (see my other thread).

But practical experirence shows that the disk behaves the same in win 2008. (that practical experiences are the reason I did the iPEAK test in the first place)

(maybe I'll try to "transplant" the MSAHCI driver into win 2003 to try...)

(if I'll have time)

Regards,

David

Share this post


Link to post
Share on other sites

Upgraded from 750GB Seagate 7200.11 ST3750330AS to Seagate 7200.11 1TB ST31000333AS. I had to put 750GB back since things came quickly to a halt with 1TB. This is surprising.

750GB: 110MB/s single, 98MB/s double.

1TB: 120MB/s single, 22MB/s double.

Edited by 6_6_6

Share this post


Link to post
Share on other sites

You missed an important thing-- which firmware?

and I'm confused by your above post-- the 750GB/1TB comparison is with Seagate 750GB and 1TB? Or is it with Samungs? Or both?

Share this post


Link to post
Share on other sites
You missed an important thing-- which firmware?

750GB is the same one I have been using from the very beginning -- SD15.

1TB is CC... something (not affected by firmware bugs).

and I'm confused by your above post-- the 750GB/1TB comparison is with Seagate 750GB and 1TB? Or is it with Samungs? Or both?

Single instance of all drives is about 100-120MB/s.

7200.11 ST3750330AS 750GB: 98MB/s double.

7200.11 ST31000333AS 1TB: 22MB/s double.

Samsung Spinpoint F1 1TB: 22MB/s double.

Edited by 6_6_6

Share this post


Link to post
Share on other sites

How good that I found this thread, I am really puzzled about my drives multithreaded performance! I have a new Seagate 7200.12 1TB and an old Samsung HD501LJ. The Seagate has amazing single threaded performance, the Samsung a little less, but when I try just 2 reading threads they almost die. I mean both of them get an average of 7.8 mb/s (or 15.6 mb/s if you say double) and they really lock up your system when you write to the system drive.

I am on a X38 chipset based motherboard with ICH9R. I tried all different Intel RAID and AHCI drivers but it's exactly the same. OS is XP 64-bit.

hdspeed2.png

hdspeed4.png

Share this post


Link to post
Share on other sites
How good that I found this thread, I am really puzzled about my drives multithreaded performance! I have a new Seagate 7200.12 1TB and an old Samsung HD501LJ. The Seagate has amazing single threaded performance, the Samsung a little less, but when I try just 2 reading threads they almost die.

Have you tried WD Caviar Blue or Caviar Black drives? According to my experience, WD drives are really considerably faster in these tasks.

Share this post


Link to post
Share on other sites

I posted in another topic, i will add here too.

I have not come across any other hard drive except 7200.11 750GB that does not stutter my system and they are not even boot drives (no hd activity).

I have an Intel SSD and it doesnt even cut it.

Intel X25-M (Boot - NCQ)... Copy a 10GB file to it... And try playing a movie. It stutters and skips and can't even open very lowprint Miranda IM. Lag.

Seagate 1TB 7200.11 (Non-Boot - NCQ): Copy a 10GB file to it... And try playing a movie. Stutters and skips.

Seagate 750GB 7200.11 (Non-Boot - NCQ): Copy a 10GB file to it... Run a virus scan... Encode a 1GB video... Zip a 10GB file... And try playing a movie... It works smooth, everything is responsive.

Intel X25-M (Non-Boot - NCQ): Same crap.

Seagate 750GB 7200.11 (Boot - NCQ): Everything works fantastic even with 5-6 simultaneous disk-operations even though this drive is boot drive.

So for me, Intel X25-M did not make a difference at all. I would take Seagate 750GB any time of the day. Now I have SSD boot drive and I needed more space so replaced 750GB with 1TB... What a torture! I could not take the it more than a day and back to 750GB. Tried 1TB WDs, Samsungs and still no go.

Share this post


Link to post
Share on other sites

Now, forget about benchmarks, MBs, etc. I will time a common operation i do regularly on command line and see who completes the task faster.

I used timeit.exe from 2003 Server Res Kit. Free download from MS.

TASK (about 30GB read)

2 runs of MD5 sum on a total of mixed 15GB [1800 files] data (COMMAND: md5.exe /R /T:R)

1 run of Clam Antivirus on a total of mixed 5GB [4000 files] data (COMMAND: clamscan.exe -d D:\ProgramData\.clamwin\db -r --quiet)

2 runs of winzip on a total of mixed 5GB [4000 files] data (COMMAND: wzip -en -p -r -bE:\SPEED_TEST\DIR4\ dir4.zip *)

1 run of copy single 5GB file (COMMAND: copy d:\5GB .)

NOTES:

1. System has been rebooted before every test run.

2. Zip, etc all use the same drive for temp files.

3. Single file has been copied from SSD, and in case of SSD, from another drive.

4. Intel x25-M always boot drive, system has no disk activity.

5. Data at teh beginning of the drive in all drives except SEAGATE 1TB where it is at the end (sorry, cannot move around 1TB now)

SUBTASK (same but just read performance)

2 runs of md5 sum

1 run of Clam Antivirus

CONTENDERS

INTEL X25-M 80GB (SSDSA2MH080G1GC Fw:8820)

SEAGATE 7200.11 750GB (ST3750330AS Fw:SD15)

SEAGATE 7200.11 1TB (ST31000333AS Fw:CC1H)

WD GREEN 1TB (WDEADS-00L5B1 Fw:01.01A01)

RESULTS

1 . When all tasks run sequentially one after the other, all drives took almost the same time to finish: 9 Mins

2. When READONLY portion of tests is run simultaneously [3 threads](Best first):

1:24 Intel

4:50 Seagate 750GB

9:56 WD

21:23 Seagate 1TB

3. When ALL tasks run simultaneously [6 threads](Best first):

7:42 Seagate 750GB

7:55 Intel

10:26 WD

23:45 Seagate 1TB

CONCLUSION

Gentelmen, I am not talking about 1 or 2 percent difference in sustained transfer rates, access times and fvck-knows what . WE ARE TALKING ABOUT 5 MINUTES DIFFERENCE IN A 10 MINUTES RUN OF SOMETHING I FREQUENTLY DO!!!

Seagate 1TB has awful results but that is understandable. It was almost full and data was at the end of the drive. Nonetheless, I tested that drive when i bought it first and it was crap. I dont find it necessary to test it again just to enter the numbers. I assume it is somewhere near WD.

Intel SSD is a bigger piece of crap. Yeah it hits 250MB/s constant even when you read 10 threads from it but so what? I guess it is only good for Linux LiveCDs! It locks the system when there is big (4-5GB) read or write to it. This lock/freeze issue has been with every drive except Seagate 750GB. You can't run a movie without stuttering in any of these drives (except 750GB) while there is any sort of disk access.

This issue is also not an NCQ matter as the topic wrongly titled... It is more AHCI matter (the very same Seagate 750GB sucked without AHCI).

Edited by 6_6_6

Share this post


Link to post
Share on other sites

6_6_6 =>

I have made myself similar comparison among Seagate 7200.11 ST3640323AS, WD6400AAKS and WD6401AALS. All have 640GB, 2 platters, 7200RPM. I used Linux and Ext3 filesystem. The result of WD drives were similar, but both kicked Seagate's ass with more than 20% advantage in time.

So. I don't understand, why your results differs so much, and I don't think they are incorrect. But I don't remember anybody else claiming that 7200RPM Seagate drives work well in real multitasking multiuser environment. Oh. I have another question. You tested Caviar Green. Have you tried Caviar Black?

And have you any experience with Barracuda 7200.12?

Share this post


Link to post
Share on other sites
So. I don't understand, why your results differs so much, and I don't think they are incorrect. But I don't remember anybody else claiming that 7200RPM Seagate drives work well in real multitasking multiuser environment.

Dude, I don't know what the heck you are talking about but I AM ***USING*** THIS. And I have been using it since a year. Which part of this you are not getting?

And I started using it on RECOMMENDATION from somebody else on this forum! Reference at the first post of the topic.

I don't care about your WD benches, my Intel SSD kicks it multitude of times instead of meager percentages and it even does not cut it for me.

Now, go do a zip, md5 checksum, a big file copy and a virus scan simultaneously on your WD and tell me if you can do anything else or if your computer comes to a halt. If it doesnt come to a halt, please post results.

Share this post


Link to post
Share on other sites
Now, go do a zip, md5 checksum, a big file copy and a virus scan simultaneously on your WD and tell me if you can do anything else or if your computer comes to a halt. If it doesnt come to a halt, please post results.

I did it many times, but because of this conversation, I have tried it once again:

WD6401AALS drive, Opensuse 11.1 64bit, Ext3 filesystem:

- 3 simultanous instances of tar+gzip compression +/- 11GB of files each.

- copying large amount of files

- playing video in SMPLayer

- seaching file on the disk

No problem at all. System does not hang up, video is smooth, all of these task proceed and i can do more and more.

In my experience, trying this with Baracuda 7200.11 640GB resulted in to slower environment, occasionally hanging up, and tasks completed noticeably later. I don't have any Barracuda here, to make some comparison once again. That's why I'm asking and collecting information. I don't understand, what's your problem with it. B)

Share this post


Link to post
Share on other sites

This conversation is about Windows. Do all these in Windows and post your results here. I had no problem doing any of these with my 10-year old 40GB fireball on a Slack.

Edited by 6_6_6

Share this post


Link to post
Share on other sites

6_6_6 =>

I don't have any Windows installation. And as I said, I trust your results. The only thing I was asking was experiences of Barracuda 7200.12 and Caviar Black in this topic.

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