czr

NCQ performance on Seagate 1TB 7200.11

Recommended Posts

Let me start of by say WOW! I'm no longer disappointed with NCQ implementation, why the hell did it take so long to get right!

Finally a drive that performs the way I had expected them to from when NCQ was introduced!

I used this program http://www.steelbytes.com/?mid=20 for testing (fantastic utility)

Ran 9 instances of it, "Position" of 0%, 5%, 10%, 15%,.....,35% using "Block Size" of 256KB

Windows XP Performance Monitor, specifically "Disk Bytes/sec"

The 2 drives I compared were, with the following results.

Seagate ST31000340AS (7200.11)

Firmware : SD01

Size : 1TG

1 instance of the program at 0% Position ~ 100MB/s (to determine baseline performance)

9 instances of the program ~ 90MB/s (WOW!!!!)

Seagate ST3750640AS (7200.10)

Firmware : 3.AAE

Size : 750GB

1 instance of the program at 0% Position ~ 77MB/s (to determine baseline performance)

9 instances of the program ~ 20MB/s (really disappointing)

And check this out, the difference is actually AUIDABLE, you can hear the 7200.10 seeking madly away with 9 instances, where as the 7200.11 is near silent without any noticeable seek noise (what I typically assumed NCQ was suppose to deliver but never did, i.e. less seeks)

So my question is was NCQ broken or not implemented at ALL or not implemented properly in previous generation of drive? It makes me really mad to see just how NCQ was marketed in previous generations of drives but failed to deliver.

All I can say it then god NCQ has finally come of age and well done Seagate for the 7200.11 1TB drive, highly recommended for multi user environments, and people understand that a multi user environment is on everyone's desktop, i.e. pretty much all P2P Applications.

Share this post


Link to post
Share on other sites

@thegreek

We don't sell Hitachi drive normally, so I can't.

Next time we get a WD Raptor in I'll try it with them, but from memory I have a feeling I've already tried this and they did not perform that well either (not as well as I'd have expected).

@Mei

Wether it is related to larger cache and/or buffered I/O (which is possible, but my instincts tell me maybe not) they have definitely done something right this time around.

Share this post


Link to post
Share on other sites

@czr Did You try to do some tests without using sata2 ??

For example on sata(1) or with ncq off ?? What are the results ??

Could You test it again using HDTune and HDTach ??

ps

sorry for my english ;)

Share this post


Link to post
Share on other sites

Just did some more testing, this time with a WD drive

Western Digital WD2500YS

Firmware : 01SHB1 (??)

Size : 250GB

1 instance of the program at 0% Position ~ 63MB/s (to determine baseline performance)

9 instances of the program ~ 22MB/s (still crap)

So as you can see shitty performance in multi user wasn't just limited to Seagate

@luckass

No, I have not carried test out at SATA 1, well not recently (if you can consider such a standard actually exist) or with NCQ off, as this requires more time than I have free ATM

WRT HDTune and/or HDTach, I refuse to use those applications as I consider them to be totally rubbish (Garbage to you Americans) and of no real value.

Share this post


Link to post
Share on other sites

Also in my original post I meant to say "less head/actuator movement" and not "less seeks"

BTW I think the "edit" feature is badly implemented, the "time out" is far too short

Share this post


Link to post
Share on other sites
BTW I think the "edit" feature is badly implemented, the "time out" is far too short
NIce tests. :) Very good to know the results.

And the edit feature's short timeout is quite deliberate. Thanks. :)

Share this post


Link to post
Share on other sites

As promised, here are the results for a 74GB Raptor

Western Digital WD740ADFD

Firmware : 00NLR5 (??)

Size : 74GB

1 instance of the program at 0% Position ~ 84MB/s (to determine baseline performance)

9 instances of the program ~ 36MB/s (still crap, but better)

Well I’d say that’s still an indication that NCQ is not up to the task on this drive, granted the performance IS better, but it’s what I’d expect form a 10K drive which has much faster seeks, with 9 instances of the software running the drive is very loud and madly seeking away.

So far the Seagate 7200.11 1TB drive seem to be the best at multi-reads, I wonder if the rest of the 7200.11 will perform the same, if so then the smallest in that series is going to be my main O/S drive.

Now to start building some custom NAS/SAN with those drives and see how they perform

Share this post


Link to post
Share on other sites

Just did testing with R10 - 4x36Gb 15k sff

PCIe HP Smart Array P800 512Mb SAS RAID Firmware 4.06 / Drivers 6.8.0.64

4 x HDD HP 36G 15k SAS 2.5" [sT936751SS] Firmware HPD4

Size: 67.7 Gb

1 instance of the program at 0% Position ~ 204MB/s. Only two hdd activity leds blinking.

2 instances at 0% and 51% position ~ 391MB/s. All four hdd activity leds blinking.

9 instances of the program approx. 135-161MB/s. All hdd blinking.

Share this post


Link to post
Share on other sites
@czr Did You try to do some tests without using sata2 ??

For example on sata(1) or with ncq off ?? What are the results ??

Good idea. It will be interesting to see if performances are nice with NCQ disabled, please czr ;)

Share this post


Link to post
Share on other sites
Let me start of by say WOW! I'm no longer disappointed with NCQ implementation, why the hell did it take so long to get right!

Seagate ST31000340AS (7200.11)

Firmware : SD01

Size : 1TG

1 instance of the program at 0% Position ~ 100MB/s (to determine baseline performance)

9 instances of the program ~ 90MB/s (WOW!!!!)

All I can say it then god NCQ has finally come of age and well done Seagate for the 7200.11 1TB drive, highly recommended for multi user environments, and people understand that a multi user environment is on everyone's desktop, i.e. pretty much all P2P Applications.

Can you please run some benchmarks with NCQ OFF vs ON on this drive?

Multi-user sometimes goes along well with multi-tasking, so i would like a test with a few typical applications running at the same time, like my typical config: outlook, firefox, uTorrent, excel, an antivirus and WMP.

It seems a great performer. I'm debating wheater to get the WD greenpower or this. I got tired of waiting for the Samsung F1 :)

Share this post


Link to post
Share on other sites

does anyone know if the ST31000340AS has TLER enabled on them? Seagate's website mentions that they are for desktop raid so this would imply yes but it's not stated anywhere. Was thinking about some of these for my home system opposed to the ES.2's my desktop computer here uses an areca raid controller.

Share this post


Link to post
Share on other sites

Yes they do, although if you note, WD is the only maker who actually advertises TLER, probably because of WD's horrid reputation with their HD's in RAID. ;)

Share this post


Link to post
Share on other sites
Multi-user sometimes goes along well with multi-tasking, so i would like a test with a few typical applications running at the same time, like my typical config: outlook, firefox, uTorrent, excel, an antivirus and WMP.

I strongly disagree with this statement. Multi-user =/ multi-tasking, the latter of which is much more closely represented by the single-user DriveMarks than something that almost artificially keeps concurrency at levels virtually unmaintanable in a non-server environment.

Share this post


Link to post
Share on other sites

I can confirm that the 500GB model of the 7200.11 series performs equally as well as the 1TB

ST3500320AS

Share this post


Link to post
Share on other sites
Can you please run some benchmarks with NCQ OFF vs ON on this drive?

Hello :),

I did some tests witch NCQ On and Off and can say it's definitely NCQ thing :).

Below is screenshot from the first HD_Speed and in the middle of the graph I've started a second one (HDD: 7200.11 500GB, NCQ On):

HD_Speed-7200.11.png.xs.jpg

Performance with one instance: 98MB/s, with 2 instances (summary in Performance Monitor) also 98MB/s :D.

Next screenshot with the same disk and NCQ Off:

HD_Speed-7200.11_non_NCQ.png.xs.jpg

Performance with one instance: 98MB/s, with 2 instances (summary in Performance Monitor) dropped to 62MB/s.

As you can see HDD behave completely different trying to transfer data in both instances of HD_Speed at the same time, while with NCQ On it tried to do it alternately.

Someone asked also for Hitachi tests :). Below test of 7K1000 750GB with NCQ On:

HD_Speed-7K1000.png.xs.jpg

Performance with one instance: 82,8MB/s, with 2 instances (summary in Performance Monitor) 27MB/s :( and disk is behaving like 7200.11 with NCQ Off.

Edited by Szymon

Share this post


Link to post
Share on other sites
I got tired of waiting for the Samsung F1 :)

I'm still waiting :) Anyone thoughts on what to expect from ncq performance on a new Samsung drive vs the 7200.11?

Share this post


Link to post
Share on other sites

Could I expect the same performance for a 500gb version? its like 100 $ (in denmark anyway) if this is the right one "Seagate Barracuda 7200.11 500 GB"

Share this post


Link to post
Share on other sites
I can confirm that the 500GB model of the 7200.11 series performs equally as well as the 1TB

We are talking about ncq implementation here. Does this hold true for the 250GB 7200.10 ST3250410AS single-platter drive, which is atleast based on the same platter technology as the 7200.11? I don't know to what extend ncq is based on hardware or firmware. I would very much like to know the answer to this question in light of a possible purchase decision.

Share this post


Link to post
Share on other sites

No, the 250GB 7200.10 (3.AAA firmware) NCQ performance sucks ass like most drives I've tested so far, I'd wager that the 250GB 7200.11 will be fast, I'm just waiting on availability and as soon as I get one I'll run some test on it.

Also a few weeks ago I tested the drive using DC++ (LanDC++) and was serving up about 6-8 users simultaneously, all were averaging between 10-14MB/s, something the previous gen drive use to choke on, i.e. I'd be lucky you see more then 2MB/s each with that many users

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