Sign in with
Sign up | Sign in
Your question

Two identical drives....very different benchmarks.

Last response: in Storage
Share
February 14, 2012 4:56:35 AM

Hello all, I'm new here. I've got two WDC3200AAKX-001CA hard drives in my rig, both bought at the same time. One is running Windows 7 and formatted with NTFS, the other is running Xubuntu and is formatted with ext4. I have benchmarked both drives in Windows with HDTune, and in linux with DIsk Utility, and gotten very similar results with both programs.

The problem is that one drive (the one with linux on it) benchmarks significantly slower, both in data transfer and in access time. I have checked the s.m.a.r.t. attributes on both drives, everything is good. Then I ran a scan for bad sectors and came up with zero. I also notice that the graphs for data transfer for the two drives are completely different, the fast drive drops very little toward the end, while the other one drops quite a bit. The DCM number on the fast drive is EBNNHTJCH, and the DCM of the slow one is DHNNHVJCH...I know this means there are some differences, but what would they be? Is there maybe a different number of heads or platters? Even if that is the case, the 15.2ms access time seems horrid. I've got a 5.5 year old Samsung SP2504C in this same rig, formatted with ext4, and its access time is 14.3, though its data transfer rate is significantly slower than the slower WD drive.

Could there be something physically wrong with the drive? I'm OK with it being a little slower, that isn't a huge deal, but I do want to know if its possibly a future problem child?

Here are the graphs of the two drives:

Fast Drive http://imgur.com/kLo84

Slow Drive http://imgur.com/j3P7j

And just for comparison, heres the old Samsung http://imgur.com/pu9Z4
February 14, 2012 5:38:41 AM

It's not easy to tell since they have different operating systems installed, and they're not benchmarked by the same software doing exactly the same tests. If you want to be sure, either put Windows 7 on the second disc as well so you can run identical benchmarks on identical OS, or switch OSs alltogether and rerun the benchmarks. If the Linux drive is still slower, it's a file system/OS/benchmark issue. If the Windows Drive is now slower, it could be a drive issue.

Your graphs show almost identical maximum speed, but the two drives just seem to behave differently, wich could be a software issue. Finally, make sure both drives have the same firmware installed.

February 14, 2012 6:23:23 AM

Thing is, I did run tests from both systems. I used HD Tune to check both the Windows drive, and the linux drive, and likewise used Disk Utility to check the linux drive, and the Windows drive. I guess what concerns me most is the difference in access times, and the fact that my "slow" drive is being beat out in that respect by an old, otherwise much slower drive that has the same filesystem on it. Does the filesystem on a drive even make a difference to a drive benchmark like HD Tune? Both drives have the same firmware installed, 15.01H51.
Related resources
February 15, 2012 2:55:02 AM

So....after I got home today I formatted the slow drive with NTFS and tested in Windows with HD Tune. When I did, the results came back exactly the same, except my access time actually got worse, all the way up to 15.7ms, so it looks like the filesystem/OS has nothing to do with it. Any ideas?
a c 155 G Storage
February 15, 2012 3:33:18 AM

I once picked up a few WDC drives and one had higher access times because a feature called AAM was on. Makes the drive more quiet at the cost of access times.

The one drive with fast access times looks very strange to be honest, most drives start fast and slowly drop in speed. The only time i have seen like your fast drive was with a drive short stroked, but that does not to seem to be setup for your drive.

Is the info tab the same for both drives as well? firmware version ect?
February 15, 2012 5:26:46 AM

I thought it was a little odd that the fast drive has such a flat graph too, but could that even be a bad thing? The drive seems to perform well.

As for the AAM thing...is there a way to tell if its on and if so can I turn it off? The info on the two drives is identical.
February 15, 2012 6:13:01 AM

tylere1228 said:
I thought it was a little odd that the fast drive has such a flat graph too, but could that even be a bad thing? The drive seems to perform well.

As for the AAM thing...is there a way to tell if its on and if so can I turn it off? The info on the two drives is identical.


You can check and control the AAM/APM status with this:

http://crystalmark.info/software/CrystalDiskInfo/manual...

On or off, it won't hurt your drive. AAM is just used to minimize audible noise from the drive.
February 15, 2012 6:53:47 AM

I downloaded and the crystalmark program, and it looks like I can't change the AAM/APM settings on either one of my WD drives, the slider is greyed out, so I suppose that's a dead end. I've looked everywhere and cant figure what the difference could be, everything seems to check out the way it should. I've looked over the BIOS, there's nothing there to change (gigabyte neutered this one). Both drives are running on the same SATA controller. I've tried different cables even...I'm kinda at a loss with it. Thanks for the help so far guys.
February 15, 2012 8:45:50 AM

I think there is an "enable" button on that menu you have to click to be able to adjust the slider. It's been a long time since I used it so I don't remember exactly.

What about the status? Do you se wether AAM and APM are enabled or disabled on both drives?
a c 155 G Storage
February 15, 2012 12:43:36 PM

HD tunes info tab should also tell you if AAM is on or off. It should show it as check marks in the boxes
February 15, 2012 6:15:41 PM

I tried the enable button, i fiddled with it for a while, seems like its pretty much all around disabled. On the main screen of the program it shows them both greyed out. Also, HD Tune doesn't tell me anything about either of the drives, I had to get the info on them from the linux Disk Utility program. I'm not sure why....maybe its the chipset drivers I have installed? Its the nForce 630a chipset, and I'm using Nvidia's drivers for it. Also, I'm using the free version of HD Tune, if that matters any.
a c 155 G Storage
February 15, 2012 10:49:37 PM

If HD Tune does not show info, it is most likely as you say because of the drivers or SATA mode.

For instance, my media center has very little in the way of info in HD Tune, while my other system shows everything.

When i disabled AAC, i did it with hitach feature tool. It was a CD i burned and used to make the change. This was a long time ago, but should still work with today's drives.

With most of this software, you need to set your sata ports to IDE mode(just while using the bootable cd/dvd) for the software to see it right.

The strange thing is AAM does not effect actual speed too much from what i remember.

Toms did an article on AAM a while back

http://www.tomshardware.com/reviews/terabyte-hitachi-ac...
a c 302 G Storage
February 17, 2012 7:57:42 AM

Here are direct links to your images:

http://i.imgur.com/kLo84.png
http://i.imgur.com/j3P7j.png
http://i.imgur.com/pu9Z4.png

AISI, the "slow" WD drive and the Samsung drive both show the typical 2:1 ratio between the transfer rates at the outermost and innermost zones. This is to be expected since the diameter (and circumference) of the outer and inner tracks is in the same ratio. That is, the innermost zone has half as many bits per track as the outermost zone, and therefore half as many bytes pass under the head during each revolution. The WD graph also shows 16 or so steps corresponding to each zone.

If you now compare the "fast" WD graph against the slow one, it appears that there may be only two or three zones. In fact the transfer rate at the 100% mark on the fast graph corresponds to the 35% mark on the slow graph. This suggests that the fast drive has been shortstroked, and that its full-stroke capacity is more like 914GB (= 320GB/0.35). Therefore I suspect that the fast 320GB drive is actually a 1TB drive that has had its capacity truncated. Shortstroking would also explain why its access time is significantly better.

Another clue as to what may be going on is the DCM (Drive Configuration Matrix). The following thread would suggest that there is a difference in the head stacks of the two WD drives:
http://forum.hddguru.com/western-digital-what-dcm-t6488...

I would think that the 1TB drive would have 3 platters and 6 heads, whereas the 320GB drive would have 1 platter and 2 heads.

One final observation I'd like to make is that the maximum sustained transfer rate (105 MB/s) is a lot less than what is touted in the datasheet (126 MB/s).

http://products.wdc.com/Library/SpecSheet/ENG/2879-7012...

In fact 126 MB/s is what you would expect from a 7200 RPM drive with 500GB platters. Instead your drives appear to have 350GB platters (333GB ???).

((105 / 126)^2) x 500 = 347
a c 155 G Storage
February 17, 2012 12:32:57 PM

I have ran many a drive short stroked(my first post mentioned that possibility), the strange thing is that they both have the same model number as well.

short stroke setup from 2009 :) 


Has WDC ever sold one in that config?
a c 302 G Storage
February 17, 2012 4:58:47 PM

"This file was removed due to violation of ImageShack's terms of service."
a c 155 G Storage
February 17, 2012 5:47:26 PM

hmmmmm. imageshack has honestly lost lots of images(worse part is they are part of how to guides) over the years and now is going to charge for anything over 500 images anyway.



Uploaded with ImageShack.us
a c 302 G Storage
February 17, 2012 6:05:32 PM

I'm suggesting that WD may have provided a 1TB drive in place of a 320GB, possibly as a consequence of supply issues, and then recertified it at the factory as a WDC3200AAKX-001CA.

I have seen another thread where the performance of a WD 500GB drive was shown to be consistent with a shortstroked 750GB, so it would appear that this practice is not unusual.

See http://www.tomshardware.com/forum/261937-32-wd5001aals-...
February 17, 2012 7:29:32 PM

Hmm.....so you're saying my "fast" drive is the oddball here? If thats the case, I'm OK with that :)  The high access time on the other one still seems odd though...
a c 302 G Storage
February 17, 2012 8:15:26 PM

Access time = seek time + rotational latency

Average rotational latency = 4.16 ms for a 7200 RPM drive (8.33ms per revolution)

ISTM that your seek time is reasonable for a 3.5" drive.

a c 155 G Storage
February 17, 2012 8:49:18 PM

Well i wish they gave me one of those as a replacment :) 

My WDC black 640 was replaced with a 1tb blue. And while faster then the 640 gig black, it did not fit back into the array.

The fast one does seem to be the odd ball. But in a good way.
!