Performance of Seagate Barracuda IV

Silencing hard drives, optical drives and other storage devices

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Performance of Seagate Barracuda IV

Post by Shmanky » Sat Aug 17, 2002 6:47 am

I own a Seagate Barracuda IV 40GB ST340016A and when I benchmark this disk using SiSoftware Sandra, I get a performance rating nowhere near the compared level of an ATA100 7200rpm disk. Are you people getting decent performance in benchmarking the Barracuda? Is the wonderful quietness of the Barracuda offset by poor performance? Are the benchmarking tests not realistic? Is the Barracuda comparable in speed to say those IBM disks? Are there drivers/tools on the Seagate website I should be using? I remember my friend telling me about his Maxtor disk having a tool that can swtich the Maxtor between two modes: "fast & loud, and quiet & slow." Is there such a thing for the Barracuda? I'm running the Barracuda on an Asus A7V mainboard, on the ATA100 IDE channel, and the operating system is Windows XP Professional, and I have the latest Via 4in1 driver set. Thanks.

Gerwin
Posts: 41
Joined: Fri Aug 16, 2002 11:35 pm
Location: Amsterdam
Contact:

Post by Gerwin » Sun Aug 18, 2002 12:24 pm

A couple of things to consider:
1. A lot if Via chipsets have a bug in the PCI bus that hampers the theoretical performance of the harddisk. VIA knows this and has acknowledged this. But since a single harddisk doesn't get near the maximum throughput of the PCI bus, this will probably not be the cause of your dissappointing performance.
2. IBM has a tool that can switch the harddisk between fast/noisy <--> slow/quiet. You will find it somewhere on www.ibm.com, support, downloads. It states that it only works with IBM disks, but this is not true, it works on all modern HD's. The advantage of this tool is that it not only lets you choose between two settings, but between about 15 combinations of fast/noisy <--> slow/quiet. Try it, there is a nice testing mode in it that lets you test the setting right away. I still don't think that it matters much, because even in the most quiet and slowest mode, the Barracuda is quite fast. Only seek times are affected by this, not max throughput, and even the seek times differ no more than 10%.
3. How did you test your HD? I hardly know of any reliable tools to test drives, and a lot of tools that pretend to test the harddrive but actually test the caching performance of windows.
4. Are you sure your harddrive is working in proper UDMA mode? If the CPU is highly loaded during copying operations, it may be in PIO mode.

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sun Aug 18, 2002 11:19 pm

2. Thanks, I'll check IBM for their tool.

3. I used a benchmarking programme called Sandra whose homepage is either www.3bsoftware.com or www.sisoftware.co.uk I'm not sure as I found it in both locations. I'm surprised you haven't heard of it, I thought it was a rather popular benchmarking application.

4. That's another good question because in Windows XP Professional my Barracuda is recognized as a SCSI disk!!! Which clearly it isn't. I think that's because of the way WinXP sees the Promise IDE controller on my Asus A7V mainboard. So in the properties of the disk I can't set or change between UDMA or PIO.

Thanks for your input Gerwin (:

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sun Aug 18, 2002 11:22 pm

Okay Gerwin you've reminded me about this SCSI issue. I've read others inquiring about it on other forums. I should persue this issue as the main cause of my performance troubles. And I'll check on other hard disk forums like www.storagereview.com as this is no longer a "silent pc" issue. Thanks again (:

quokked
Posts: 333
Joined: Sun Aug 11, 2002 3:26 pm
Location: Melbourne, Australia
Contact:

Post by quokked » Mon Aug 19, 2002 12:26 am


quokked
Posts: 333
Joined: Sun Aug 11, 2002 3:26 pm
Location: Melbourne, Australia
Contact:

Post by quokked » Mon Aug 19, 2002 12:29 am

Oh yeah and somebody send get in touch with me to send me a image of the file I don't have a floppy disk drive for it to find :roll:

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Mon Aug 19, 2002 9:22 am

Thank you quokked. I admit I am apprehensive about running this type of software on a non-IBM disk, but I will read the documentation on IBM's website to be sure.

Although I really don't understand what it is you're requesting in your second post (??) :?

quokked
Posts: 333
Joined: Sun Aug 11, 2002 3:26 pm
Location: Melbourne, Australia
Contact:

Post by quokked » Sat Oct 12, 2002 12:18 am

yeah I just read it back again and I don't know what I'm saying myself :oops:

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sat Oct 12, 2002 9:19 am

quokked wrote:yeah I just read it back again and I don't know what I'm saying myself :oops:
hehe

Queue
Posts: 61
Joined: Mon Sep 09, 2002 9:53 pm

Post by Queue » Sat Oct 12, 2002 6:58 pm

I'm setting up a box right now...
(Using Sandra 2001.5.8.11 to get benchmarks...)

relevant details:
ECS K7S5A MOBO (SiS 735 Chipset, ATA100 onboard Controller)
512 MB PC2700 DDR (Crucial)
AMD 1600+ XP CPU
Win Me

In it are 2 HDDs, both at default AAM settings (which I assume to be "fast and loud")
'Cuda IV 40GB
Outer 20GB partition Sandra Drive Index=27539
Inner 20GB partition Sandra Drive Index=24294

Maxtor D740x 6L040L2 (The FDB model, ATA133)
Outer 20GB partition Sandra Drive Index=27439
Inner 20GB partition Sandra Drive Index=24695

Sandra's Reference ATA U100 7200rpm 30GB Sandra Drive Index=24000

Some tweaks I've done to boost disk performance:
You may notice the outer partitions outperform the inner partitions. I use the outers for disk intense apps, inners for backups/storage.
All partitions were formatted with larger sectors per cluster size switch /z:64
The Typical Role is set to "network server" not "desktop"

Queue

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sat Oct 12, 2002 10:46 pm

Queue wrote:I'm setting up a box right now...
My comments on your experiment, Queue....

I find that Windows ME always outperforms Windows XP.. on virtually all fronts, whether that be disk access or Direct 3D rendering. So I expect you would get lower results on an NT based machine.

I am *fairly certain* that by default the Seagate Barracuda IV ships with AAM set to SLOW AND QUIET, not fast and loud. That's how mine shipped, and that's what Red Dawn claims in this thread.

A score of 24,000+ is absolutely astounding! I am getting a mere ~18,000 in loud mode, on Windows XP. I got about the same value when I was running the disk in Windows ME in slow mode. How the heck did you get 24,000+? Could it be that you have double the ram I do? Here are my specs:

Asus A7V mainboard (ATA100 Promise onboard controller)
Athlon 1.0 @ 1.0GHz
256 MB PC133 SDRam (NEC)
- up to date BIOS, latest Via 4-in-1's, et cetera.

I have never heard someone say that the physical location of the partition on the disk makes a difference in performance. But it's something I'll now consider. Although I don't know how you would go about configuring that with FDISK or the NT setup :?: And that "larger sectors per cluster size" idea is new to me also.

Thanks for your report, Queue.

Red Dawn
Posts: 169
Joined: Fri Oct 04, 2002 11:46 am
Location: Stockholm

Post by Red Dawn » Sun Oct 13, 2002 3:26 am

Having large sector sizes can be beneficial to performance, but can also be it's achilles heel. If you have a large number of small files, for example word files; it would only eat away large chunks of your hd and would not always boost performance, however larger files should benefit a great deal from this.

For each bit of data it takes up the same formatted size (64 in this case), whether it would actually be that size or not, so clearly, it might not always be wise. Not trying to critize you or anything, just stating what you probably already know. :)

Queue
Posts: 61
Joined: Mon Sep 09, 2002 9:53 pm

Post by Queue » Sun Oct 13, 2002 8:59 am

Because I'm Mr. Drive Bay...

I popped the 'Cuda in a similar box (it has a AMD XP 1700+, and a slightly different model of the K7S5A) that has XP PRO OS.

Partition......ME.........XPPro
Outer........27539......27609
Inner........24294......24370

Then I found that you could check/control AAM through DTemp. You and Red Dawn were correct about the default AAM setting (if I understand Dtemp's interface.) So I cranked the thing to "Fast and Loud" and retested.

Partition......ME
Outer........27780
Inner........24516

I don't think Dtemp can get to AAM in XP?

In response to some questions/statements:
"ME always outperforms XP" maybe that should read Fat32 outperforms NTFS?

Red Dawn's comments about sector size are absolutely correct, but I tend to have large audio/video files on these HDDs, so I've got them optimized for such.

"How did you get 24,000+?" There could be a number of factors at play here. First I tend to lean toward drive performance when optimizing. Second, chipset. The ECS K7S5A performs quite well in this category. Third, what version of Sandra are you using?

Lastly, on the Outer vs. Inner Partition performance. It is true, (and proven by my results). Think of it this way: More distance is covered taking one lap around the outer edge of a HDD than one lap around the inner edge. Therefore, you can pick up more data during the same lap on the outer track than the inner track.

Queue

Red Dawn
Posts: 169
Joined: Fri Oct 04, 2002 11:46 am
Location: Stockholm

Post by Red Dawn » Sun Oct 13, 2002 9:29 am

About the AAM issue in WinXP/Win2k using dtemp: in win9x (and WinMe) it uses 'smartvsd' to control AAM, in xp and 2k however that is disabled, you'll have to go the 'normal' route of running ibm feature tool on a boot disk, or some third party software designed to alter the AAM.

While you can't change the AAM values in xp/2k, you can view them using dtemp, under drive features/automatic acoustic management/ you can find the value, or at least a rough estimate (mine says "Current AAM value: 80h (minimum acoustic emanation level)").

Very nifty program I must say, I find it a very useful tool, especially when trying out new cooling techniques.

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sun Oct 13, 2002 12:20 pm

Queue wrote: In response to some questions/statements:
"ME always outperforms XP" maybe that should read Fat32 outperforms NTFS?
Well I've installed Windows XP on a FAT32 partition before and I still could not get as high a framerate in Unreal Tournament as I could when playing in Windows ME.
Queue wrote:
"How did you get 24,000+?" There could be a number of factors at play here. First I tend to lean toward drive performance when optimizing. Second, chipset. The ECS K7S5A performs quite well in this category. Third, what version of Sandra are you using?
I was using the latest version of Sandra as of writing, which I got from download.com and is titled SANDRA Standard 2002.68.97. You seem to be an older version. I don't know why you're using an older one. Perhaps you paid for that one. I was using the free/shareware demo. But my version still has the same ATA100 7200rpm = 24,000 reference.

Oh well, I'll just assume it's your chipset that's the winner in this case.

Also, in my system, my Barracuda IV 40GB disk appears in the Device Manager as a SCSI disk. This is clearly not true. Is it just because this is how Windows XP sees my onboard Promise ATA100 controller? Could there be a performance hit with this? I believe in WinME it was recognized as a normal ATA100 disk.

Shmanky

Red Dawn
Posts: 169
Joined: Fri Oct 04, 2002 11:46 am
Location: Stockholm

Post by Red Dawn » Sun Oct 13, 2002 2:52 pm

Schmanky: Put simply, framerate and drive performance aren't necessarily one and the same. Load times may differ if you use a better drive, but in general, fps is not affected, that is directly related to the work your cpu in conjunction with your gfx card can handle (at least when playing at low resolutions, at high resolutions, the gfx card takes over most of the processing).

I also have the Asus A7V, such as yourself, and while I'm unsure whether or not you would suffer from a performance hit using the pdc20265 promise ultra100 ide controller or not, I can honestly say that if such were the case, at least I haven't felt any decreased performance in my system. :)

I don't really know why they have the whole device (the controller, plus the drives) showing up as scsi, I just know they do, and that's fine with me, as long as it works, heck - maybe it gives you better performance, who knows?
The only hassle with it are during installs such as win2k where you have to have a floppy ready with all the drivers during the first install phase (and you have about 1-2 seconds to press down a button to let it install a scsi interface), but that's basically it. If you're getting bad results, it could be because of the drivers for that specific chip, I use the latest I could find (which are 1.60 build 33). When A7V was still active, ie. being sold, there was talk that burst performance of the versions that came bundled with the motherboards were poor, and pretty soon thereafter another version came out. At least it wouldn't hurt to try if you haven't already! :-)

Hope that clears a few things up.

PS. I would conduct benchmarking on my seagate, but it's a 2-platter version (60gb) and would therefore not be representative for your little 'investigation'. :) DS.

Shmanky
Posts: 54
Joined: Tue Aug 13, 2002 5:25 pm
Location: Ontario, Canada
Contact:

Post by Shmanky » Sun Oct 13, 2002 10:38 pm

Red Dawn wrote:Schmanky: Put simply, framerate and drive performance aren't necessarily one and the same.
Sorry, I did not mean to suggest that Direct 3D rendering was related to hard disk performance. Yes it's mostly in the GPU. I was talking about the difference between how the NT and 9x kernels allow access to video hardware for rendering. But that's beyond the scope of this thread so we won't get too much into that.
Red Dawn wrote:The only hassle with it are during installs such as win2k where you have to have a floppy ready with all the drivers during the first install phase
Yup I've gone through that 2k install. Fortunately XP has the promise driver on the CD. Similarly the recent releases of various distributions of Linux are able to detect Promise controller disks! Fhew!! I can't imagine having to compile my own ATA100 driver into a Linux kernel. My eyes would go cross-eyed :shock:
Red Dawn wrote:I also have the Asus A7V, such as yourself
Okay if you don't find it a problem having your ATA disk show up as SCSI... then I won't worry about it either.
Red Dawn wrote:PS. I would conduct benchmarking on my seagate, but it's a 2-platter version (60gb) and would therefore not be representative for your little 'investigation'. :) DS.
Well, yes I would be interested in seeing performance results across the entire Barracuda IV line, with differing size and numbers of platters. I forget, does more platters mean better or worse access times?

Red Dawn
Posts: 169
Joined: Fri Oct 04, 2002 11:46 am
Location: Stockholm

Post by Red Dawn » Mon Oct 14, 2002 1:43 am

Yes, there is that difference between 9x and NT kernels, and their access to video hardware for rendering, but as you said, that's a bit off topic here. ;-)

And I will benchmark it as soon as I un-lazy myself and plug it in, I'm feeling sooo lazy today it's incredible! :D
Don't know about the access times though, we'll just have to find out, won't we?

Storm
*Lifetime Patron*
Posts: 81
Joined: Tue Aug 27, 2002 2:10 pm
Location: UK, Northern Ireland

Post by Storm » Tue Oct 15, 2002 8:48 am

Detailed review of the barracuda V out now at SR
http://www.storagereview.com/welcome.pl ... 23A_1.html
Its quieter, but barely faster than the IV and at times slower.

Post Reply