The Test

In our last few hard drive reviews, we used an Intel based platform that helped eliminate any bottlenecks aside from the hard disk itself. We have since then decided to focus our attention on drive performance on an AMD platform. Again, we chose hardware that would help eliminate any bottlenecks from the CPU, memory, and GPU, and shift them directly to the hard disk drives that we have benchmarked.

Our test bed:

AMD Athlon64 3500+ (2.2Ghz)
Giga-byte GA-K8NXP-SLI
NVIDIA 6600GT SLI Edition (single 128MB card)
1GB (512MBx2) Corsair XMS4400

Our motherboard is an nForce4 based board that features support for the SATA II standard, up to 3Gbps/sec SATA transfer rates, and NCQ and TCQ.

We used the following nForce platform drivers in conjunction with our testbed:

nForce4 Chipset Driver 6.31
NVIDIA Forceware 71.89
Windows XP SP1 w/out further updates

Business Winstone IPEAK - a playback test of all of the IO operations that occur within Business Winstone 2004.
Content Creation IPEAK - a playback test of all of the IO operations that occur within Multimedia Content Creation Winstone 2004.
SYSMark 2004 - the official SYSMark 2004 test suite.
Business Winstone 2004 - the official Business Winstone 2004 test suite.
Multimedia Content Creation Winstone 2004 - the official Multimedia Content Creation Winstone 2004 test suite.
Half-Life 2 Level Load Test - a timed test of loading a level in Half-Life 2.
Doom 3 Level Load Test - a timed test of loading a level in Doom 3.
Command & Conquer: Generals Level Load Test - a timed test of loading a level in Command & Conquer: Generals.
Real World File System Task Tests - timed tests of basic file system tasks including zipping/unzipping and copying files.
Service Time and Transfer Rate Tests - Synthetic tests for average service time and transfer rate of hard disk at the beginning and the end of a full disk read.
Business Winstone 2004 Multitasking Test - Synthetic tests for overall system multitasking performance.
Real World Multitasking Test - timed tests of basic multitasking processes, timing a file zip operation while importing Outlook data.

More details about each individual test will appear in the section of the review dedicated to that particular test.

The Competition


Max Capacity
Platter Density
# of Platters
Spindle speed
(RPM)
Average Seek Time
Average Latency
Interface
Buffer Size
Seagate 7200.8
(NCQ)
400GB
133GB
3
7200
8ms
4.16ms
SATA
8MB
Seagate 7200.7
120GB
80GB
2
7200
8.5ms
4.16ms
SATA
8MB
Hitachi 7K400
400GB
80GB
5
7200
8.5ms
4.17ms
SATA
8MB
Maxtor DiamondMax 10
(NCQ)
300GB
60GB
5
7200
9.0ms
4.17ms
SATA
16MB
Western Digital Raptor 740
74GB
37GB
2
10,000
4.5ms
2.99ms
SATA
8MB
Samsung SpinPoint SP1614C
160GB
80GB
2
7200
8.9ms
4.17ms
SATA
8MB
Samsung SpinPoint SP1614N
160GB
80GB
2
7200
8.9ms
4.17ms
PATA
8MB
Samsung SpinPoint SP1604N
160GB
80GB
2
7200
8.9ms
4.17ms
PATA
2MB
Maxtor DiamondMax Plus 9
160GB
80GB
2
7200
9.3ms
4.17ms
PATA
2MB
Maxtor DiamondMax 16
160GB
80GB
2
5400
12.5ms
4.17ms
PATA
2MB

Hard Drive Buffer: Does Size Really Matter? Pure Hard Disk Performance - IPEAK
Comments Locked

44 Comments

View All Comments

  • PuravSanghani - Wednesday, April 20, 2005 - link

    mjz5: With our nForce4 platform there is an option under the drive controllers options tab called "Enable command queuing". By checking this option and restarting the system, command queuing will be enabled. Some boards, however, enable NCQ/TCQ by default through the BIOS. You may want to check with your motherboard manual on that.

    Take care,

    Purav
  • mjz5 - Wednesday, April 20, 2005 - link

    Nighteye2 has a good question. How does NCQ work with RAID arrays? Is it better, worse???

    How would I know if TCQ is enabled on my 74 raptor?
  • xsilver - Wednesday, April 20, 2005 - link

    #21 LOL --- you wouldnt want that space anyways even if it was there.... its cant be guaranteed reliable so would you trust 100gb's of your drive that could die at any moment???
  • quorm - Wednesday, April 20, 2005 - link

    I have one of the 300gb 7200.8 drives. It's mentioned in the article that all of the 7200.8 drives use a 3x133gb platter configuration. I was wondering if there is any hack to allow access to the remaining 100gb of disk space. Anyone?
  • AtaStrumf - Wednesday, April 20, 2005 - link

    Hey, where did all the WD drives (apart from Raptor obviously) go??? I can get a 200 GB PATA model pretty cheap, so I'm seriously considering it. Any advice anyone?
  • n7 - Wednesday, April 20, 2005 - link

    Thanx for the review guys :)

    flatblastard: I'd agree.

    The Raptors may not win all the benches, but i find they feel so much snappier than my other 7200RPM drives.

    I certainly wouldn't mind adding a 400 Gb Seagate to my collection though :)

  • bob661 - Wednesday, April 20, 2005 - link

    Can you guys post a UT2004 for load time graph please.
  • flatblastard - Wednesday, April 20, 2005 - link

    I'm using the raptor for my OS, and the 250GB seagate 7200.8 for everything else. I really can't tell which one is faster at loading games...but the raptor is MUCH quicker loading anything else.
  • Icehawk - Wednesday, April 20, 2005 - link

    Where were the heavier real-world multi-tasking tests like in the Intel DC previews? In those articles it appeared that NCQ offered some performance boost in heavy I/O situations - here it seems to offer zero benefit.
  • Houdani - Wednesday, April 20, 2005 - link

    I dunno. Neither the Seagate nor the Maxtor NCQ drive really impressed me. They didn't stand out from the peleton. For most performance needs, I'd have to give the yellow jersey to the Raptor, although the idle heat is a noteworthy ding.

    For extra capacity one of the larger models would be prudent, but for a primary drive the Raptor is fairly impressive.

Log in

Don't have an account? Sign up now