Maxtor’s DiamondMax 10 hard drive

Thankfully, Native Command Queuing (NCQ) brings a little spice to the otherwise drab world of desktop hard drives. NCQ intelligently re-orders I/O requests to minimize the performance penalty associated with a drive’s rotational latency. That’s not the sexiest sales pitch, but for PC enthusiasts, NCQ’s potential performance benefits are certainly tantalizing.

Maxtor’s first NCQ-enabled desktop drive is the DiamondMax 10. Available in capacities up to 300GB, the DiamondMax 10 comes equipped with as much as 16MB of cache and the promise of improved performance. Read on to see how command queuing affects the DiamondMax 10’s performance and how the DiamondMax 10 stacks up against the competition.

The specs
Before we tackle the DiamondMax 10’s performance, let’s see how the drive’s specs compare to its predecessor, the DiamondMax Plus 9.

Like its predecessor, the DiamondMax 10 has a 150MB/sec Serial ATA interface and spins at 7,200 RPM. The drive’s seek time and rotational latency are a little faster than the Plus 9, but not by much. The DiamondMax 10 does manage to pack an additional 20GB or storage per platter, though.

Additional cache is one of the DiamondMax 10’s more notable features, but it’s limited to the largest drives. 250 and 300GB DiamondMax 10 drives come with 16MB of cache—twice what’s available with 80, 120, 160, and 200GB drives.

The DiamondMax 10’s most intriguing new feature is support for Native Command Queuing (NCQ). NCQ minimizes the performance impact of a hard drive’s mechanical latencies by queuing I/O requests and intelligently executing them in a more optimal order. Command queuing can improve performance in multi-user environments and with randomized access patterns, but it doesn’t do much for streaming transfers.

Maxtor’s warranty policy for DiamondMax drives is a little odd. Units sold as bare drives are covered by a three-year warranty, but those sold in retail packaging only get a one-year pact. It seems counter-intuitive that the retail drive kits would get less warranty coverage than bare drive equivalents, but there’s no reason for enthusiasts to buy more expensive retail kits, so I can’t complain. Perhaps Maxtor is only serving up extra warranty coverage for bare drives because it knows that those who buy bare drives are conditioned to expect better warranty coverage. It’s unfortunate that the unwashed masses who are buying retail drive kits get shafted with a single-year warranty, though.

Although it’s nice to see bare drives getting a little extra warranty love, three years isn’t all that special. Seagate covers all its desktop hard drives with a five-year warranty, which equals the design life of the DiamondMax 10. Maxtor offers a five-year warranty on its higher end MaXLine “Enhanced-reliability” drives, though.

The DiamondMax 10 from above…

And below

You’ll need a Serial ATA power connector for this one

Test notes
I elected to use Promise’s FastTrak TX4200 Serial ATA RAID card on our test platform because it supports both Native Command Queuing and Tagged Command Queuing (TCQ). Although the only drive that currently supports TCQ is Western Digital’s Raptor WD740GD, the drive is widely available and reasonably popular among enthusiasts. NCQ is definitely the future of Serial ATA command queuing, but TCQ is certainly an alternative for now, albeit a considerably more exclusive one.

To contrast NCQ and TCQ’s impact on performance, I tested the DiamondMax 10, 7200.7, and Raptor with command queuing enabled and disabled. This should give us a good idea of where command queuing matters and where it doesn’t for each drive.

Finally, I should preempt purists who will no doubt be perturbed that I’m testing 7,200-RPM desktop drives against 10K-RPM Raptors. The Raptors are technically enterprise-class drives, while the 7,200-RPM the drives are desktop products, and marketing types don’t like see the two thrown in the ring together. However, arbitrary labels have never stopped PC enthusiasts from running enterprise-class hardware in their desktop PCs, so they’re certainly not going to stop me. I’ve even thrown in a 10K-RPM SCSI drive into some of our tests to illustrate how Native and Tagged Command Queuing compare with SCSI’s command queuing. The horror!

Our testing methods
All tests were run three times, and their results were averaged, using the following test systems.

The Serial ATA and SCSI cards were used in different PCI-X slots on different PCI-X busses. Each card had an entire PCI-X bus to itself, so bandwidth sharing shouldn’t be an issue.

A special thanks goes out to the Computer Repair Shop and KickAss Gear for kicking in the Western Digital Raptor WD360GD and WD740GD we used for testing.

Also, all of our test systems were powered by OCZ PowerStream power supply units. The PowerStream was one of our Editor’s Choice winners in our latest PSU round-up.

We used the following versions of our test applications:

• ZD Media Business Winstone 2004 1.0.1
• ZD Media Content Creation Winstone 2004 1.0.1
• Xmpeg 5.03 with DivX Video 5.11
• ATTO ExpressPCI Utilities v163
• Intel IOMeter v2004.07.30
• Xbit Labs File Copy Test v1.0 beta 13
• TCD Labs HD Tach v3.01
• Far Cry v1.2
• DOOM 3

The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at a 75Hz screen refresh rate. Vertical refresh sync (vsync) was disabled for all tests. All of the 3D gaming tests used the high detail image quality settings.

All the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

HD Tach
We tested HD Tach with the benchmark’s full variable zone size setting.

Although command queuing can’t help the DiamondMax 10’s performance in HD Tach’s streaming transfer rate tests, the drive is pretty speedy considering its 7,200-RPM spindle speed. The drive’s transfer rates are consistently faster than Seagate’s Barracuda 7200.7 and a huge improvement over the old DiamondMax Plus 9.

Despite being faster than the 7200.7 in HD Tach’s transfer rate tests, the DiamondMax 10’s random access time is more than half a millisecond slower than the Seagate drive’s. CPU utilization is pretty even across the board, though; all results are within HD Tach’s +/- 2% margin for error.

IOMeter – Transaction rate
To compare how TCQ and NCQ scale with SCSI command queuing, I’ve included scores from a Maxtor Atlas 10K V SCSI drive in our IOMeter results. The FastTrak TX4200 is only validated for loads up to 128 outstanding I/Os and it generated errors in IOMeter with a load of 256 outstanding I/Os.

The DiamondMax 10’s IOMeter transaction rates are a little disappointing. With NCQ enabled, the Maxtor drive’s performance lags that of the Barracuda 7200.7, especially with the web server test pattern. Without native command queuing, the DiamondMax 10’s performance is much slower than the 7200.7.

Of course, none of our 7,200-RPM drives can match the performance of our 10K-RPM Raptors in IOMeter. Also notice how much more aggressively our 10K-RPM SCSI drive’s performance ramps as the load increases—that’s a mature command queuing implementation in action.

IOMeter – Response time

IOMeter response time results show the DiamondMax 10 behind the Barracuda again, both with and without NCQ. As we saw with the transaction rate results, the DiamondMax 10 falls even further behind when command queuing is disabled.

IOMeter – CPU utilization

With NCQ disabled, the DiamondMax 10’s CPU utilization in IOMeter doesn’t distinguish itself from the 7200.7. However, when we turn command queuing on, the Barracuda’s CPU utilization climbs, while the DiamondMax 10’s remains consistently low through three of four test patterns. The DiamondMax 10 doesn’t have a CPU utilization advantage over the 7200.7 with the web server test pattern, which is dominated by read operations.

Of course, none of the drives even comes close to the Raptor WD740GD’s horrendous CPU utilization with Tagged Command Queuing (TCQ) enabled.

ATTO
ATTO tests read and write performance with various transfer sizes. We tested the drives with 1MB transfers that are well within their caches and with 32MB transfers that are not.

The DiamondMax 10’s 1MB ATTO write speeds are much quicker than the competition, but read performance is considerably more pedestrian. Note that in the write test, the DiamondMax 10’s performance doesn’t suffer with NCQ enabled—that’s not the case with the Barracuda 7200.7, which is much slower with NCQ than without in this test.

Moving to 32MB transfers, the DiamondMax 10’s read and write transfer speeds trail only the Western Digital Raptor WD740GD. For a 7,200-RPM desktop drive to keep pace this well with a 10K-RPM enterprise drive is impressive indeed.

File Copy Test
File Copy Test is a pseudo-real-world benchmark that times how long it takes to create, read, and copy files in various test patterns. File copying is tested twice: once with the source and target on the same partition, and once with the target on a separate partition. Scores are presented in MB/sec.

The DiamondMax 10’s file creation speeds are second only to the Raptor WD740GD, which is pretty good company to keep. Notice how little impact NCQ has on the DiamondMax 10’s performance; that’s not the case with the Barracuda 7200.7, which is much slower with NCQ enabled than without.

FC-Test’s read speed tests also show the DiamondMax 10 in a favorable light. Again, the drive keeps pace with the fastest Raptor and is largely faster than the Barracuda 7200.7.

The results of FC-Test’s file copy tests are a little more mixed for the DiamondMax 10. The drive’s performance goes from just trailing the fastest Raptor to being slower than the 7200.7 to beating the entire field, depending on which test pattern you’re looking at. The DiamondMax 10’s huge performance advantage over the DiamondMax Plus 9 is consistent throughout, though.

FC-Test’s partition-to-partition copy results closely mirror what we saw with the standard file copy test. The DiamondMax 10 is ahead of the 7200.7 with some test patterns and behind it with others.

WorldBench overall performance
WorldBench uses scripting to step through a series of tasks in common Windows applications and produces an overall score. More impressively, WorldBench spits out individual results for its component application tests, allowing us to compare performance in each. We’ll look at the overall score, and then we’ll show individual application results alongside the results from some of our own application tests.

The DiamondMax 10 is the only drive in the field to see a performance benefit from enabling NCQ in WorldBench. In fact, the 7200.7 is actually slower with command queuing enabled. Hard drive choice doesn’t influence WorldBench’s overall score much, but let’s explore WorldBench’s individual tests to see where NCQ gives the DiamondMax 10 an edge.

Multimedia editing and encoding

MusicMatch Jukebox

Xmpeg DivX encoding
We used the default settings for the DivX codec to encode a 1000-frame sequence from a DVD-formatted MPEG2 source file.

Windows Media Encoder

Adobe Premiere

VideoWave Movie Creator

NCQ saves the DiamondMax 10 a few seconds in Premiere and gains the drive extra half a frame per second in Xmpeg, but that’s about it for our multimedia apps. Notice that the Barracuda 7200.7 is actually much slower in Premiere with NCQ enabled.

Image processing

Adobe Photoshop

ACDSee PowerPack

Neither Photoshop nor ACDSee benefits from NCQ with the DiamondMax 10. Unlike the 7200.7, which is slower in ACDSee with NCQ than without, the DiamondMax 10’s performance is consistent regardless of whether command queuing is enabled.

picCOLOR image processing
Dr. Reinert H. G. Mueller has a hard drive benchmark built into his picCOLOR image analysis tool. The disk benchmark writes the contents of a 16MB test pattern buffer eight times to create a 128MB file on the disk. Performance is measured in MB/sec.

picCOLOR’s hard drive benchmark shows the DiamondMax 10 just trailing the fastest Raptor and well ahead of the Barracuda 7200.7. Again, notice that the DiamondMax 10 is the only drive to perform better with command queuing than without.

Multitasking and office applications

Microsoft Office

Mozilla

Mozilla and Windows Media Encoder

NCQ doesn’t do much for the DiamondMax 10 in our selection of office and multitasking tests.

Other applications

Winzip

Nero

The DiamondMax 10 is a little sluggish in Winzip and Nero, at least when compared with the Barracuda 7200.7. Still, the DiamondMax 10 enjoys a clear performance boost from command queuing where other drives do not.

Boot and load times
To give you an idea of how much—or how little—hard drives can affect boot and load times, I busted out my trusty stopwatch.

Although its boot time is a few seconds off most of the field, the DiamondMax 10 continues to enjoy a performance benefit from NCQ. At least the drive’s much faster to boot from than its predecessor, the DiamondMax Plus 9.

The DiamondMax 10 performs well in our level load tests, but with the exception of the DiamondMax Plus 9, all the drives are pretty close.

Noise levels
Noise levels were measured with an Extech 407727 Digital Sound Level meter 1″ from the drives at idle and under an IOMeter file server test pattern load.

While not quite as silent as the Barracuda 7200.7, the DiamondMax 10’s noise levels are competitively low.

Power consumption
Power consumption was measured for the entire system, sans monitor, at the outlet. I used the same idle and load environments as the noise level tests.

The DiamondMax 10’s power consumption is a little higher than some of the other dives, but only by a few watts.

Conclusions
Perhaps the most striking thing about the DiamondMax 10 is that it doesn’t suffer from the same command queuing-induced performance penalty in some applications as the Barracuda 7200.7 and Raptor WD740GD do on our test platform. The DiamondMax is able to overcome whatever’s slowing the Barracuda and Raptor down in tests like WorldBench, and I suspect that the drive’s larger 16MB cache may be at least somewhat responsible.

Apart from its consistently amicable relationship with command queuing, the DiamondMax 10’s strengths clearly lie with common desktop applications and streaming transfers. Despite its 7,200-RPM spindle speed, the drive’s transfer rates are very impressive and generally faster than the Barracuda 7200.7’s, particularly in FC-Test.

Despite its strong desktop application performance and speedy transfer rates, the DiamondMax 10’s performance in IOMeter is uninspiring. IOMeter is a veritable playground for command queuing, and while the DiamondMax 10’s performance improves with NCQ enabled, the drive can’t keep up with the Barracuda 7200.7’s transaction rates or response times. To Maxtor’s credit, the DiamondMax 10 consumes fewer CPU cycles than the Barracuda 7200.7 when command queuing is enabled, but the I/O performance is still a step behind, making the drive less attractive for low-end server and multi-user applications.

Performance aside, the DiamondMax 10’s two-tier warranty leaves something to be desired. You definitely don’t want to get stuck with a retail drive that’s only covered by a one-year warranty, and although the bare drive’s three-year warranty is pretty standard among 7,200-RPM desktop drives, it’s less impressive next to Seagate’s five-year warranty.

At the end of the day, the DiamondMax 10 is best suited for single-user desktops where its fast transfer rates should keep performance-hungry enthusiasts happy. Unlike the 10K-RPM Raptors, DiamondMax 10 drives also offer plenty of storage—up to 300GB for under $200 online sounds good to me.