Western Digital’s Raptor 150 GB with RAID-0

Holy Molee...4 of em config in RAID0...cost??? bout $1500 smakers for 1 smack down setup

Can not get the link to work so until I can...
Taking A Test Spin : Western Digital’s Raptor 150 GB with RAID-0


Author : Chris Connolly Date : 1/12/2006



It may seem hard to believe, but Western Digital's immensely popular Raptor 74 GB hard disk has been on the market for over two years ago, as it was launched in late 2003. At the time of its launch, the Raptor 74 GB only boasted minor architectural improvements over its predecessor, the Raptor 36 GB, but this second generation Raptor drive somehow struck a chord with customers. The Raptor 74 GB became the drive of choice for gamers and enthusiasts who wanted an ultra-fast, but small storage area for their heavily used applications/games, not to mention workstation users who loved the drive's price/performance ratio versus comparable SCSI hard drives.
Since the Raptor 74 GB launched, Western Digital's competitors have come out with drives based on newer standards, boasting larger cache sizes and bigger capacities, but none have matched the Raptor 74GB's performance for a Serial ATA hard drive. In this day and age, it's pretty rare that one product owns the high-end market for more than a few months, much less over two years. Even now, in early 2006, the Raptor is known as the highest performance drive on the market. In some ways, we feel that Western Digital should be congratulated for producing a product with such a lasting effect on the market. However, we also feel that we should scold other hard drive manufacturers, who have not stepped up to the plate and attempted to produce a 10,000 RPM SATA disk to compete against the Raptor, which seems ludicrous to us, as Western Digital has proved there is a major market for this kind of hard disk.

In any case, the Raptor 74 GB's reign at the top of the Serial ATA market had to end eventually. What better company to produce a better product than the Raptor 74 GB than the company who made the original product. Western Digital is back once again with the third generation (3G) Raptor drive, the Raptor WD1500ADFD, or as most people will refer to it, the Raptor 150. The Raptor 150 includes nearly every feature which enthusiasts have been clamoring for, including double the storage space (150 GB vs. 74 GB), double the cache (16 MB vs. 8 MB), and support for Native Command Queuing (NCQ) through the use of a new native Serial ATA controller design (previous generation Raptor drives used PATA to SATA converter chips).

Reports around the web have already shown that this new Raptor is a speed demon - however, we wanted to take it one step further. We rounded up a posse of these drives and went about testing them in a high-speed RAID-0 configuration, to see how much speed is possible from a Serial ATA hard drive configuration. Not only did we test with the fastest Serial ATA drives on the market, we also utilized the fastest RAID card on the market today as well, Areca's ARC-1220 PCI Express x8 controller, to grab every bit of performance possible from these new drives. Let's get those platters spinning and those bits flying.



600 GB of 10,000 RPM Serial ATA Raptor 150 storage.



--------------------------------------------------------------------------------


Raptor 150 Details
Western Digital's Raptor 150 has a similar, but distinctly different appearance compared to previous generation Raptor disks. The latest generation Raptor still maintains a heavier weight compared to standard desktop hard drives, which likely is due to the heavy-duty heatsink fins on the left side of the drive. The drive's appearance has been altered slightly, and now has a sleek all black appearance, similar to Western Digital's new RE2 and SE16 lines of hard drives. Click here for a side-by-side comparison image of the Raptor 74 and 150 drives.
As many of you know, Western Digital will also be producing this drive with a clear plastic panel, allowing you to see the platters spinning with their "Raptor X" variant, which should ship in a few months. That drive should be identical in terms of performance as the drive we’re looking at today, although today’s Raptor 150’s have far less flash.





Note the manufacturing date. These drives are less than three weeks old from the factory.
As with previous generation Raptor drives, the new Raptor 150 can be run with a new Serial ATA-class power connector or a standard 4-pin Molex power connector.




In terms of specifications, the Raptor 150 does not differ too greatly from the previous generation Raptor 74. The new Raptor 150 has an identical 10,000 RPM spindle speed, and latencies and access/seek times are virtually un-changed between generations (seek time has gone up from 4.5 ms to 4.6ms) . The big changes are a doubled amount of onboard cache (8 MB to 16 MB) and a new "native" Serial ATA/150 interface, instead of a bridged parallel ATA/133 to SATA-150 interface like previous generation Raptors. The native Serial ATA interface not only cuts down on chips on the Raptor's PCB, but also ushers in new SATA-specific technologies such as NCQ (Native Command Queuing), Staggered Spin-Up, and improved hot-plugging/hot-swapping support.

Obviously, the drive has double the storage capacity as the previous generation Raptor, up from 74 GB to 150 GB (139.73 GB when formatted). Western Digital didn't simply double the amount of platters in the disc in order to get the added capacity, as the Raptor 150 GB is still based on a two platter design. Western Digital has increased the density of those two platters to provide double the capacity without increasing platter count. This is important, as heat buildup, power consumption, and noise levels rise with the more platters which are utilized.

While the drive still maintains a two platter design like previous Raptors, the new Raptor 150 is rated quite a bit higher in terms of noise levels. The Raptor 74GB is rated at 32 dBA Idle / 36 dBA seek, whereas the new Raptor 150 GB is rated at 39 dBA Idle, 46 dBA Seek, which is a fairly major jump for the Raptor line. While this concerned me quite a bit when looking over the specifications for the first time, I was actually relieved when using the drive for the first time, that the Raptor 150 GB had a similar sonic signature to that of the prior generation. The drive was audible when seeking, and multiple Raptor 150 GB drives can be quite noisy during benchmarks, but in a single drive scenario, the drive was very tolerable and similar in noise level to most modern 7,200 RPM hard disks.

One of the biggest disappointments to many has been that the drive does not utilize a Serial ATA-II/300 class interface, which many of Western Digital's Caviar drives are being equipped with. This interface level limitation should not hinder performance, but does make the drive less enticing on paper, especially if you're buying a new motherboard or controller card with full SATA-II/300 support. The drive does support NCQ technology, which the previous generation Raptor did not. However, NCQ support can be a mixed bag, as we'll show on the next page.



--------------------------------------------------------------------------------


To NCQ or Not To NCQ
One of the Raptor 150's major new features is the ability to support NCQ (Native Command Queuing) technology. NCQ, when active, allows the drive to re-order read/write commands to process them quicker, instead of processing commands in the order in which the drive received them. In theory, this should mean NCQ is faster in just about every scenario, especially during heavy multi-tasking where you're reading data off of different sectors of the hard drive. NCQ, heralded as the "next big thing" for hard drives, is suffering from a backlash of sorts recently. Many users are claiming that you can receive better performance by turning NCQ off.
Sure enough, while running our disk benchmarks, we were confounded that the new Raptor 150 drives were not performing quite as well as expected. Luckily, the Areca 1220 PCIe RAID card we tested with allows you to enable and disable NCQ technology through its software interface, and doing so improved our benchmarks by quite a lot in some cases. Now, we had expected that disabling NCQ would help in some of our benchmark scenarios, but amazingly, nearly all of our benchmarks showed performance improvement with NCQ disabled. Here's a quick breakdown of the performance differences we saw with a four-disk RAID-0 array of Raptor 150 disks with NCQ enabled and disabled.

4 x Raptor 150 RAID-0 (NCQ Enabled) 4 x Raptor 150 RAID-0 (NCQ Disabled) Performance Difference
HDTach Burst Speed 609.1 MB/s 656.2 MB/s +7.7%
HDTach Random Access Time 8.7 ms 8.6 ms +1.1%
HDTach CPU Utilization 6 % 4 % +33%
HDTach Sustained Transfer Rate 302.4 MB/s 303.8 MB/s +0.5%


4 x Raptor 150 RAID-0 (NCQ Enabled) 4 x Raptor 150 RAID-0 (NCQ Disabled) Performance Difference
Iometer Workstation (1) 160.63 I/Os p/s 162.11 I/Os p/s +0.9%
Iometer Workstation (2) 257.87 I/Os p/s 268.06 I/Os p/s +4.0%
Iometer Workstation (4) 368.10 I/Os p/s 389.16 I/Os p/s +5.7%
Iometer Workstation (8) 452.09 I/Os p/s 503.41 I/Os p/s +11.4%
Iometer Workstation (16) 526.33 I/Os p/s 562.04 I/Os p/s +6.8%


4 x Raptor 150 RAID-0 (NCQ Enabled) 4 x Raptor 150 RAID-0 (NCQ Disabled) Performance Difference
Iometer File Server (1) 136.40 I/Os p/s 137.52 I/Os p/s +0.8%
Iometer File Server (2) 229.75 I/Os p/s 233.32 I/Os p/s +1.6%
Iometer File Server (4) 329.88 I/Os p/s 345.75 I/Os p/s +4.8%
Iometer File Server (8) 394.54 I/Os p/s 434.09 I/Os p/s +10.0%
Iometer File Server (16) 464.85 I/Os p/s 484.63 I/Os p/s +4.3%


4 x Raptor 150 RAID-0 (NCQ Enabled) 4 x Raptor 150 RAID-0 (NCQ Disabled) Performance Difference
DiskBench Single 1 GB File Create 107.4 MB/s 104.6 MB/s -2.6%
DiskBench Dual 1 GB Threaded File Create (Average) 75.85 MB/s 74.09 MB/s -2.3%


As you can see, the majority of our benchmarks simply ran faster with NCQ disabled - greater than 10% faster in some cases. Of course, these benchmarks don't mean NCQ is a total loss, and we're sure that certain RAID scenarios will show the Raptor 150's NCQ abilities in a more positive light. However, if raw disk performance is what you crave, we would recommend disabling NCQ on these drives at this time. Most controllers will let you enable or disable NCQ through their driver/software control panels, although most will have this feature enabled by default.

However, we should note that in a single drive scenario, leaving NCQ enabled generally did not hurt performance - and in some cases did help slightly. However, when we went to a RAID setup, the more disks which were added to our array, the more NCQ ended up hurting performance. For a single drive, enabling or disabling NCQ will not make a huge difference in performance, but for multi-disk arrays, we would recommend turning it off.



--------------------------------------------------------------------------------


Testbed and Notes

AMD Athlon64 X2 System Configuration
Processor(s) AMD Athlon64 X2 4400+ (2.2 GHz) - 2 x 128k L1, 2 x 1 MB L2 Cache
Toledo Dual Core Architecture - 0.09 Micron

Memory 2 x Infineon DDR-400 (PC-3200) Memory - 1 GB Total
CAS 2.5,3,3-1T Latency at DDR-400 Speeds

Motherboard Asus A8N-SLI Premium Motherboard
Asus 1.009 BIOS Installed
nVidia nForce4 SLI Chipset
nVidia Forceware 6.70 Drivers Installed

Graphics Card nVidia GeForce 7800 GTX 256 MB PCI Express x16
nVidia Forceware 81.98 Drivers Installed

OS Hard Disk Seagate Barracuda 7200.9 500GB - 7,200 RPM / 16 MB - Serial ATA-II/300

Test Hard Disks 4 x Western Digital Raptor 74GB - 10,000 RPM / 8 MB - Serial ATA/150
4 x Western Digital Raptor 150GB - 10,000 RPM / 16 MB - Serial ATA/150

Controller Cards Areca ARC-1220 PCI Express x8 - 1.02 Driver

Operating System Windows XP Professional Edition - Default Install
Windows XP Service Pack 2 and Direct X 9.0C Installed






Benchmarking Software

Iometer 2004.07.30
NodeSoft DiskBench 2.4.3.1
Simpli Software HDTach 3.0.1.0




Notes


The operating system hard disk was kept separate from the testing hard drives.

Iometer and HDTach benchmarks were run on the drives without a volume and/or file system.

DiskBench benchmarks require the use of a file system, so we used NTFS. Disks were formatted to their full capacity.

Disks were set to use Windows XP "Dynamic" disk mode instead of standard "Basic" disk mode for tests which required a volume/partition.

Areca 1220 RAID card was installed in PCI Express x16 slot, run at PCI Express x8 speed.


--------------------------------------------------------------------------------


SimpliSoftware's HDTach 3.0 is a synthetic windows storage benchmark.
Tests sequential read speeds along with burst, max, and min speeds.
Marks are recorded from unformatted, unpartitioned disk drives.




HDTach 3.0 - Random Access Time
(Lower times mean better performance)


4 x Raptor 150 GB (RAID-0) 8.6



3 x Raptor 150 GB (RAID-0) 8.5



2 x Raptor 150 GB (RAID-0) 8.3



1 x Raptor 150 GB 8.1



4 x Raptor 74 GB (RAID-0) 7.9



3 x Raptor 74 GB (RAID-0) 7.8



2 x Raptor 74 GB (RAID-0) 7.8



1 x Raptor 74 GB 7.7



0 10






HDTach 3.0 - Sustained Transfer Rate
(Higher scores mean better performance)


4 x Raptor 150 GB (RAID-0) 303.8



3 x Raptor 150 GB (RAID-0) 226.7



2 x Raptor 150 GB (RAID-0) 156.2



1 x Raptor 150 GB 77.9



4 x Raptor 74 GB (RAID-0) 263.4



3 x Raptor 74 GB (RAID-0) 195.7



2 x Raptor 74 GB (RAID-0) 131.5



1 x Raptor 74 GB 65.8



0 400






HDTach 3.0 - Burst Transfer Rate
(Higher scores mean better performance)


4 x Raptor 150 GB (RAID-0) 656.2



3 x Raptor 150 GB (RAID-0) 645.8



2 x Raptor 150 GB (RAID-0) 625.0



1 x Raptor 150 GB 554.4



4 x Raptor 74 GB (RAID-0) 650.2



3 x Raptor 74 GB (RAID-0) 637.7



2 x Raptor 74 GB (RAID-0) 606.9



1 x Raptor 74 GB 568.4



0 800





HDTach shows some interesting results. First off, the random access time test shows that the Raptor 150 does indeed have a slightly higher (slower) access time compared to the previous generation Raptor 74. Access times increased at a greater rate when more drives were added to our array, but in the grand scheme of things, these access times are still fantastic.

Sustained Transfer Rate, the key benchmark here, shows the Raptor 150 certainly outpacing the Raptor 74. The Raptor 150 peaks out at 77 MB’s compared to the Raptor 74’s 65 MB’s for a single disk, and as more disks are added to the array, the larger the Raptor 150’s lead grows. With four Raptor 150 GB disks, we’re pushing over 300 MB’s sustained disk read speeds, which is some serious workstation-class storage horsepower.

Burst Transfer Rates are insanely high, since we used an Areca RAID controller card with 128 MB of onboard DDR cache, which is more representative of the controller speed rather than the disk speeds. However, interestingly enough, the Raptor 150’s burst rates are indeed a bit higher when in a RAID array, but slower out of the array.



--------------------------------------------------------------------------------


Iometer is a purely synthetic benchmark designed to test disk/controller I/O.
We tested with the industry standard file system / workstation patterns.
2 workers were used with a varying level of 1-16 outstanding I/O's (total)





Iometer - File System Pattern - 1-16 Outstanding I/O's
(Higher graph lines mean better cooling performance)







Iometer - Workstation Pattern - 1-16 Outstanding I/O's
(Higher graph lines mean better cooling performance)





Iometer gave us some perplexing results, especially when NCQ was enabled (all of the results shown were run with NCQ disabled, which delivered better performance across the board). Both the File System and Workstation Iometer patterns delivered a greater amount of I/O’s on the Raptor 74 models (shown in various shades of blue) compared to the 150 (shown in various shades of red/orange).

While both drives are able to deliver fantastic numbers, we certainly were expecting the Raptor 150 to take the cake in these tests, but in the majority of our tests, the Raptor 74 proved to be more efficient to the tune of 5-10% better performance, a sharp contrast to the raw transfer rates we saw on the previous page.



--------------------------------------------------------------------------------


Nodesoft's DiskBench is a synthetic benchmark compiled with Microsoft .net.
DiskBench tests the time to write newly created data to a hard drive partition.
Results are reported in both MegaBytes per second and time to write the data.




DiskBench - 1 x 1 GB Write Speed Test
(Higher scores mean better performance)


4 x Raptor 150 GB (RAID-0) 104.6



3 x Raptor 150 GB (RAID-0) 87.9



2 x Raptor 150 GB (RAID-0) 72.8



1 x Raptor 150 GB 46.2



4 x Raptor 74 GB (RAID-0) 91.4



3 x Raptor 74 GB (RAID-0) 83.5



2 x Raptor 74 GB (RAID-0) 71.8



1 x Raptor 74 GB 41.6



0 150






DiskBench - 2 x 1 GB Multi-Threaded Write Speed Test
(Higher scores mean better performance)


4 x Raptor 150 GB (RAID-0) 74.0



3 x Raptor 150 GB (RAID-0) 71.9



2 x Raptor 150 GB (RAID-0) 66.2



1 x Raptor 150 GB 36.6



4 x Raptor 74 GB (RAID-0) 73.9



3 x Raptor 74 GB (RAID-0) 65.1



2 x Raptor 74 GB (RAID-0) 53.5



1 x Raptor 74 GB 27.7



0 100





In contrast to read speeds, we ran a set of disk write speed tests on these drives, testing how quickly the drives could write 1 GB and 2 GB file sizes to the disk in 100 MB chunks. Surprisingly, the Raptor 74 and 150 drives were pretty close, although the Raptor 150 once again takes the cake here – delivering roughly 10-20% greater disk write speeds in a single disk scenario (the numbers get a little closer when comparing the drives in RAID).



--------------------------------------------------------------------------------


The Final Word
Western Digital once again has stepped up and delivered a fantastic product with the Raptor 150, delivering just about everything enthusiasts were clamoring for as a follow-up for the Raptor 74. The new Raptor 150 boasts double the storage capacity, double the cache, and a native Serial ATA interface, not to mention the sleek black exterior which cements the drive’s image as a serious, enterprise class device. Gamers and enthusiasts will no doubt flock to the drive as well, just as they have done with previous Raptor drives.
The only obvious feature missing here is a Serial ATA-II/300 interface, which was left off due to quality concerns with new SATA-II/300 components. Serial ATA-150 is a more stable choice at this point, and this fact alone should not affect performance in any way. However, it does make the drive look antiquated to someone who is not familiar with these facts, as many will want a “fast” Serial ATA-II/300 hard drive to pair with their new SATA-II/300 compliant motherboard, ignoring the fact that this Serial ATA-150 disk will be faster than any Serial ATA-II/300 disk you can possibly buy. There are rumors picking up that WD will introduce a “gaming” version of this drive with a Serial ATA-II/300 interface – although they have not confirmed such a product yet.

In terms of raw performance, the Raptor 150 is unbeatable at this point. Most new Serial ATA-II/300 class 7,200 RPM hard drives can push around 50-55 MB/s sustained transfer rates, whereas the Raptor 74 GB could push around 65 MB/s. The Raptor 150 GB one ups them all, maxing out at around 77 MB/s for a single drive, and improvement of roughly 18% over their previous generation product. Write speeds are improved as well, as our previous page of benchmarks showed. Take a few of these drives in RAID-0, and you have some heavy duty enterprise class storage. With four disks in RAID-0 we were able to push over 300 MB/s of sustained data read speed. Simply an amazing amount of storage power at one’s fingertips.

However, it's not all roses at this point. The Raptor 150 GB has slightly higher random access times, although the Raptor 150 is still far quicker compared to every 7,200 RPM disks. In addition, the Raptor 150 GB does run a bit louder and a bit hotter compared to its predecessors. However, neither of these aspects really sways our opinion on the drive too much, as the changes are negligible for most environments.

Since this drive is brand new, there is also a fairly hefty price premium on these drives, putting the drives at roughly $2 / GB range. At $300 and over, the Raptor 150 GB drive is more expensive than Western Digital's Caviar SE16 400 GB, their flagship consumer (7,200 RPM) class hard drive, which is an impressive performer in its own right. It’s a tough call here, the Raptor certainly will deliver greater performance levels, but you will only receive about 1/3rd of the capacity compared to a high-end 7,200 RPM hard disk. Prices will no doubt drift down over time, but it’s unlikely that the Raptor will lose its high-end allure any time soon.

With the launch of the Raptor 150, Western Digital continues to dominate the performance oriented Serial ATA hard disk market, and now their competitors will try even harder to match their performance levels.