Crossing Velociraptors

Crossing Velociraptors

You've seen the benchmarks. Everyone's been cooing about Western Digital's Velociraptor drive--us included--since the moment the early versions of the drives started hitting reviewers' doors. Some sites reviewed early versions of the drive as-is, others noted that these models were engineering samples. In practice, I think the latter is the best route, and that's just what Maximum PC did. Especially since Western Digital promised speed improvements in upwards of ten percent once they finished tweaking the drive's firmware for its final iteration.

Now that the fateful day has come, I've been curious to see just how much extra juice they could pull out of this speedy little monster. Here's a look at the before-and-after:

BENCHMARKS
  WD Velociraptor (old) WD Velociraptor (new) Percent change
HDTach Burst (MB/s) 255.1 249.7 -2.11%
HDTach Random Access (ms) 7.1 7.1 0.00%
HDTach Average Read (MB/s) 104.6 108.4 3.63%
HDTach Average Write (MB/s) 96.7 100 3.41%
PCMark05 Overall 9457 9450 -0.07%
All HDTach scores use HDTach 3.0.1.0. Best scores are bolded.

What do these numbers tell us? Western Digital's tweaked the performance of the drive, obviously. But the faster read and write speeds are nothing to write home about. Remember, HDTach is more a diagnostics tool than anything else. It's great for determining potential problems for your drives, but its access patterns don't replicate the real-world experience you would encounter. For that, we turn to PCMark05. And as you can see, the performance difference is negligible--probably even within an acceptable random variance were we to run the benchmark ten separate times.

I'm still going to review finished hardware. It's the Maximum PC way. And the Velociraptor is still the fastest consumer-grade hard drive one can buy. But its performance -- while mildly different in a synthetic benchmark -- isn't as "tweaked" as we expected given Western Digital's claims.

Check out the full review of Western Digital's Velociraptor drive!

5

Comments

+ Add a Comment
avatar

bcweir

What a cop-out! Western Digital cheated on the performance to goose the specs! MaximumPC ought to invalidate the previous test results, and make the post-release specs the corrected results on this drive.

WD's competitors shouldn't have to compete against a "doctored" drive! I'm glad MaximumPC gave us users the straight scoop on this.

I seem to recall that ATI and NVidia got caught in a big stink when ATI cheated on a benchmark by dithering down the resolution in a background to boost their benchmark score.

I hope that neither MaximumPC nor computer users intend to give WD a pass on this attempt to cheat on the benchmark.

avatar

TheMurph

Huh?

The previous test results were on a blatantly labeled "engineering sample" of the drive. Western Digital told everyone, on multiple occasions, that the final performance of the drive will vary as they were not yet done tweaking the firmware for the as-shipped consumer versions.

In short, you can't *buy* an engineering sample. The drive in this article is what you, the consumer, will pick up at the store. And that's what we reviewed. I just wanted to see *how much* of a performance upgrade WD was able to achieve. They were hinting somewhere in the 10% range, but as you can see, that was not the case.

But they aren't cheating benchmarks by any means. I don't see how you can draw that point.

avatar

sheldon

i feel so outta place without the podcast :-(

~ *CompTIA A+ Certified IT Technician* ~___Accomplished 2007___Age 16

Working on; ~ CISCO CCNA

avatar

Spiffy

The Gordon soundboard can only fill so much void.

Surely now that you have Norm, you can make him do the work and you take 1 hour to mak podcas.
From hell's heart, I stab at thee.

avatar

Block_Dude

jw, where's the podcast been these past few weeks?

Log in to MaximumPC directly or log in using Facebook

Forgot your username or password?
Click here for help.

Login with Facebook
Log in using Facebook to share comments and articles easily with your Facebook feed.