AJ's Benchmarking data! (for your consideration).
Table of Contents- Intro
- Test Setup
- Benchmarks
- 5K500.B vs WD5000BEVT vs 5400.6
- Raid 0 numbers
- Real-world performance boost? (compared to original raid 0 config)
=========================
INTRO
=========================
Greetings, all. As promised, here is the benchmarking data I compiled off of my 2 new 5k500.b drives. Apologies on the delay in posting these; I had a really hard time getting my HDD image to work.
=========================
TEST SETUP
=========================
I went through 3 different imaging programs (and about 16 hours) before I found one (Norton Ghost), which finally worked for me. The following data features mostly synthetic benchmarks measured over an eSATA connection. The drives were completely empty at the time. One of the tests (SpinRite: a program that runs below the OS) was not a synthetic test, and features direct communication with the HDD as follows:
taken from the creator of SpinRite:
Quote:
|
SpinRite's drive performance benchmarking system is another unique feature. SpinRite incorporates a new performance measurement index (called Sector Access Velocity, or SAV) which - for the first time ever - actually corresponds quite closely to the perceived performance of the hard disk system being measured. SpinRite also has a new data throughput test that is aware of, and not fooled by, hard disk drive internal caching. It separates and independently measures a drive's buffered and unbuffered data transfer rates.
|
Quote:
Since modern drives "translate" their cylinder, head, and sector counts to anything the system suggests, no one really knows how many storage "cylinders" a modern drive has. This makes the traditional "average seek time" across (unknown) cylinders completely meaningless. What's more, if a drive had many heads, or a large number of sectors per track, it would be storing much more data on each of the drive's cylinders so the heads wouldn't need to move as far.
SpinRite's new concept of "Sector Access Velocity" (SAV) measures the speed in megabytes per second at which the drive's heads are able to traverse the disk. It's the rate at which the drive moves through data you don't want to get to the data you do want! We've found that this SAV score seems to correlate precisely with the preceived performance of the drvive. It's a great way to really see how well different drives perform!
|
http://www.grc.com/files/sr5_manual.pdf
For the rest of the tests, you're all familiar with. HDTune, SciSoft Sandra Pro, Crystalmark). Unfortunately, I don't have any other of our favorite 5400rpm hard drives, so I'll be relying on others results later on.
On to the tests...
=========================
BENCHMARKS
=========================
===========
SpinRite 6
===========
Random sector access (milliseconds) -
18.074
Sector access velocity (bytes/ses) -
9,049,036
Burst transfer rate (bytes/sec) -
72,005,912
Sustained transfer rate (bytes/sec) -
92,393,196
Smart polling speed (seconds) -
0.0055
===========
HDTune 2.55
===========
=============
CrystalMark 2.2
=============
===========
Sandra Pro
===========
(my drive is the
red line)
=========================
5K500.B vs WD5000BEVT vs 5400.6
=========================
Here is my 5K500.B benchmark juxtaposed over
John Ratesy's Scorpio Blue and 5400.6 findings.
Left to right, 5K500.B., WD5000BEVT, 5400.6.
I submit to you all that these 3 benches represent an equitable comparison between the 3 drives, as none of these 3 drives were the C drive at the time of testing.
5K500.B doesn't lag very far behind in anything, and wins most comparisons going away.
==================
Raid 0 numbers (FYI)
==================
Crystal Mark
My original Raid 0 configuration was of 2 Fujitsu 4200rpm HDDs. Their results in HDTune and SciSoft Sandra Pro are below:
HDTune 2.55
Fujitsu MHZ2500b (4200rpm)
Hitachi 5K500.B (5400rpm)
This set of numbers represent the
read performance only. Sandra doesn't allow write performance on any drive with any content to be bench-marked in that way.
That being said, the difference is pretty clear on HDTune & Sandra
But how about real life feel?
==========================================
"REAL WORLD" PERFORMANCE BOOST? (compared to original raid 0 configuration)
==========================================
Boot-up speeds
measured from power button press to the final startup app loaded.
Average, original 4200rpm raid 0 (seconds)* -
127
Average, 5K500.B 5400rpm raid 0 (seconds)* -
98
Conclusion: Hitach drives provide
~30% faster boot time.
UnRAR speeds
Extracting a 4.87GB file
Average, original 4200rpm raid 0 (seconds)* -
~179
Average, 5K500.B 5400rpm raid 0 (seconds)* -
~114
Conclusion: Hitachi drives provide
~57% faster UnRAR time.
Photoshop CS4
Opening Photoshop + 15MB PSD, measured from icon click to the image and all menus opening.
Average of initial opening after reboot, original raid 0 (seconds)* -
~2:59
Average of initial opening after reboot, 5K500.B raid 0 (seconds)* -
~1:54
Conclusion: Hitachi drives provide
~89% faster time. Secondary open (cache'd open) provides equal launch time (~5.8 seconds). Undoubtedly because the app is still in memory, as expected. Certainly excellent gains.
=====================
FINAL THOUGHTS
=====================
I'm impressed with the increased speed and the synthetic benchmarks compared to other well-known 5400rpm drives. Based on the synthetic data, I am inclined to believe that
SpinRite is indeed an excellent app for REAL (non-synthetic) benchmarking. Those who enjoy testing these drives might want to consider employing it more often.
The drives are indeed absolutely
silent. I have noted, though, that when they've been asleep and wake up, they make
an awkward clunking sound. Just once as it's warming up and maybe once every few hours. I assume this is normal, as none of my drive health tests suggested any problems (if it's not, someone please tell me). Otherwise, I couldn't hear them when I had the external enclosure open (even when holding my head to the drive). I also feel no vibration.
I have not done a power drain test yet to see how much longer the battery life might be. I will sooner or later, though.
It took awhile to get the last benchmarks so I could post these, but I'm glad it's done. I hope this information is of value to you guys.
I will have a pair of 7200.4's soon, and they'll be tested in the same fashion and numbers directly compared. if the 7200.4's don't provide a meaningful speed gain over these, I'll be returning those and keeping these. And I wish I could have personally compared it to the Scorpio Blue. Oh well.
That being said, the 5K500.B appears to be the best 5400rpm HDD available today. I wouldn't hesitate to pick them up. I got them from for $80/drive. You can't beat the performance for this price.
Verdict: Winner. Solid, inexpensive, silent, fast.
Pick one up.
If there are questions, I'll be sure to answer. thanks
