Recommended Hard Drives

Our reference article about HDD noise, along with complete acoustics-ranked listings of all SPCR-reviewed HDDs since late 2008. Latest update 4 October 2014.

• Oct 04, 2014: Complete overhaul & update
• April 28, 2010: Complete overhaul of article, with only
  HDDs tested (or retested) in the anechoic chamber.
• June 13, 2008: WD single-platter 320GB, Velociraptor and
  Samsung F1 drives added
• Dec 15, 2007: Western Digital Green Power WD7500AACS added
• April 3, 2007: Samsung Spinpoint T HD400LJ and WD Scorpio
  120 added; other minor changes.
• June 30, 2006: Minor changes and added the WD5000KS as
  the quietest currently available drive.
• October 4, 2005: Complete revision of text and rankings
  based on reviews with our latest HDD
  testing methodology.
• January 15, 2005: Another reevaluation of the whole article
  and rankings.
• April 11, 2004: A long overdue overhaul of the entire article
  and ranking table, including information from reviews over the past year,
  expanded HDD discussion, a new notebook drive table, and a section on some
  excluded drives.
• May 11, 2003: Info from recent reviews added, including
  Samsung SP1604N, Seagate Barracuda IV & 7200.7, and IBM 180GXP. Also revised
  rankings slightly.
• Jan 11, 2003: IBM 180GXP added
• Dec 10, 2002: Minor changes + addition of HDD Noise Reduction
  Products
• Updated Sept 18, 2002
• First published July 17, 2002

SPCR began publishing hard drive reviews in March 2002. From the beginning,
the focus was on noise. An improved methodology for testing hard drives was
introduced in May 2005. Since then, we have made one major change: All acoustic
testing has been conducted in our
own anechoic chamber since the fall of 2008. This allows us to obtain accurate
SPL measurements down to about 10~11 dBA@1m. Previously, we were limited by
ambient noise and limitations in our audio measurement equipment to about 18
dBA@1m during the day. The upgrade in our acoustic testing coincides happily
with a wider range of new 5400 / 5900 RPM HDDs that bump against the noise floor,
even in our anechoic chamber.

HDDs

HDDs are racing to ever higher capacity, and ever lower prices as well. From
the $/GB perspective, 5400 and 5900 RPM 3.5″ drives remain king. The HDD
is not going away any time soon with the massive amounts of data growing ever
larger in the data centers, and the similar acceleration of image and video
accumulation in personal computing devices. As I write this, mainstream HDDs
have reached 6TB capacity, and 8~12 TB will be achieved in the very near future.

Corporate consolidations in the past decade have shrunk the number of hard
drive manufacturers to just three: Samsung’s HDD division is now a part of giant
Seagate, and only the 2.5″ drives have retained the Samsung label thus
far. Hitachi’s HDD division is now part of Western Digital, the other big player,
but its brand continues relatively unchanged. Toshiba also makes a full range
of drives, though best known for its notebook drives, and retains ~15% of the
drive market, with Seagate and WD splitting the lion’s share.

There have been no dramatic revolutions in hard drive technology in recent
memory; rather, a series of progressive evolutionary improvements that have mainly
targeted capacity by increasing the areal density of the platters. Performance
improvements do come simply because higher areal density means shorter head
movements to read different sectors of the drive. The fastest high capacity
7200 RPM drives now reach 150 MB/s sustained file transfer rates. Big 5400 RPM
aren’t far behind at around 120MB/s. The slowest current SSDs are at least 2X
faster than the fastest HDD.

Sub-6000 RPM 3.5″ HDDs have been our go-to quiet high capacity data storage
devices for years. The best of these now have performance barely perceptibly
slower than typical 7200 RPM drives, and achieve a level of noise and vibration
that is staggeringly low by standards less than a decade ago.

SSDs

Solid State Drives continue making inroads to the mainstream. We tend to mark
Intel’s entry into the market in Q3 2008 as the starting point of SSDs becoming
viable storage options for PC enthusiasts. The prices continue plummeting, so
it’s just a matter of time before they are in every computer. SSDs are generally
silent, make no vibrations, and run extremely cool, all of which makes them
ideal for silent computers. We recommend using them for the operating system,
at least, and keep to the models that have the best reliability. No SSDs are
on our recommended lists; noise is not a criteria to choose them by, though
we have run into one or two samples that made some low level electronic whine/squeal.

SSDs for consumers come mainly as 2.5″ SATA drives topping out at around
1TB, with the latestest conforming to the SATA3 standard which allows up to
6.0 Gbit/s native transfer rate. There’s no need for the extra room of the 3.5″
drive form factor, so none have appeared in that size. They are also available
in smaller, bare PCB formats for direct plug-in to mobile and desktop motherboards,
mSATA and M.2. The fastest, highest capacity and most capable configurations
have taken PCIe form, used generally for enterprise and data center applications.
The latest development in PCIe SSD technology is NVM Express (NVMe), a scalable
host controller interface designed to provide a platform for ever faster data
delivery.

Longevity of SSDs is directly related to flash memory’s write endurance, which
is finite. Intel rates their top consumer drives for “up to 70GB writes
per day for five years (compared to the industry typical 20GB).” Kingston
rates their top consumer SSDs as capable of 3 Drive Writes Per Day and warranties
them for 3 years. The
Tech Report started a stress test project on six SSDs in August 2013 to
answer the question of how many writes consumer SSDs can take before dying.
The answer, after more than a year of continuous hammering, is surpringly upbeat:
3 failures at 700~800TB of writes, another failure at 1.2 Petabytes (1000TB),
and 2 more still running at over 1.5PB. None of the SSDs are rated for over
~200TB of writes, and none died without some advance warning.

NOTEBOOK DRIVES

We used to recommend 2.5″ notebook drives for the quietest desktop systems
for many years. Notebook drives generally have these advantages over 3.5″
drives: Lower noise, less vibration due to lower moving mass, much lower power
consumption, and correspondingly less heat.

Despite all of this, some sub-6000 RPM 3.5″ desktop drives are now just
about as quiet, and in a few cases, quieter, with about as low vibration. Notebook
drives are generally slower than desktop drives, and cannot hold a candle to
SSDs, which is the main reason we’re no longer as enthusiastic about notebook
drives.

The typical enthusiast PC configuration today is an SSD of 120~500 GB for the
OS and programs, combined with a high capacity 3.5″ HDD, say 2 to 6 TB.
Even in a small PC, a single 3.5″ HDD can be mechanically decoupled in
some way to reduce vibration induced noise to a minimum.

The only scenario in which 2.5″ HDDs have practical application for a
quiet desktop PC these days is when the case is too small for a 3.5″ data
drive. The most obvious example is an Intel NUC or Gigabyte BRIX with an mSATA
slot and room for a single 2.5″ drive. The obvious choice of the latter
is a large capacity 2.5″ HDD, which get up to 2 TB now.

HDD RANKINGS

We used to short-list only the quietest drives on the recommended lists. For
this iteration, we’ve changed our methodology: Every drive tested since late
2008 (when they began to be tested our anechoic chamber) is listed, ordered
by its level of noise and vibration. Study the data, read the reviews, and pick
your own poison.

We test drive performance, but these results don’t come into play in our rankings.
The performance test results are available in the reviews. We caution against
expecting measured differences to be easily or always perceptible. Generally,
performance is similar for models of similar basic specifications: Spindle speed,
capacity, and areal density. Small differences in HDD performance are almost
impossible to appreciate in actual use because there are umpteen bottlenecks
and overheads in the PC that obscure such differences. (For an entertaining,
informative exposition of this theme, check out Dan’s Data’s How
fast is a hard drive? How long is a piece of string?)

Two assessment factors are worthy of note:

• Sample variance is a hurdle we cannot overcome without examining random samples from many production batches over a period of time. This is not feasible. The reviews and our rankings are based on a careful assessment of our samples only.

• Manufacturers sometimes revise products without notice or any change
  in model number. Even an updated firmware can affect drive noise, as seek
  strategies can be changed. Keep this in mind when perusing the recommended
  list; the date of manufacture and the firmware version of our sample is listed.

TYPES OF DRIVE NOISE

The noise of a disc drive mounted in a case comes in two forms:

1. Airborne acoustics is what all drive manufacturers currently specify as the HDD noise. It is the sound that comes from the drive through the air to the observer. This value is measured with the drive suspended in space by wires.
2. Structure-borne acoustics induced by the drive’s vibration during idle and seek is not quantified by HDD makers. This vibrational energy is transmitted to the PC chassis and causes the chassis to act much like a sounding board.

Structure-borne acoustics is the dominant source of HDD-induced PC noise. Seagate’s testing has shown that changes in stand-alone drive acoustics had little effect on the overall system acoustics when drives were hard mounted in the chassis. Hence the dramatic noise reduction evident with decoupled mountings such as the NoVibes, SPCR’s own elastic suspension, or simply placing the drive on soft foam. The noise emitted by even drives with very quiet stand-alone performance is greatly effected by how it is mounted to a chassis. There is also a useful forum thread on the effectiveness of various HDD decoupled mounting techniques.

There are also two
main types of noise:

• Idle noise – typically a smooth hum or whoosh, caused by the spinning
  motor and its bearings. Non-FDB drives often exhibit a high pitched whine.
• Seek noise – a rough, intermittent “clacking” or rapid
  “chugging” noise caused by head actuator movement during seek, read
  and write.

Idle and seek noise have both airborne and vibration-induced components. The
relative balance between airborne and vibration-induced noise is influenced
by the case and the method of installation (i.e, soft vs. hard
mounting). In other words, setup affects how a hard drive will sound. This is
why we do not produce a unified rating for drive noise, but measure and report
both airborne noise and vibration.

FACTORS AFFECTING DRIVE NOISE

Until about 2005, the majority of hard drives on the market used ball-bearing
motors, which had a characteristic high pitched whine and other objectionable
airborne noise. Since then, the industry has shifted to much quieter FDB (Fluid
Dynamic Bearing) motors, with the result that most recent drives are significantly
quieter than older drives, sometimes by as much as 10 dBA@1m. All major drive
manufacturers now use FDB motors in their current lineups. If you have a typical
non-FDB drive, the simplest way to achieve lower noise (and improved performance)
is to swap it for a new drive, almost any new drive.

There are three other factors that affect drive noise:

The number of platters in the drive. The acoustic difference between
a single platter drive and a four platter drive is much smaller than between
a ball-bearing drive and a FDB drive. In other words, the noise penalty for
using a higher capacity drive is not great, especially with FDB bearings.
Still, the quietest drives tend to be single platter FDB models.

The difference between idle and seek noise. While a drive idling quietly
may not be intrusive, if seek noise is considerably louder than idle noise,
it will certainly be noticeable. The smaller the difference between seek and
idle noise, the less audible the drive will be. The sound quality of seek
in some drives is more tonal and more prominent than in others, even when
the measured SPL is the same as in other drives.

Automatic Acoustic Management (AAM). This technology sacrifices some
seek latency in favor of softer, quieter seeks. The performance hit is often
small enough that AAM is well worth using. With the notable exception of Seagate,
all manufacturers these days allow AAM to be enabled, although its effectiveness
varies from drive to drive. Not all manufacturers provide a utility to enable
and adjust AAM, but many drives work with Hitachi’s
feature tool.

USEFUL LINKS

In SPCR

• SPCR’s Hard Drive Testing Methodology:
  Detailed information about drive acoustics and our testing methods.
• Hard Drive Silencing: Sandwiches &
  Suspensions: One of SPCR’s first articles, but still a good reference
  for how to silence a hard drive.
• HDD
  vibration & noise reducing methods – ranked: A useful forum thread
  on the effectiveness of various HDD decoupled mounting techniques. Filled
  with many anecdotal accounts from users.
• Is the Silent PC Future 2.5-inches wide?
  The quietest mechanical drives tend to be 2.5″ drives originally meant
  for use in notebooks. This article documents how — and why — these
  drives can be used in a desktop system.
• An Anechoic Chamber for
  SPCR: How SPCR’s far-flung but dedicated community helped
  to fund the creation of an anechoic chamber and the acquisition of new audio
  test gear suitable for testing low noise

On the Web

• Disc
  Drive Acoustics: Comprehensive, informative and readable white paper
  from Seagate (PDF) dated 2001 but still pertinent today.
• Storage Review:
  Excellent resource for drive performance information. Like SPCR, they specialize,
  and they are good at what they do.
• Hard
  Disk Drives: Thorough explanation of how hard drives work from The
  PC Guide.
• Introducing
  the SSD Endurance Experiment: Just how long do they last, anyway?
  The Tech Report’s well-conceived systematic 1+ year long hammering of six consumer SSDs to answer the question.
• How
  long do disk drives last? Ongoing HDD life study at Backblaze, which
  has used >25,000 consumer HDDs to provide online backup for over five years.

* * *

Discuss
this article in the forums.

GUIDE to the TABLES

Drive: Manufacturer and model number.

Mfg date: Date of manufacture of our sample model.

Firmware: Firmware version of our sample model.

Vibration: The level of vibration produced by the drive at idle, evaluated
relative to standard reference drives on our
standard vibration box. 10 = zero vibration, 1 = jackhammer level.

Airborne Acoustics: The measured decibel level, A-weighted (dBA),
at 1m distance, during idle and seek, with AAM enabled and disabled (when
this is a switchable option).

Power Consumption: The amount of power in watts used by the drive.
Can be used to gauge thermal behavior.

NOTE: Only HDDs that were tested (or retested) in SPCR’s anechoic
chamber are listed in this guide (post-2008). This does not mean that previously
tested and recommended HDDs are not quiet, only that we lack sufficiently
accurate acoustic data to put them in context. Keep in mind that drives
that date back more than a couple years may be discontinued.

HDD NOISE REDUCTION PRODUCTS

Hard drive noise reduction products typically take the form of enclosures,
although the category is not strictly limited to such devices. Noise reduction
products are tested using the loudest and most vibration-prone drives in the
lab. The noise levels are compared with and without the device. The measurements
below reflect the sonic improvement effected by the products.

NOTE: Demand for such products has fallen with the wide popularity of quiet, low vibration drives, especially the 5xxx RPM drives, and the increasing capacity & falling price of SSDs. As a result, these products have very limited distribution and may be difficult to find. Grow Up Japan is the maker of the Smart Drive, and a search with that name will work better. NoVibes is marketed under some other brand names as well as NoiseMagic.

* * *

Discuss
this article in the forums.