Samsung 840 Pro 256GB SSD User Hardware Review

[rubix_1011]

Hello TH Forum members!

Recently, a few of us were given the opportunity to put a Samsung 840 Pro 256GB SSD drive through it's paces for a real-life-usage scenario for our day to day lives. This thread contains these steps and tests according to many daily tasks that I do every day between work and home (mostly work).

I wanted to do a comparison of the Samsung 840 Pro against other drives that I have at home on my desktop PC; a Western Digital Black 1TB 7200 RPM drive and an OCZ Vertex 3 120GB SSD drive; the latter of which has been used as my main OS drive for about a year. This will allow me to demonstrate differences between a user upgrading from a normal platter HDD (the WD 1TB) and someone already with an SSD drive (Vertex3).

Looking at this SSD drive, initially what strikes me is how attractive it is. It's a black, brushed aluminum shell and feels very lightweight...almost as if it could float on water. The packaging is very straight forward, and includes a disc with some basic software for updating firmware, drive cloning and benchmarking. A very generic pamphlet with simplistic images shows the steps on installing the drive either in a laptop or a desktop computer; both of which omit several steps that an uninformed user would take for initial connection, but I believe that it's implied when buying an SSD, either you know how to install it, or you have someone you know that does. A couple of very 'Intel-like' stickers adorned with 'Samsung SSD' and 'activated' round out the packaging (see bottom right corner of Samsung packaging above). One thing is missing, I feel...something very important.

The 3.5" to 2.5" adapter bracket for a desktop computer.

My OCZ Vertex 3 came with one, which was what I expected in this scenario as well. Sadly, Samsung must think their SSDs will only be installed into laptop devices or that their customers enjoy spending another $5-$10 after the fact to run around and find the bracket to install the drive they just purchased. I truly feel this is a cheap way out for Samsung and I hope they reconsider the idea of packaging the drive bracket...the drive bracket that likely costs a handful of pennies to produce and include. I have always felt that retail-packaged hardware should always come with all items to at least make it functional, especially in the sense that 2.5" drives and adapters are still not always standard fare on new PC cases and especially not for that 3 year old case someone might be upgrading with a Samsung SSD drive.

That all being said, my case is a homemade, hand built case that I put together about a year and a half ago out of wanting to have something original and still handle all my watercooling gear. If interested, you can view it in the watercooling gallery and my build log here:

Build Log | Project: Askew

Since this build, the hardware has been updated to what exists below.

I simply ran a SATA 3 cable and power to the drive to allow me to quickly swap between drives and tests. This actually made my life very easy for all tests and allowed very fast access to the motherboard to connect/reconnect SATA cables.

Test System:

Intel i7 2600 Core 2 Quad
ASUS Sabertooth P67 Motherboard
G.Skill Ripjaws DDR3 1600 - 16GB
EVGA nVidia 560TI (x2 in SLI)
BFG 1200w Powersupply
HannsG 30" monitor
Scythe KazeMaster Fan Controller
self-built case

Full, custom watercooling system:
D-tek CPU block
Swiftech MCW60 (x2) GPU blocks w/RAMsinks
Swiftech MCP655 (D5 vario) pump
Swiftech MCR320 (x2) radiators w/6 Ultra Kaze fans (2000/3000rpm)
Bitspower Multi 80 reservoir
distilled water/biocide/killcoil


Installed Software:
Windows 7 x64 Ultimate SP1 + all updates
Notepad++
AVG Free
Dropbox
Google Chrome
TeamViewer 8
7zip 9.20 (x64)
TeraCopy 2.27
nVidia Driver 320.18
GIMP 2.8.4
MirthConnect 2.2.2.6388.b54
Java x86 & x64
Java SDK x64 7u25

Test Scenarios:

All tests will be performed on each drive's desktop <--> desktop for all applicable processes.

1) Install Windows 7 x64 Ultimate, SP1 - USB key to desktop 'ready'

2) MirthConnect XML Transformer - Transform 10,000 XML files (21.3GB) by re-encoding and outputting as XML Pretty Print.

3) Handbrake - Convert ISO of 'Monty Python and the Holy Grail' Standard DVD (Special Edition) into .mp4 with priority level set to 'normal', all other defaults and 'Normal' profile.

4) 7zip - XML file compression of the same, original 10,000 XML files (21.3GB) with large memory pages set in the 7zip options.

5) GIMP - image batch processing of 217 (607MB) .jpg photos from a charity golf tournament that I shot last June. Batch configurations include:
Color - Enable, Auto levels
Resize - Enable, Relative 0.35
Sharpen - Enable, Defaults
Output - .JPG format, Quality 100%, rest defaulted

6) Launch of SQL Developer from quick launch to working status. (Typically a resource intensive Java reliant application for database interaction.)

7) Use Samsung's Packaged 'Samsung Magician' software to benchmark each drive's I/O.


SSD-Specific Drive Configurations and Considerations:

SSD additoinal configurations and optimzations outside of normal Windows install:


Turn off or minimize the system paging file for more than 8GB of RAM (I run 16GB RAM on my desktop, so I typically only set to 1GB)
Start menu > right click 'Computer', select 'Properties'
Choose 'Advanced System Settings' from left hand options
From System Properties box, choose 'Advanced' tab and in the 'Performance' box, hit 'Settings'
In the 'Performance Options' box, choose the 'Advanced tab, and in 'Virtual Memory' box, choose 'change'
Uncheck 'Automatically manage paging file size for all drives'
Choose your C: drive (or primary system drive) and select the radio button below to 'Custom size'
Paging file can be set to 'No paging file' if you wish but I keep a minimal size just as a precaution. With 16GB RAM, I really don't have a need to page to disk. Making the minimal and maximum page file size the same helps manage the OS allocation.
Save changes, restart.


Turn off Drive Indexing on C:

Start Menu > 'Computer'
Right click your C: drive and uncheck the box in the 'General' tab 'Allow files on this drive to have contents indexed in addition to file properties'. Click Apply. You might be prompted that some files might be in use or not able to be immediately updated, just click skip all allow to complete. Click OK.


Disable Prefetch and Superfetch:

Start > Run > 'services.msc'
In the services menu, scroll down to 'Superfetch'
Double click to open, choose the dropdown and select 'Disabled'

Start > Run > 'regedit'
Expand to:
HKEY_LOCAL_MACHINE\SYSTEM\CurrentControlSet\Control\Session Manager\Memory Management\PrefetchParameters

Double click and set both values from '3' to '0' [zero] (I ensure I use 'decimal'; but 0 hexadecimal is also '0')
EnablePrefetcher
EnableSuperfetch


Ensure TRIM is functional -You shouldn't have to do this step, but it's decent to check to ensure Windows has this set correctly.

Open CMD prompt from Run or from the Accessories menu, ensure you are running 'As Administrator'

At CMD prompt type the following and hit enter:
fsutil behavior query disabledeletenotify

Based on the response, see below.
DisableDeleteNotify = 1 (Windows TRIM commands are disabled)
DisableDeleteNotify = 0 (Windows TRIM commands are enabled)


Other SSD specific steps that were not performed, but I typically do for normal installs:

Turning Off the Hibernation file:

Not done for these tests since they are temporary for this report, but this is one step I always follow as I never use hibernation as SSDs boot so quickly.


Reducing the System Protection file size:

Not done for theses tests since they are temporary for this report, but also another step to reduce the amount of SSD storage space for restore points. Good if you are diligent about drivers and applications you install; consider leaving decent size if you tinker with questionable apps or downloads often (CYA).

 

[rubix_1011]

Windows 7 x64 Ultimate SP1

Installation from USB key to working desktop, timed in min:sec

For this test, I wanted to test the install time of Windows 7 64 bit onto each of the 3 drives and record the time taken to complete the Service Pack 1 disk. All disks were tested using the same SATA 3 port on my motherboard during each run. Only the DVD drive and USB UEFI Win7 source drive were present as reported by the BIOS:

WD 1TB Black: 10:22
OCZ Vertex3 120GB: 9:23
Samsung 840 Pro 256GB: 7:56

The Samsung wins this category rather handily although I expected a closer race between the OCZ and Samsung. Based on these numbers alone, you can imagine what I assumed would take place for the results of all following tests...

Once all drives were installed with the OS, Windows Updates was run on each to install all available updates and drivers.

 

[rubix_1011]

Handbrake

"It's not a question of where he grips it! It's a simple question of weight ratios! A five ounce bird could not carry a one pound coconut."

Pardon for the comic relief of my next test, the Handbrake ISO conversion to mp4 of the standard DVD, "Monty Python and the Holy Grail"; one of my all-time favorite comedy flicks. I even confess that a friend of mine gifted me a plush 'Holy Hand Grenade of Antioch' a couple years ago and it has adorned my work desk ever since.

Using Handbrake, I only set the priority level to 'Normal' in the options. All other settings were left as default and using the 'Normal' profile:

WD 1TB Black: 7:10
OCZ Vertex3 120GB: 7:00
Samsung 840 Pro 256GB: 7:13

Interesting result, especially after the Samsung smoked both drives in the Windows 7 install test. I really don't have much explanation for this and I even re-ran the test on both SSDs again within a 1 second tolerance.

 

[rubix_1011]

GIMP

GIMP is a free photo editing application that allows a great deal of control and modification to photos and images similar to many of the tools offered in the Photoshop suite. I ran a batch editor against 217 photos taken for a charity golf tournament last summer which was a total of 607 MB orgininally snapped in 10MP .JPG format. I used Dave's Batch Processor for GIMP (check Google and YouTube for info and tutorials on this simple process) and ran them against:
Color-Enable, auto level
Resize-Enable, Relative 0.35
Sharpen-Enable, defaults
Output-JPG, Quality 100%, remaining defaults

WD 1TB Black: 3:58
OCZ Vertex3 120GB: 3:52
Samsung 840 Pro 256GB: 3:47

Seems like a small spread, but this was 'only' 217 photos, but the Samsung came out 5 seconds ahead of the OCZ drive. Depending on your point of view, this is either minuscule or quite substantial. If I were to try and apply this batch process to a few thousand photos, the differences of mere seconds would begin to add up to minutes quite quickly; especially when you still consider the WD platter HDD.

 

[rubix_1011]

Mirth Connect

Mirth Connect is a Java based engine for transforming and processing data; typically used for clinical data, but can also handle just about any need as it can utilize JavaScript commands, database readers/writers and perform dozens of simultaneous tasks. It's an application that I find incredibly useful every day at work and allows a multitude of possibilities in many tasks I regularly perform.

Mirth Connect was setup with a Java server heap of 512mb and configured to process 10,000 (21.3 GB) irregularly formatted XML files and correctly encode and transform the output into XML Pretty Print to be more 'human readable'.

This application is very dependent upon CPU and memory, but is also very I/O dependent; the faster it can read, process and write the output file, the faster it can pick up the next XML file to process.

WD 1TB Black: 35:57
OCZ Vertex3 120GB: 23:44
Samsung 840 Pro 256GB: 26:14

Another set of interesting results where the OCZ actually pulled through about 2.5 minutes faster than the Samsung. It's no surprise that the WD can't keep up in the repetitive I/O of all those pesky XML files. Consequently, I use a 320GB HDD on my work laptop and this was one of the most important tests I wanted to perform, along with 7zip and SQL Developer launch times.

 

[rubix_1011]

7zip

7zip is a free alternative to the popular WinZip and WinRAR file compression and handling applications. It's powerful, fast and works well from all menu functions.

For this test, I zipped the same 10,000 (21.3 GB) XML files used as the source for the MirthConnect test into a .zip archive. I set 'Large Memory Pages' in the 7zip options for each test:

WD 1TB Black: 7:39
OCZ Vertex3 120GB: 7:29
Samsung 840 Pro 256GB: 7:48

Not too sure what happened here with the Samsung actually getting beaten by the WD platter HDD by 9 seconds. I expected to see both SSDs perform relatively close, but the OCZ actually beats out the 3rd place Samsung by 19 seconds. I'm hard pressed to offer any kind of relevant explanation on this other than this simply comes down to the internals used by both the SSDs being used for comparison and their strengths/weaknesses, therein.

 

[rubix_1011]

SQL Developer

SQL Developer is an Oracle product aimed at users that need interaction and viewing of databases and most applicable components. This is a Java-based tool that I use constantly throughout the day, usually at the expense of opening and closing many, many times over the course of 8-10 hours of work, depending on need. On my Lenovo T430 with an i5-3210M and 8GB RAM and 7200 RPM drive, it can take upwards of 10-15 seconds to load to working state. I wanted to see that if I gave SQL Developer the ability to run a system with a Core i7 and 16GB of RAM, would that allow it to show whether the disk I/O is the bottleneck:

WD 1TB Black: 0:06 (6 seconds)
OCZ Vertex3 120GB: 0:04 (4 seconds)
Samsung 840 Pro 256GB: 0:04 (4 seconds)

Well, I guess I figured out the answer to my question- given enough raw CPU horsepower and memory, SQL Developer opens in a quick jiffy regardless of drive being used in my tests. I thoroughly expected a much bigger lag between the SSDs and the platter HDD since the SQL Dev startup is almost entirely read processes. Regardless, this Java based app is very responsive when given the system resources it demands to respond quickly to launch.

 

[rubix_1011]

I ran the bundled Samsung benchmark software, Samsung Magician, against each drive to do a comparison of what it determined the speed and performance of each was.

Following the same order as all other tests:

WD 1TB Black

OCZ Vertex3 120GB

Samsung 840 Pro 256GB

Both the Samsung 840 Pro and the OCZ Vertex3 show nearly identical sequential read (555MB/s vs. 552MB/s; Samsung/OCZ) and write scores (513MB/s vs. 519MB/s; Samsung/OCZ) while the Samsung seems to show more than a 3:1 ratio advantage in read IOPS, which doesn't make a whole lot of sense to me given their nearly identical read rates. I'll admit that I'm not a storage guru, so perhaps I'm missing something extremely fundamental with this observation. I will assume this is likely the case at this time and will just go with that for the time being, rather than attempt to find fault in Samsung's software interpretation of the OCZ performance evaluation. Either way, I find it very suspect after many of the test results listed above. While my methods might not be 100% perfect, I was intently keen on maintaining as accurate of readings as I could for each and every test.

I also feel that the the Western Digital drive's numbers were a bit low on this chart, especially for as well as it held it's own against the SSDs in other tests.

 

[rubix_1011]

The Cold Boot

No, not in the sense that you left your snow shoes outside your doorstep overnight in the winter.

From power button 'PUSH' to desktop, we're talking every golden second that passes while you either wait for your microwave to nuke your pizza rolls or maybe just having enough time to pull your chair out and sit down before the magic box shows up, ready for your interaction.

I didn't perform any fast-booting tricks like trimming the Windows GUI on boot or eliminating BIOS features or removing the splash image. Nope. We're talking Plain-Jane-out-of-the-box-Joe-User boot up. Without wasting any more seconds for those of you that had to endure a platter HDD boot-time, here are the results:

WD 1TB Black: 2:02
OCZ Vertex3 120GB: 0:47
Samsung 840 Pro 256GB: 0:46

Compared to the SSD boot times, the HDD bootup seemed like I was waiting days to reach the desktop; almost torturous. It truly is amazing when you are watching a timer tick away and how much time does pass when you perform this test one after another.

 

[rubix_1011]

Conclusion and Reflection

In a 32-bit OS system, an SSD can easily provide the extra speed boost that could otherwise be missing from a powerful processor or extra RAM or in a system with a limited upgrade path. An SSD upgrade can also be a fiscally acceptable, low budget update or build option to allow a lower-spec machine to perform at an enhanced level. The same can be said about breathing new life into an older laptop, which often have hardware memory limitations; many older single and dual core laptops are limited to 2GB RAM while many low power solutions also have limited memory capacity.

You'll often see articles and comparisons that mention 'the lowest performing SSDs still grossly outperform traditional hard drives' still remains very true, for the most part. The Western Digital Black drives are some very respectable components in terms of performance, and had I tested other 7200 rpm drives or even a 5400 rpm drive (2.5" or 3.5" form factor) we would have seen more scattered results. Regardless of how you slice it, an SSD offers the most real-world, user level performance enhancement over the best hard drive alternative, for the money.

As SSD prices drop and storage density increases, the price/performance curve of these drives will continue to rise to benefit that of the consumer. Picking up an SSD for an OS drive to run your core applications will provide you with some incredibly noticeable results, especially with day to day computing and web browsing. What I displayed with my testing did show some of those results, but also showed that some of the daily work applications I use are also very CPU and memory responsive and left less to the actual I/O of the physical drives themselves. I actually came away being fairly surprised, but in more than one way. The OCZ really held it's own and even outpaced the Samsung in a few tests. However, the Samsung just felt 'snappier' and more responsive in normal computing functionality but it's difficult to quantify, especially based on the test results listed above. It's an added benefit that the 840 also has 256GB capacity on it's side.

While we are a society of instant gratification, the short wait of the WD 1TB boot is made up by it's relative day to day performance when compared to the two SSDs. However, the Samsung does provide the speed and performance of the SSD world while also providing more capacity that allows it to strike a very solid balance between a small, fast SSD and a larger capacity hard drive. It becomes a difficult choice to be made when considering upgrade or build options for a computer and when faced with these choices, there are some items to consider.

I'm finding that where I've loved using my OCZ Vertex3, I've enjoyed testing the Samsung 840 Pro 256GB just as much. I haven't reached any capacity concerns as of yet, but they'll certainly be less of an issue when I would normally expect the time to arrive, thanks to those extra 'GBs'.

 

[EmbracingHim]

I ran the bundled Samsung benchmark software, Samsung Magician, against each drive to do a comparison of what it determined the speed and performance of each was.

Following the same order as all other tests:

WD 1TB Black

OCZ Vertex3 120GB

Samsung 840 Pro 256GB

Both the Samsung 840 Pro and the OCZ Vertex3 show nearly identical sequential read (555MB/s vs. 552MB/s; Samsung/OCZ) and write scores (513MB/s vs. 519MB/s; Samsung/OCZ) while the Samsung seems to show more than a 3:1 ratio advantage in read IOPS, which doesn't make a whole lot of sense to me given their nearly identical read rates. I'll admit that I'm not a storage guru, so perhaps I'm missing something extremely fundamental with this observation. I will assume this is likely the case at this time and will just go with that for the time being, rather than attempt to find fault in Samsung's software interpretation of the OCZ performance evaluation. Either way, I find it very suspect after many of the test results listed above. While my methods might not be 100% perfect, I was intently keen on maintaining as accurate of readings as I could for each and every test.

I also feel that the the Western Digital drive's numbers were a bit low on this chart, especially for as well as it held it's own against the SSDs in other tests.

Hello,
I didn't install "Samsung Magician", is there a need to install this software on my APPLE Macbook? I know that there was a computer disc from Samsung that came with the Hard Drive. But, APPLE told me I didn't need to use it......... I might need to look into this just to cover my base's........ is the "SAMSUNG Magician" software a program that shows you how the Hard Drive is functioning? I just tried to down load the Samsung Magician software. My computer says it is a Microsoft Windows application. So, it will not work on my system. Thus, I understand why APPLE mentioned I didn't need it.

 

[rubix_1011]

There is not a need to install this software, even on a Windows machine...it is mainly a set of tools to benchmark and evaluate your SSD. It does provide some handy 'one-click' Windows fixes for registry entries and Services, but again, would not apply to an Apple machine.

The only thing I can see being of benefit would be the firmware update functionality, but I think there are some Mac work-arounds on the web from what I have seen...or run the firmware update from a Windows computer, first.