Corsair Performance 3 256GB SSD Review

[AkG]

While SandForce and their second generation controller line is the solid state drive industry’s darling right now, there are several other “fringe” solutions that promise to do an excellent job as well. One of these is the Marvell 88SS9174 which was first introduced to the general consumer market during this year’s CES. Housed within Corsair’s Performance 3 line, it seems to have what it takes to compete with many premier SSD controller lines.

By not opting for one of the eight SandForce controllers available, and instead going with the very potent Marvell 88SS9174-BKK2, Corsair has in effect given consumers another great option to choose from. Naturally, this controller comes with its own set of strengths and weaknesses but this makes Marvell’s solution quite different from anything else out there.

​

Today we are going to look at what the $470 Performance 3 256GB can do and help shed light on whether the Marvell controller can go toe to toe with some of SandForce’s best. This in turn should help you decide if the Performance 3 is a better solution for you than Corsair’s own Force 3 line. After all, $470 is certainly a lot of money to spend and while this is still about $30 less than a Force 3 240GB, we highly doubt price will be the deciding factor for most. This decision will in all likelihood to come down what you plan on using your new drive for and which controller’s strengths best satisfy those needs.

Granted, this drive isn't the latest generation, nor does it use any cutting edge technology but it can be considered a "dark horse" in this industry since most people seem to have forgotten that it actually exists. And yes, the Performance 3 still packs a pretty mean punch.

​

Opening up the Corsair Performance 3 reveals an almost jarring juxtaposition of old and new architectures all blended together onto a 1.8” form factor PCB. Housed inside are eight dual die NAND chips and a Marvell 88SS9174-BKK2 controller chip along an external ram chip for cache.

​

The actual cache chip used on this device is a Nanya branded, NT5CB64M16DP-CF IC with a capacity of a 128 Megabytes. More importantly it is a DDR3-1333 SDRAM chip with CAS Latency of 8 which is supposed to take control of all caching operations. Some of you may remember seeing this kind of layout on several older SSDs and we’re sure it will be just as effective on the Performance 3.

​

Interestingly enough, Corsair didn’t opt for ONFi 25nm NAND. Rather they opted for super high density, 32nm Toshiba TH58TVG8D2FBA89 NAND chips.

​

As with all high performance Corsair Solid State Drives, the Performance 3 line comes with a 2.5” to 3.5” adapter.

 

[AkG]

The TRIM Conundrum & Marvell's Answer

The TRIM Conundrum & Marvell's Answer

To get the absolute best long term performance out of a Solid State Drive your system has to be able to pass on the TRIM command. This is because most modern SSD controller designs are built and designed upon the assumption that the a system will be able to provide TRIM commands to the drive’s controller.

​

For the ATA TRIM command to occur there are many different components which need to work hand in hand for the TRIM command to be sent and received by the SSD controller. The primary lynchpin here is the operating system. If you are running an older OS (pre Windows 7) the operating system itself will not know how to send the TRIM command. In addition, the motherboard’s storage controller itself must be able to process the TRIM command. AMD motherboards which predate the 800-series chipsets and Intel products using a pre-ICH9R setup simply don’t support TRIM regardless of the OS being run. Further complicating things is the prerequisite that the onboard controller be working in AHCI mode as IDE isn’t supported nor can the SSD be part of RAID array. With the exception of some exotic, factory RAID’ed Solid State drives (such as the Revo 3) RAID + SSD means no trim support.

The last crucial link is the drivers which also need to support TRIM and currently Microsoft AHCI, Intel RST or AMD’s latest versions all work well. The easiest way to relate to all of these requirements is to think of each one as a link in a chain: if any one is broken your solid state drive will not receive TRIM commands and could degrade very quickly.

​

The introduction of TRIM really did have an impact upon the perceived longevity of SSDs but there are still many consumers who don’t posses systems capable of passing on the command. Unfortunately, most SSD controller manufacturers design their chips based on the assumption that TRIM will always be a possibility. This means in order to get the best possible long term performance out of a modern SSD, you’ll need a TRIM-supporting system which could mean a costly upgrade to go along with that brand new SSD.

So what about all those consumers who don’t have TRIM capabilities? Before TRIM was implemented, controller designers created routines to minimize and alleviate performance degradation. These self-maintenance routines go by many names but are usually called “Idle Time” or “Background” Garbage collection.

​

This “old school” solution is of course the somewhat perfect answer for non-TRIM environments. Sadly, not all algorithms and routines are created equal and efficiency varies greatly from one controller maker to another. Since SandForce expects people to have TRIM capable systems, their self maintenance routines are very mild a best and thus very slow. So slow in fact that without TRIM, even the mighty SF2281 based drives can see their performance degraded very quickly. In extreme cases pauses and stuttering can even occur.

​

While SandForce and others went the route of assuming their customers had TRIM, Marvell did not. In fact, Marvell seems to have actively focused upon and have designed their controller for consumers who want a large performance boost but don’t want to upgrade their entire system to get it. This is why Marvell controllers – especially this second gen “9174” - have more aggressive self-maintenance routines which start working sooner and faster than their competitors. The high performance, high efficiency self-maintenance routines within the Performance 3 really are the secret weapon in its arsenal but as we will see later in this review, they can act as a double edged sword as well.

 

[AkG]

Testing Methodology

Testing Methodology

Testing a drive is not as simple as putting together a bunch of files, dragging them onto folder on the drive in Windows and using a stopwatch to time how long the transfer takes. Rather, there are factors such as read / write speed and data burst speed to take into account. There is also the SATA controller on your motherboard and how well it works with SSDs & HDDs to think about as well. For best results you really need a dedicated hardware RAID controller w/ dedicated RAM for drives to shine. Unfortunately, most people do not have the time, inclination or monetary funds to do this. For this reason our testbed will be a more standard motherboard with no mods or high end gear added to it. This is to help replicate what you the end user’s experience will be like.

Even when the hardware issues are taken care of the software itself will have a negative or positive impact on the results. As with the hardware end of things, to obtain the absolute best results you do need to tweak your OS setup; however, just like with the hardware solution most people are not going to do this. For this reason our standard OS setup is used. However, except for the Vista load test times we have done our best to eliminate this issue by having the drive tested as a secondary drive. With the main drive being a Phoneix Pro 120GB Solid State Drive.

For synthetic tests we used a combination of ATTO Disk Benchmark, HDTach, HD Tune, Crystal Disk Benchmark, IOMeter, AS-SSD and PCMark Vanatage.

For real world benchmarks we timed how long a single 10GB rar file took to copy to and then from the devices. We also used 10gb of small files (from 100kb to 200MB) with a total 12,000 files in 400 subfolders.


For all testing a Asus P8P67 Deluxe motherboard was used, running Windows 7 64bit Ultimate edition (or Vista for boot time test). All drives were tested using AHCI mode using Intel RST 10 drivers.

All tests were run 4 times and average results are represented.

In between each test suite runs (with the exception being IOMeter which was done after every run) the drives are cleaned with either HDDerase, SaniErase or OCZ SSDToolbox and then quick formatted to make sure that they were in optimum condition for the next test suite.


Processor: Core i5 2400
Motherboard: Asus P8P67 Deluxe
Memory: 8GB Mushkin DDR3 1300
Graphics card: Asus 5550 passive
Hard Drive: 1x Seagate 3TB XT, OCZ 120GB RevoDrive
Power Supply: XFX 850


SSD FIRMWARE (unless otherwise noted):

OCZ Vertex: 1.6
OCZ Vertex 2 100GB: 1.33
Mushkin Callisto Deluxe 40GB: 3.4.0
Corsair Force F90: 2.0
OCZ Vertex 3 240GB: 1.11
Crucial C300 128GB: 006
Corsair Force 3 GT 120GB: 1.2
Corsair Force 3 120GB: 1.2
Patriot Pyro 120GB: 3.1.9
Corsair Performance 3 256GB: 1.1

 

[AkG]

Read Bandwidth / Write Performance

Read Bandwidth

For this benchmark, HDTach was used. It shows the potential read speed which you are likely to experience with these hard drives. The long test was run to give a slightly more accurate picture. We don’t put much stock in Burst speed readings and thus we no longer included it. The most important number is the Average Speed number. This number will tell you what to expect from a given drive in normal, day to day operations. The higher the average the faster your entire system will seem.

​

While 466+ Megabytes per second may seem a touch slow compared to some other 256GB high performance Solid State Drives, it is still extremely fast. When you get past a certain point we truly doubt anyone – besides benchmark junkies – will notice thirty or forty Megabytes per second one way or the other. Nonetheless, the Performance 3 still leaves the C300 in the dust.

Write Performance

For this benchmark HD Tune Pro was used. To run the write benchmark on a drive, you must first remove all partitions from that drive and then and only then will it allow you to run this test. Unlike some other benchmarking utilities the HD Tune Pro writes across the full area of the drive, thus it easily shows any weakness a drive may have.

​

As with sequential reads, the Performance 3 does post some very good numbers. Much of the difference between the Marvell controller and SandForce based solutions comes down to the fact that SandForce drives get a fairly significant boost from easily compressible data which is exactly what HD Tune uses.

However, some of the performance change between the two types of drives could also be due to those extremely aggressive garbage collection routines Marvell is using. It seems they aren’t even waiting for large idle periods but rather taking cycles away from the controller while the drive is still being accessed.

 

[AkG]

ATTO Disk Benchmark

ATTO Disk Benchmark

The ATTO disk benchmark tests the drives read and write speeds using gradually larger size files. For these tests, the ATTO program was set to run from its smallest to largest value (.5KB to 8192KB) and the total length was set to 256MB. The test program then spits out an extrapolated performance figure in megabytes per second.

​

While the read curves look perfectly fine, the same can't be said of the write performance. If you look closely at the curve, you will notice that just past the 32KB point, the upwards trends to get severely curtailed. This we believe is a direct result of those ultra aggressive garbage collection routines kicking in too early.

As with most things, there is only so many controller cycles to go around and any cycles used on garbage collection can't be used for other things. This is the double edge sword of the Marvell 9174 controller running the stock firmware.

 

[AkG]

Crystal DiskMark / PCMark 7

Crystal DiskMark

Crystal DiskMark is designed to quickly test the performance of your hard drives. Currently, the program allows to measure sequential and random read/write speeds; and allows you to set the number of tests iterations to run. We left the number of tests at 5 and size at 100MB.

​

We are not going to sugar coat things: these numbers are quite low. The sequential and even mid file size numbers are more than decent, but both read and write performance at a 4K que depth are low and the 32 que depth numbers are also disappointing.

If Corsair and / or Marvell can tone down the garbage collection a touch, we can see the glimmers of one extremely powerful device for even TRIM enabled environments. However, this controller simply dedicated too many cycles for self maintenance.

PCMark 7

While there are numerous suites of tests that make up PCMark 7, only one is pertinent: the HDD Suite. The HDD Suite consists of numerous tests that try and replicate real world drive usage. Everything from how long a simulated virus scan takes to complete, to MS Vista start up time to game load time is tested in these core tests; however we do not consider this anything other than just another suite of synthetic tests. For this reason, while each test is scored individually we have opted to include only the overall score.

​

Thanks in no small part to its reliance on 32nm NAND, PCMark 7 seems to approve of this drive. It may not get blockbuster large numbers, but hitting above 4.5K on this test is darn impressive nonetheless.

 

[AkG]

AS-SSD / Access Time

AS-SSD

AS-SSD is designed to quickly test the performance of your drives. Currently, the program allows to measure sequential and small 4K read/write speeds as well as 4K file speed at a queue depth of 6. While its primary goal is to accurately test Solid State Drives, it does equally well on all storage mediums it just takes longer to run each test as each test reads or writes 1GB of data.

​

The usual battery of AS-SSD tests really doesn’t shine any new light on things. The garbage collection is aggressive and it costs this controller some low end performance. With that being said, the performance can be downright impressive when sequential numbers are factored into the equation.

Access Time

To obtain an accurate reading on the read and write latency of a given drive, AS-SSD was used for this benchmark. A low number means that the drive’ data can be accessed quickly while a high number means that more time is taken trying to access different parts of the drive.

​

As expected the read and write latency is more than adequate.

 

[AkG]

Anvil Storage Utilities Pro

Anvil Storage Utilities Pro

Much like AS-SSD, Anvil Pro was created to quickly and easily – yet accurately – test your drives. While it is still in the Beta stages it is a versatile and powerful little program. Currently it can test numerous read / write scenarios but two in particular stand out for us: 4K que depth of 4 and 4K que depth of 16. A que depth of four along with 4K sectors can be equated to what most users will experience in an OS scenario while 16 depth will be encountered only by power users and the like. We have also included the 4k que depth 1 results to help put these two other numbers in their proper perspective. All settings were left in their default states and the test size was set to 1GB.

​

It seems that even at lower que depths the 4K performance of this drive is severely curtailed. With all that being said, the market niche for these drives is consumers who don’t have TRIM and will most likely be coming from a standard Hard Drive background. While a SandForce drive may or may not provide a great first time experience for these customers, the Performance 3 on the other hand should be more than adequate. This really is the flip side to such obsessive compulsive garbage collection.

 

[AkG]

IOMETER

IOMETER

IOMeter is heavily weighted towards the server end of things, and since we here at HWC are more End User centric we will be setting and judging the results of IOMeter a little bit differently than most. To test each drive we ran 5 test runs per HDD (1,4,16,64,128 que depth) each test having 8 parts, each part lasting 10 min w/ an additional 20 second ramp up. The 8 subparts were set to run 100% random, 80% read 20% write; testing 512b, 1k, 2k,4k,8k,16k,3xk,64k size chunks of data. When each test is finished IOMeter spits out a report, in that reports each of the 8 subtests are given a score in I/Os per second. We then take these 8 numbers add them together and divide by 8. This gives us an average score for that particular que depth that is heavily weighted for single user environments.

​

These numbers are OK at best but we certainly wouldn't want to use this device in a more server orientated role. Of course, the Performance 3 was never meant to fulfill such a role, and thus these results can't be held against it.

 

[AkG]

Vista Start Up / Adobe CS5 Load Time

Vista Start Up

When it comes to hard drive performance there is one area that even the most oblivious user notices: how long it takes to load the Operating System. While all the other tests were run with a Windows 7 operating system, this particular test uses another older test bed's “day to day” OS (copied over to our new testbed) which has accumulated a lot of crud over the months from installs and removals. We chose the Anti-Virus splash screen as our finish line as it is the last program to be loaded on start up.

​

While forty three seconds may not sound like all that much, this is awfully quick. In fact, not that long ago it would have been considered best in class performance. Considering the TRIM enabled environment is not the environment Marvell’s engineers envisioned for this drive,we're suitably impressed by these results.

Adobe CS5 Load Time

Photoshop is a notoriously slow loading program under the best of circumstances, and while the latest version is actually pretty decent, when you add in a bunch of extra brushes and the such you get a really great torture test which can bring even the best of the best to their knees. Let’s see how our review unit faired in the Adobe crucible.

​

As with the boot times, the Adobe load times are very respectable.