What's new
  • Please do not post any links until you have 3 posts as they will automatically be rejected to prevent SPAM. Many words are also blocked due to being used in SPAM Messages. Thanks!

OCZ RevoDrive 3 X2 480GB PCI-E SSD Review

Status
Not open for further replies.

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
OCZ’s solid state drives release schedule is progressing at a breakneck pace and even we’ve had a hard time keeping up with them. Be it consumer or enterprise, value or performance - or even some blend of all the above – they will in all likelihood have a model tailor made for your needs. The all new RevoDrive 3 x2 480GB is no exception to this axiom. Unlike most SandForce based drives we have looked at, the RevoDrive 3 x2 is built around a PCI-E interface and is tailor made for price conscious workstation consumers who need performance and features only this class of storage device can offer. But before we go on, remember one thing: its focus upon the workstation market’s pricing structure still puts it well beyond the means of most home users.


The RevoDrive 3 x2 accomplishes its goals by offering unprecedented levels of performance while including features never before seen on OCZ PCI-E SSD like TRIM and a non-RAID 0 setup. Like the original RevoDrive x2 models, the “x2” in the name denotes a secondary daughter card attached to the PCB which doubles both the capacity and number of controllers (this particular example boasts four controllers). The RevoDrive 3 x2 comes in three sizes: large (240GB), massive (480GB) and enormous (960GB) which range from just over $700 to an astonishing $3,200 for the 960GB version. The 480GB drive we are reviewing today goes for around $1,600. Hey, no one ever said cutting edge performance would be cheap!

OCZ_RevoDrive3x2_board.jpg

On paper, the RevoDrive 3 x2 480GB houses a custom ARM based Serial Attached SCSI (SAS) RAID controller coupled with four SF2281 SATA controllers and a veritable fortune in NAND ICs. This is all wrapped up on a PCI-E X4 gen 2 form factor board along with an attached daughter card.

So where are the cost saving measures here? Well, OCZ decided to go with ONFi 1.0 NAND and a non native PCI-E solution rather than ultra expensive ONFI 2.0 modules and a high-end controller interface.

OCZ_RevoDrive3x2_board2.jpg

In the past, ONFi 1.0 NAND based SF2281 drives have vacillated from being absolutely great performers to less than optimal solely based on usage patterns. Considering this 480GB drive has a sky high asking price, the temperamental aspects of past drives do raise some concerns but as we will see on the upcoming pages this isn’t to say OCZ has sitting back and doing nothing about the perceived limitations of their high end solution.

zoom_sm.jpg
nand2_sm.jpg

Even though it uses a PCI-E x4 form factor and has NAND and controllers are laid bare for the world to see, the Revodrive’s architecture is actually quite similar to standard SATA-based drives. Both the primary PCB and its daughter card contain a pair of SF2281 controllers per board, with each controller populated with and responsible for 16 ONFi 1.0 Asynchronous NAND ICs. In grand total there is one SuperScale processor, one SAS to PCIE x4 gen 2 bridge, 64 NAND ICs and four SF2281 controllers. In essence this drive is setup with the equivalent of four 120GB Agility 3 drives all being funneled into a custom SAS controller and PCI-E bridge.

OCZ_RevoDrive3x2_controller_sm.jpg

Along with several solid state drives worth of hardware, the main board itself also houses the small heatsink covering its ARM controller. While OCZ is coy about who makes this chip recent rumors have it pegged as a Marvell controller with custom firmware and happens to be the true difference between the original RevoDrive and this new RevoDrive 3 lineup.

The RevoDrive original used a 4 port SiL RAID controller and an additional secondary PCI-E / SATA bridge chip. The RevoDrive 3 x2 also uses a two chip solution but the specifications are very, very different this time around. The first part of this equation is the 4-port 6Gbps SAS to generation 2 PCIe x4 bridge while the second integral part is what OCZ dubs a VCA 2.0 enabled, “SuperScale processor”. It may sound like complicated stuff but we’ll get into these items a bit later in the review. For the time being, let’s just say that OCZ has designed a very, very unique PCI-E SSD here.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
A Look at DuraWrite, RAISE and More

A Look at DuraWrite, RAISE and More



Corsair_Force_sandforce_logi.jpg

Let’s start with the white elephant in the room and explain why this 480GB drive is in reality a 512GB drive. The OCZ RevoDrive 3 x2 has sixty four 8GB NAND chips onboard which gives it a capacity of 512GB, but is seen by the OS as 4800GB. Manufacturers use this to help increase IOPS performance and also extend life via wear leveling (as there are always free cells even when the drive is reported as “full”) and even durability since the drive has cells in reserve it can reassign sectors to as the “older” cells die.

dura.jpg


As we said, over-provisioning is usually for wear leveling and ITGC as it gives the controller extra cells to work with for keeping all the cells at about the same level of wear. However, this is actually not the main reason SandForce sets aside so much. Wear leveling is at best a secondary reason or even just a “bonus” as this over-provisioning is mainly for the Durawrite and RAISE technology.

Unlike other solid state drives which do not compress the data that is written to them, the SandForce controller does real time loss-less compression. The upside to this is not only smaller lookup tables (and thus no need for off chip cache) but also means less writes will occur to the cells. Lowering how much data is written means that less cells have to be used to perform a given task and this should also result in longer life and even fewer controller cycles being taken up with internal house cleaning (via TRIM or ITGC).

Corsair_Force_Fact5.jpg


Longevity may be a nice side effect but the real purpose of this compression is so the controller has to use fewer cells to store a given amount of data and thus has to read from fewer cells than any other drive out there (SandForce claims only .5x is written on average). The benefit to this is even at the NAND level storage itself is the bottleneck for any controller and no matter how fast the NAND is, the controller is faster. Cycles are wasted in waiting for data retrieval and if you can reduce the number of cycles wasted, the faster an SSD will be.

Compressing data and thus hopefully getting a nice little speed boost is all well and fine but as anyone who has ever lost data to corruption in a compressed file knows, reliability is much more important. Compressing data means that any potential loss to a bad or dying cell (or cells) will be magnified on these drives so SandForce needed to ensure that the data was kept as secure as possible. While all drives use ECC, to further ensure data protection SandForce implemented another layer of security.

Corsair_Force_Fact4.jpg


Data protection is where RAISE (Redundant Array of Independent Silicon Elements) comes into the equation. All modern SSDs use various error correction concepts such as ECC. This is because as with any mass produced item there are going to be bad cells while even good cells are going to die off as time goes by. Yet data cannot be lost or the end user’s experience will go from positive to negative. SandForce likes to compare RAISE to that of RAID 5, but unlike RAID 5 which uses a parity stripe, RAISE does not. SandForce does not explicitly say how it does what it does, but what they do say is on top of ECC, redundant data is striped across the array. However, since it is NOT parity data there is no added overheard incurred by calculating the parity stripe.

Corsair_Force_Fact2.jpg


According to SandForce’s documentation, not only individual bits or even pages of data can be recovered but entire BLOCKS of data can be as well. So if a cell dies or passes on bad data, the controller can compensate, pass on GOOD data, mark the cell as defective and if necessary swap out the entire block for a spare from the over-provisioning area. As we said, SandForce does not get into the nitty-gritty details of how DuraWrite or RAISE works, but the fact that it CAN do all this means that it most likely is writing a hash table along with the data.

SandForce is so sure of their controller abilities that they state the chances of data corruption are not only lower than that of other manufactures’ drives, but actually approaches ZERO chance of data corruption. This is a very bold statement, but only time will tell if their estimates are correct. In the mean time, we are willing to give the benefit of the doubt and say that at the very least data corruption is as unlikely with one of these products as it is on any modern MLC drive.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
Say Hello To VCA 2.0

Say Hello To VCA 2.0


Virtualized Controller Architecture 2.0 - or VCA 2.0 for short - is a new flash virtualization layer created by OCZ for their enterprise grade devices. More importantly this new layer is what resides between the RAID controller hardware on the RevoDrive3 x2 and the operating system to facilitate communication and also happens to provide features never before seen on the OCZ RevoDrive series.

vca.jpg


The original RevoDrive also made use of a RAID controller - albeit a different one - but since VCZ 2.0 hadn't quite made it into OCZ's product stack, it was unable to do many things that users of the RevoDrive 3 will take for granted. The features packed into the SuperScale processor include firmware updating through OCZ's Toolbox software, the reporting of SMART information and implementation of OCZ’s Complex Command Queuing Structure (CCQS) which utilizes both Native and Tagged Command Queuing.

Most importantly though, VCA 2.0 allows for the RevoDrive 3 x2 to support the TRIM command along with Queue Balancing Algorithm (QBA) which balances drive loading across all four controllers.

sas.jpg


What OCZ calls QBA - or Queue Balancing Algorithm – underlines how much the RevoDrive 3 x2 has changed when compared to previous versions. At its heart, the unnamed “SuperScale” processor / SAS controller is a RAID controller but the custom firmware and VCA 2.0 implementation an entirely different beast than the original's RAID 0 setup.

Rather than having the SAS controller stripping data across all four SF2281 controllers, it uses real time dynamic load balancing and sends IO commands to each of the controllers on a case by case basis. For example, if one of the four “drives” on the PCBs is busy doing a read or write, another can be implementing an emergency block erase, while a third and even fourth could be doing something completely different. In theory four commands can be carried out at the exact same time without the need for the custom deep command queuing (aka CCQS) this drive implements.

sata.jpg


While a unique configuration and custom software layer is indeed interesting, it is the TRIM capabilities on these high end PCI-E solutions which many have been waiting for. Sadly, while this device has the POTENTIAL to pass on the TRIM command to the four SF2281 controllers, it lacks this ability right now.

It is important to remember that the RevoDrive 3 is basically a SCSI device (the controller is a SAS or Serial Attached SCSI controller) and accepts SCSI commands over PCIe. This means the operating system will have to send the TRIM command via the MS StorPort SCSI driver rather than the standard AHCI method. Unfortunately Microsoft’s StorPort does not currently support sending the TRIM command so the RevoDrive is still S.O.L. in this area.

ms.jpg


While OCZ states they are working closely with Microsoft to get an updated driver as soon as possible, but we don't realistically expect anything soon.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
RevoDrive 3 x2 and CPU Utilization

RevoDrive 3 x2 and CPU Utilization


Since the RevoDrive 3 x2’s SuperSale processor is at heart a software rather than hardware-based solution, higher CPU utilization is to be expected but exactly how much is the question.

To obtain an accurate picture of exactly how much CPU horsepower this device truly requires we have configured the RevoDrive 3 x2 480GB device as a secondary “D” data drive and let the system idle with a minimum of processes running. Using Windows' built in Performance Monitor we can see exactly how much processing power is being dedicated to a given storage solution. For comparison's sake we have also include the results from the original RevoDrive 120GB and also a SATA 6GB/s Solid State. Lastly, we run Crystal DiskMark while monitoring CPU utilization.

The processor used is a Core i5 2500K which is running at stock speed.


idle_v3.jpg

idle_revo1.jpg


idle_revo3.jpg

idle_revo3_cdm.jpg



They say that a picture is worth a thousand words, but in this case it is worth about a 50% increase in CPU utilization. With no PCI-E based storage device attached to the system, the idle state is basically as close to zero as you can come for a modern Operating System. While the RevoDrive 120GB does indeed have much greater utilization with an average of 22%, this is down right tame in comparison to what the RevoDrive 3 demands.

Regardless of the processor you are using, the RevoDrive 3 will never let your system enter a low power state. While Turbo Boost never kicked in, having a single drive use 75% percent of a modern quad core CPU running at 3.3GHZ system is simply mind boggling. Either VCA 2.0 simply needs to be refined a bit more or the new controller needs some tweaking. Remember, these numbers are with an idling system that is doing absolutely nothing.

Thankfully, modern Intel systems do have Turbo Boost and with the CPU speed boosted to 3.7GHz it “only” uses 87% during Crystal DiskMark and only hit around 95% when dealing with 64 queue depths. Luckily, throughout our testing we never ran into a true CPU bottleneck but for the first time ever, we can easily foresee a high speed quad core CPU holding back things in our storage test bed rather than the PCH or the storage device itself.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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 2500
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 2 100GB: 1.33
OCZ Vertex 3 MI 240GB: 2.1.3
Corsair Force 3 GT 120GB: 1.3
Patriot Pyro 120GB: 3.2.0
Kingston HyperX 240GB: 3.2.0
Crucial M4 256GB: 009
Mushkin Chronos 120GB: 3.2.0
OCZ Agility 3 240GB: 2.1.3
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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.

read.jpg


While good, these results underscore how difficult it will be to access the full power of this device. We know that in sequential reads it should be good for three times these numbers, but it appears that in single queue depth tasks -such as single large sequential reads- most of the four controllers will be idle at any given time.

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.

write.jpg



Once again these numbers are certainly outstanding for an ONFi 1.0 based drive, but they are no where near where you would want them to be. This is partially due to HD Tune using slightly older (but no outdated) write calls but once again we have to say that the RevoDrive 3 just isn't designed to excel in these particular tests.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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.

atto_r.jpg


atto_w.jpg


Once you get beyond the simply mind numbing large file performance of this drive, you can see that the low end reflects what the read and write charts on the previous page showed. This may be due to the higher than typical demands the RevoDrive 3 x2 places on the CPU or the fact that it relies on ONFi 1.0 NAND. It seems like Load balancing and other fancy maneuvers can only get a drive so far in certain tasks. Nonetheless, some of the results here simply beggar belief.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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.

cdm_r.jpg


cdm_w.jpg


The reason these numbers are only half – or less – of the rated performance is because of the Asynchronous NAND ICs the RevoDrive 3 x2 relies upon. As we have seen in the past, when dealing with highly incompressible data like what's used by Crystal Diskmark, the SF2281 / ONFi 1.0 combination seem to choke. This is especially true of the single queue depth 4K numbers where only one of the four controllers is doing most of the work at any one time and the SuperScale processor is unable to obfuscate this well known limitation. When the queue depth gets deeper the performance increases dramatically as the load balancing kicks in and more effectively masks it.


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.

pcm7.jpg


While a touch north of 6,100 points is nothing to sneeze at, we were expecting a bit more in this test as well. OCZ claims that if we had switched to a 2600K processor, the 8,500 point mark could have been surpassed. It is anyone's guess whether or not this underscores exactly how much power the RevoDrive 3 x2 really needs. But what we can say is that moving from a 4 thread to an 8 thread processor will increase the PCMark score regardless of the storage subsystem being used.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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 64. 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.

asd_r.jpg


asd_w.jpg

As with Crystal DiskMark, these numbers are certainly impressive but it seems the 4K reads and writes are somewhat lacking and are only slightly above the results posted by the Vertex 3 Max IOPS . Nonetheless, its other numbers are simply staggering.


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’s data can be accessed quickly while a high number means that more time is taken trying to access different parts of the drive.

random.jpg


You certainly can’t argue with performance numbers like this. It appears that the hybrid RAID / VCA 2.0 configuration results in no increased latency and in fact is able to reduce it via load balancing across all four controllers.
 
Last edited by a moderator:

AkG

Well-known member
Joined
Oct 24, 2007
Messages
5,270
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 queue depth of 4 and 4K queue depth of 16. A queue 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 queue 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.

anvil_r.jpg


anvil_w.jpg


As has become the reoccurring theme of the RevoDrive 3 x2 and our synthetic benchmarks, this drive has only moderately impressive single queue depths but as things pick up, so too does this OCZ drive. The result is some otherworldly numbers in our Write tests.
 
Last edited by a moderator:
Status
Not open for further replies.

Latest posts

Twitter

Top