Review: Corsair’s Neutron and Neutron GTX solid-state drives

In the world of solid-state drives, Corsair has a reputation for, well, getting around. The enthusiast-oriented component maker has hooked up with all the big names in the controller business: Indilinx, Marvell, SandForce, and Samsung. It even had a tryst with JMicron, a dalliance that can probably be blamed on a combination of alcohol and low self esteem.

Corsair’s willingness to experiment with controller technologies has allowed the company to hedge its bets, offering multiple product lines based on different solutions. The Force series has used the latest SandForce controllers, while the Performance line has featured Marvell silicon. The Indilinx-based Nova model still serves the budget market to this day, and there are Accelerator drives designed for caching, too.

All of those drives are part of a diverse family of offerings, but none of them are really unique in the wider context of the market. The fact is, everyone and his mother is making SSDs these days. Most vendors, Corsair included, are pulling from the same limited selection of NAND, controllers, and stock firmware.

Corsair’s Neutron SSDs are different, though. They’re the first and thus far only drives on the market using Link_A_Media Devices’ new LM87800 controller. Corsair has exclusive access to the controller right now, and it has deployed the chip alongside two flavors of NAND memory in the Neutron and Neutron GTX. We’ve tested both to see how they stack up against not only each other, but also their peers. Before diving into those results, let’s meet Corsair’s new muse.

Inside the Neutron

Based in Santa Clara, California, Link_A_Media Devices got its start in 2004 and specializes in storage-related SoCs. Yes, those are underscores in the name—how edgy. To avoid having to type them constantly, I’m going to refer to the firm as LAMD. Besides, the company has sold out already. LAMD was acquired by memory maker SK Hynix earlier this year, not long after the then-upcoming Neutrons were announced as the first consumer-grade SSDs based on LAMD controllers. According to Corsair, LAMD SSD silicon was previously used only in enterprise-class products.

Both Neutrons rely on the same LAMD LM87800 controller. This chip features dual ARM processor cores, one dedicated to the host and the other to the NAND. On the host side is a 6Gbps Serial ATA interface. On the flash side, the chip supports NAND from the ONFI and Toggle DDR camps.

LAMD is quick to tout eBoost, a combination of error correction and “adaptive signal estimation techniques” that purportedly enables the LM87800 to extend the endurance of both kinds of NAND. There’s no SandForce-like write compression involved here, just advanced signal processing. eBoost promises enterprise-class reliability for client SSDs, including those based on the upcoming 1x-nm generation of flash memory.

The LM87800 also has a built-in redundancy scheme that protects the drive from physical flash failures. Storing redundancy data requires some extra capacity, which explains why the Neutrons come in 120GB and 240GB sizes rather than the more traditional 128GB and 256GB. Corsair has a 480GB version of the GTX in the works, as well, and the controller supports up to a terabyte of flash.

Model Max
sequential (MB/s)
4KB
random write (IOps)
Price
Read Write
Neutron
120GB
555 211 80,000 $120
Neutron 240GB 555 370 85,000 $210
Neutron GTX 120GB 555 330 80,000 $140
Neutron GTX 240GB 555 511 85,000 $250

Like its contemporary counterparts, the LM87800 has eight memory channels. Each channel can connect to up to four chips, allowing the controller to address a maximum of 32 individual NAND dies. The 240GB Neutrons tap that full capacity, serving up at least 32 NAND dies each. With fewer NAND connections, the 120GB models can’t take advantage of as much controller-level parallelism, which is why their write performance ratings are lower. The drops in claimed sequential write throughput are much larger than the small dips in 4KB random write performance.

As the write performance ratings clearly illustrate, the standard Neutron is slower than the GTX. The Neutron uses ONFI-compliant synchronous NAND from Micron, while the Neutron GTX relies on Toggle DDR flash made by Toshiba. The memory chips for both come from 2x-nm nodes: the Micron stuff is fabbed on a 25-nm process, and the Toshiba chips are built using 24-nm tech.

The Micron NAND has two 64Gb dies per package, which means the Neutron needs 16 individual packages to offer 240GB of total capacity. Corsair splits the NAND evenly between the two sides of the Neutron’s circuit board. With eight 32Gb dies in each of its Toshiba flash packages, the GTX can hit 240GB while populating only one side of the circuit board with eight islands of NAND.

We’re used to seeing 8-16 NAND packages on SSDs of this capacity. However, we’re not using to seeing the flash laid out on such stubby circuit boards.

The circuit boards on both drives are about 20% shorter than usual, as evidenced by the leftover space inside the Neutrons’ enclosures. Incidentally, the cases eschew the screws used by most SSDs in favor of snap-on covers that probably save a cent or two. If you want to see the Neutrons in the nude, the covers can be pried off using a flat-head screwdriver. Amusingly, doing so won’t disturb the warranty stickers covering the screw holes on the undersides of the drives.

Corsair offers five-year warranties with both Neutrons, equaling the best coverage we’ve seen among consumer-grade SSDs. Three-year warranties are more common, and the extra coverage will surely help to justify the Neutrons’ asking prices. The GTX is notably more expensive than some of its direct competitors, which have equivalent capacities but cost $200 or less.

Our testing methods

We have a full suite of performance results for literally dozens of different SSDs, but today, we’ve narrowed the field to include only the highest-capacity models available in the Benchmarking Sweatshop. Most of these drives offer 240-300GB of capacity, so they’re comparable to the Neutrons. The OCZ Octane and Vertex 4 drives weigh in at a heftier 512GB, but those models don’t carry substantially higher performance ratings than their 256GB counterparts. That’s true for all the SSDs we tested; models in the 240-256GB range tend to have nearly identical performance ratings to their 480-512GB siblings. Even though we have a mix of capacities, our results should give us a good sense of how each SSD performs in its optimal configuration.

We’ve included a Western Digital Caviar Black mechanical desktop drive for reference, which gives us more than enough fodder for overstuffed graphs. Our test methods and systems haven’t changed in probably a little too long, so the scores on the following pages can be compared to those in any of our storage reviews dating back to last September. Ideas for tests to include in our next storage suite are percolating already.

If you’re familiar with our test methods and hardware, the rest of this page is filled with nerdy details you already know; feel free to skip ahead to the benchmark results. For the rest of you, we’ve summarized the essential characteristics of all the drives we’ve tested in the table below. Our collection of SSDs includes representatives based on the most popular SSD configurations on the market right now.

Interface Cache Flash controller NAND
Corsair Force Series 3 240GB 6Gbps NA SandForce SF-2281 25-nm Micron async MLC
Corsair Force Series GT 240GB 6GBps NA SandForce SF-2281 25-nm Intel sync MLC
Corsair Neutron 240GB 6GBps 256MB LAMD LM87800 25-nm Micron sync MLC
Corsair Neutron GTX 240GB 6GBps 256MB LAMD LM87800 26-nm Toshiba Toggle DDR
Crucial m4 256GB 6Gbps 256MB Marvell 88SS9174 25-nm Micron sync MLC
Intel 320 Series 300GB 3Gbps 64MB Intel PC29AS21BA0 25-nm Intel MLC
Intel 510 Series 250GB 6Gbps 128MB Marvell 88SS9174 34-nm Intel MLC
Intel 520 Series 240GB 6Gbps NA SandForce SF-2281 25-nm Intel sync MLC
OCZ Octane 512GB 6Gbps 512MB Indilinx Everest 25-nm Intel sync MLC
OCZ Vertex 4 512GB 6Gbps 1GB Indilinx Everest 2 25-nm Intel sync MLC
Samsung 830 Series 256GB 6Gbps 256MB Samsung S4LJ204X01 2x-nm Samsung Toggle DDR
WD Caviar Black 1TB 6Gbps 64MB NA NA

We used the following system configuration for testing:

Processor Intel Core i7-2500K 3.3GHz
Motherboard Asus P8P67 Deluxe
Bios revision 1850
Platform hub Intel P67 Express
Platform drivers INF update 9.2.0.1030

RST 10.6.0.1022

Memory size 8GB (2 DIMMs)
Memory type Corsair Vengeance DDR3 SDRAM at 1333MHz
Memory timings 9-9-9-24-1T
Audio Realtek ALC892 with 2.62 drivers
Graphics Asus EAH6670/DIS/1GD5 1GB with Catalyst 11.7 drivers
Hard drives Corsair Force 3 Series 240GB with 1.3.2 firmware

Corsair Force Series GT 240GB with 1.3.2 firmware

Crucial m4 256GB with 010G firmware

Intel 320 Series 300GB with 4PC10362 firmware

Intel 510 Series 250GB with PWG2 firmware

WD Caviar Black 1TB with 05.01D05 firmware

OCZ Octane 512GB with 1313 firmware

Samsung 830 Series 256GB with CXM03B1Q firmware

Intel 520 Series 240GB with 400i firmware

OCZ Vertex 4 512GB with 05.10.30 firmware

Corsair Neutron 240GB with M206 firmware

Corsair Neutron GTX 240GB with M206 firmware

Power supply Corsair Professional Series Gold AX650W
OS Windows 7 Ultimate x64

Thanks to Asus for providing the systems’ motherboards and graphics cards, Intel for the CPUs, Corsair for the memory and PSUs, Thermaltake for the CPU coolers, and Western Digital for the Caviar Black 1TB system drives.

We used the following versions of our test applications:

Some further notes on our test methods:

  • To ensure consistent and repeatable results, the SSDs were secure-erased before almost every component of our test suite. Some of our tests then put the SSDs into a used state before the workload begins, which better exposes each drive’s long-term performance characteristics. In other tests, like DriveBench and FileBench, we induce a used state before testing. In all cases, the SSDs were in the same state before each test, ensuring an even playing field. The performance of mechanical hard drives is much more consistent between factory fresh and used states, so we skipped wiping the HDDs before each test—mechanical drives take forever to secure erase.

  • We run all our tests at least three times and report the median of the results. We’ve found IOMeter performance can fall off with SSDs after the first couple of runs, so we use five runs for solid-state drives and throw out the first two.

  • Steps have been taken to ensure that Sandy Bridge’s power-saving features don’t taint any of our results. All of the CPU’s low-power states have been disabled, effectively pegging the 2500K at 3.3GHz. Transitioning in and out of different power states can affect the performance of storage benchmarks, especially when dealing with short burst transfers.

The test systems’ Windows desktop was set at 1280×1024 in 32-bit color at a 75Hz screen refresh rate. Most of the tests and methods we employed are publicly available and reproducible. If you have questions about our methods, hit our forums to talk with us about them.

HD Tune — Transfer rates

HD Tune lets us present transfer rates in a couple of different ways. Using the benchmark’s “full test” setting gives us a good look at performance across the entire drive rather than extrapolating based on a handful of sample points. The data created by the full test also gives us fodder for line graphs, which we’ve split by drive maker. You can click the buttons below each line graph to switch between the different SSDs.

To make the graphs easier to interpret, we’ve greyed out the mechanical drive. The SSD results have been colored by drive maker, with the Neutrons set apart from Corsair’s other offering in different shades of green.


Impressive. The Neutrons get off to a good start, nearly equaling the sequential read throughput of the fastest drive we’ve tested, Samsung’s 830 Series. There’s essentially no difference between the standard and GTX versions of the Neutron in this test. Even the line graph shows overlapping results. Things get a little more interesting with sequential writes, however.


The Neutrons post painfully low average write speeds, and it takes no more than a quick glance at the line graph to determine why. Both drives alternate between prolonged lows around 100-150MB/s and brief spikes to over 450MB/s. We see similarly erratic behavior from the drives based on recent SandForce controllers, except the lows aren’t nearly as deep, resulting in higher average speeds.

HD Tune runs on unpartitioned drives, which isn’t always an ideal case for SSDs. For another perspective, we ran a handful of the SSDs through CrystalDiskMark’s sequential transfer rate tests, which call for partitioned drives. We used the app’s default settings: a 1GB transfer size with randomized data.

CrystalDiskMark’s write speed results show no indication of inconsistent performance from the Neutrons. The GTX leads the field by a fair margin, while the standard Neutron pulls up behind the OCZ Vertex 4 and Samsung 830 Series. This time around, the difference in write speeds between the two Neutrons is quite substantial—just under 120MB/s.

The Neutrons offer similar performance in CrystalDiskMark’s sequential read speed test, with the GTX posting a slightly lower score than its sibling. Both Neutrons finish off the podium here.

HD Tune’s burst speed tests are meant to isolate a drive’s cache memory.

Although the Neutrons’ 256MB DRAM caches score very well in HD Tune’s read burst speed test, the Neutrons are well behind a number of other SSDs when it comes to burst writes. Interestingly, about half of the solid-state drives, including the Neutrons, have slower write burst speeds than our mechanical hard drive.

HD Tune — Random access times

In addition to letting us test transfer rates, HD Tune can measure random access times. We’ve tested with four transfer sizes and presented all the results in a couple of line graphs. We’ve also busted out the 4KB and 1MB transfers sizes into bar graphs that should be easier to read without the presence of the mechanical drive.

As the line graph illustrates, the difference in access times between solid-state and mechanical storage typically works out to at least one order of magnitude. The gap narrows somewhat at the largest 1MB transfer size, but even then, the access times of our mechanical hard drive are at least four times higher than those of the SSDs.

Focusing on the 4KB and 1MB transfer sizes reveals no difference in access times between the two Neutron SSDs. The drives may sit near the middle of the pack, but they’re only marginally slower than the fastest examples in each test.

Our solitary mechanical hard drive continues to be embarrassed in HD Tune’s random write speed tests. The comparatively instantaneous access times of SSDs are the reason they feel more responsive than traditional hard drives.

Among the other SSDs, the Neutrons are fairly competitive. They offer nearly identical access times with 4KB random writes and are less than one hundredth of a millisecond shy of first place. In the 1MB test, the GTX has much quicker access times than the standard Neutron, which lands in the middle of the pack. The GTX’s access times are only slightly higher than those of the lead group of three, all of which are based on SandForce controllers.

TR FileBench — Real-world copy speeds

Concocted by resident developer Bruno “morphine” Ferreira, FileBench runs through a series of file copy operations using Windows 7’s xcopy command. Using xcopy produces nearly identical copy speeds to dragging and dropping files using the Windows GUI, so our results should be representative of typical real-world performance. We tested using the following five file sets—note the differences in average file sizes and their compressibility. We evaluated the compressibility of each file set by comparing its size before and after being run through 7-Zip’s “ultra” compression scheme.

Number of files Average file size Total size Compressibility
Movie 6 701MB 4.1GB 0.5%
RAW 101 23.6MB 2.32GB 3.2%
MP3 549 6.48MB 3.47GB 0.5%
TR 26,767 64.6KB 1.7GB 53%
Mozilla 22,696 39.4KB 923MB 91%

The names of most of the file sets are self-explanatory. The Mozilla set is made up of all the files necessary to compile the browser, while the TR set includes years worth of the images, HTML files, and spreadsheets behind my reviews. Those two sets contain much larger numbers of smaller files than the other three. They’re also the most amenable to compression.

To get a sense of how aggressively each SSD reclaims flash pages tagged by the TRIM command, we run FileBench with the solid-state drives in two states. We first test the SSDs in a fresh state after a secure erase. They’re then subjected to a 30-minute IOMeter workload, generating a tortured used state ahead of another batch of copy tests. We haven’t found a substantial difference in the performance of mechanical drives between these two states.

Fresh from a secure erase, the Neutrons fare well in FileBench. The GTX copies the larger files in the movie, RAW, and MP3 sets faster than anything else we’ve tested. The regular Neutron isn’t too far off in those tests, although it falls behind the Vertex 4 and Samsung 830 Series when copying movies and MP3s.

Both Neutrons fall to the middle of the pack when we switch to the smaller files of the TR and Mozilla sets. The drives are more evenly matched in those tests, but they can’t keep up with the Force GT and the Intel 520 Series.

The relative performance picture for the Neutrons doesn’t change dramatically when the drives are put into a used state. The GTX still takes top honors with the larger file sets, and it has a comfortable lead over the standard model. The Vertex 4 isn’t as competitive this time around, letting the vanilla Neutron slip into third place in the first three tests.

In the TR and Mozilla tests, the Neutrons again slip to the middle of the pack behind a collection of SandForce-based SSDs. The Marvell-based Octane and Intel 510 Series both prove faster at copying smaller files, as well.

Scrolling between the fresh- and used-state results reveals that the Neutrons offer fairly consistent copy speeds regardless of the condition of the drives. The Neutrons support TRIM, naturally, and they seem to be fairly aggressive at reclaiming previously used flash pages.

TR DriveBench 1.0 — Disk-intensive multitasking

TR DriveBench allows us to record the individual IO requests associated with a Windows session and then play those results back as fast as possible on different drives. We’ve used this app to create a set of multitasking workloads that combine common desktop tasks with disk-intensive background operations like compiling code, copying files, downloading via BitTorrent, transcoding video, and scanning for viruses. The individual workloads are explained in more detail here.

Below, you’ll find an overall average followed by scores for each of our individual workloads. The overall score is an average of the mean performance score for each multitasking workload.

Not bad. The Neutron and Neutron GTX lock up fourth and third place overall, respectively. They’re wedged between Corsair’s own Force Series 3, which pairs SandForce’s latest controller with asynchronous NAND, and the company’s Force Series GT, which combines the same controller with synchronous flash. Let’s see if the individual test results reveal anything interesting.

Indeed, they do. The Neutrons fare the best in the file copy test, taking the top two spots. They can’t keep up with the pack-leading SandForce drives in the other tests, though.

The GTX is faster than the standard model in all but the virus scanning test, and the gaps aren’t trivial. However, the difference in performance between the two Neutrons is narrower than the delta between Corsair’s Force Series SSDs.

TR DriveBench 2.0 — More disk-intensive multitasking

As much as we like DriveBench 1.0’s individual workloads, the traces cover only slices of disk activity. Because we fire the recorded I/Os at the disks as fast as possible, solid-state drives also have no downtime during which to engage background garbage collection or other optimization algorithms. DriveBench 2.0 addresses both of those issues with a much larger trace that spans two weeks of typical desktop activity peppered with multitasking loads similar to those in DriveBench 1.0. We’ve also adjusted our testing methods to give solid-state drives enough idle time to tidy up after themselves. More details on DriveBench 2.0 are available on this page of our last major SSD round-up.

Instead of looking at a raw IOps rate, we’re going to switch gears and explore service times—the amount of time it takes drives to complete an I/O request. We’ll start with an overall mean service time before slicing and dicing the results.

Just 0.02 milliseconds separate the mean service times of the top five SSDs. The Neutrons bookend that pack, with the GTX in the lead and the base model bringing up the rear. Even in that position, the plain-Jane Neutron has quicker service times than more than half of its peers. Let’s starts slicing and dicing the results to see what we can learn.

Despite having the lowest mean service time overall, the Neutron GTX manages no better than third place when we split our DriveBench access times along read and write lines. The synchronous SandForce drives from Corsair and Intel come out on top with reads, while the Vertex 4 and Samsung 830 Series are quicker with writes. Consistency appears to be the GTX’s greatest asset here.

Once again, the Neutron and Neutron GTX remain closely matched when we focus on a read-oriented metric. Only when we examine write performance does the GTX pull ahead.

There are millions of I/O requests in this trace, so we can’t easily graph service times to look at the variance. However, our analysis tools do report the standard deviation, which can give us a sense of how much service times vary from the mean.

The Neutrons have among the lowest standard deviations of any of the drives, indicating more consistent service times than most of the competition. Combine that consistency with a low mean, and you have a pretty attractive performance proposition.

Another way to characterize the consistency of service times is to sort them. We’re going to close out our DriveBench analysis with one more set of graphs, this time focusing on the distribution of service times. The results have been split between I/O requests that completed in 0-1 milliseconds, 1-100 ms, and those that took longer than 100 ms.

The graphing order is based on the percentage of service times under one millisecond, which makes up the majority of the distribution for all the drives. The Neutrons are right up among the leaders based on those results. More importantly, they suffer barely any extremely long service times over 100 ms.

IOMeter

Our IOMeter workloads feature a ramping number of concurrent I/O requests. Most desktop systems will only have a few requests in flight at any given time (87% of DriveBench 2.0 requests have a queue depth of four or less). We’ve extended our scaling up to 32 concurrent requests to reach the depth of the Native Command Queuing pipeline associated with the Serial ATA specification. Ramping up the number of requests also gives us a sense of how the drives might perform in more demanding enterprise environments.

We run our IOMeter tests using the fully randomized data pattern, which presents a particular challenge for SandForce’s write compression scheme. We’d rather measure SSD performance in this worst-case scenario than using easily compressible data.

Did we mention that LAMD has been making SSD controllers for enterprise customers? Our IOMeter results certainly hint at that server pedigree, because the Neutrons completely dominate the competition. The file server, workstation, and database tests mix read and write operations, and that’s where the Neutrons are at their finest. They offer higher transaction rates than any of the other SSDs regardless of the load.

The beatdown isn’t as overwhelming in the web server test, which is made up exclusively of read operations. With that workload, the Neutrons trade blows with the Samsung 830 Series, racing out to an early lead before trailing as the load exceeds eight concurrent I/O requests.

As one might expect given the other performance results we’ve seen already, the standard Neutron shadows the GTX in the read-dominated web server test. The gaps between the Neutrons are wider in the other tests, but never by all that much.

Boot duration

Before timing a couple of real-world applications, we first have to load the OS. We can measure how long that takes by checking the Windows 7 boot duration using the operating system’s performance-monitoring tools. This is actually the first test in which we’re booting Windows 7 off each drive; up until this point, our testing has been hosted by an OS housed on a separate system drive.

The Crucial m4 may be the quickest-booting SSD of the bunch, but it’s barely ahead of Corsair’s latest. Interestingly, the Neutron loaded our Windows install slightly faster than the GTX. Really, though, all the SSDs are very fast here. Barely more than a second separates the fastest one from the slowest. Our mechanical hard drive, on the other hand, takes almost twice as long to load the OS.

Level load times

Modern games lack built-in timing tests to measure level loads, so we busted out a stopwatch with a couple of reasonably recent titles.

With only one exception, all of the SSDs are within a second of each other when loading our game levels. Given the hand-timed nature of these tests, I’m hesitant to draw too many conclusions based on such small differences. That said, it’s clear the solid-state drives load game levels noticeably faster than the average mechanical hard drive. After moving to an SSD in my primary gaming rig, I’d never go back.

Power consumption

We tested power consumption under load with IOMeter’s workstation access pattern chewing through 32 concurrent I/O requests. Idle power consumption was probed one minute after processing Windows 7’s idle tasks on an empty desktop.

LAMD claims the LM87800 is a low-power design, but the Neutrons certainly aren’t leaders in that department. The drives consume just over a watt at idle, which isn’t exceptional, and they’re among the most power-hungry SSDs when crunching our IOMeter load.

Keep in mind that the Neutrons deliver substantially better IOMeter performance than the other drives, though. If you look at IOps per watt, the new Corsair SSDs are much more efficient than the power consumption numbers suggest. The Neutron cranks out 3300 IOps/W, which is only just shy of the Crucial m4’s 3400 IOps/W. Meanwhile, the Neutron GTX delivers a more impressive 3800 IOps/W.

The value perspective

Welcome to another one of our famous value analyses, which adds capacity and pricing to the performance data we’ve explored over the preceding pages. We used Newegg prices to even the playing field, and we didn’t take mail-in rebates into account when performing our calculations. Although the Intel 510 Series has been discontinued and is no longer for sale, we’ve included it to illustrate how much the last generation of SSDs used to cost.

First, we’ll look at the all-important cost per gigabyte, which we’ve obtained using the amount of storage capacity accessible to users in Windows.

The Neutron GTX comes within a few cents of slipping below the dollar-per-gigabyte threshold. Meanwhile, its standard sibling costs 21 cents less per gig, putting that drive very close to the most affordable SSDs we’ve tested. As a relatively new entrant into the market, there hasn’t been much time for the Neutrons’ prices to fall from their initial levels. We’ve watched other drives tumble by double-digit percentages over just a few months, so it wouldn’t surprise us to see Corsair cut prices on the Neutrons.

Our remaining value calculations use a single performance score that we’ve derived by comparing how each drive stacks up against a common baseline provided by the Momentus 5400.4, a 2.5″ notebook drive with a painfully slow 5,400-RPM spindle speed. This index uses a subset of our performance data described on this page of our last SSD round-up. The Intel 510 Series was actually worse than our baseline in one of the tests—100+ ms writes in DriveBench 2.0—so we’ve fudged the numbers a little to prevent that result from messing up the overall picture. (The Intel 510 Series was given the same score as the baseline in that test.)

There’s a clear separation between our leading pack of five drives and their next-closest competitors. The Samsung 830 Series has the highest overall score, and it’s followed closely by the Neutron GTX and the Intel 520 Series. Corsair’s own Force Series GT trails just behind those two, with the regular Neutron nipping at its heels.

Our performance index shows just how little difference there is between the Neutron and the Neutron GTX overall. On this scale, the 21 percentage points separating the two work out to a difference of only 2%.

Now for the real magic. We can plot this overall score on one axis and each drive’s cost per gigabyte on the other to create a scatter plot of performance per dollar per gigabyte. The best place on the plot is the upper-left corner, which combines high performance with a low price.

The Neutrons sit to the right of the Force GT and Samsung 830 Series on our scatter plot. They’re in the same top performance tier, but Corsair’s new hotness costs more per gigabyte. The GTX is particularly pricey, matching the position of the Intel 520 Series almost exactly. In fact, those two drives overlap completely on the plot.

With a full terabyte of storage capacity, the Caviar Black’s low cost per gigabyte obviously puts it way over to the left side of the plot. The mechanical drive’s plodding performance is slow enough that it’s really a different class of product, though. Rather than trying to decide between solid-state and mechanical storage, we recommend using both.

Conclusions

Corsair wouldn’t tell us when its exclusive access to LAMD’s LM87800 controller runs out, so it’s hard to say how long the Neutrons will remain unique. One thing is certain, though: these drives are good enough to keep up with the fastest 2.5″ SSDs around. In some cases, particularly IOMeter’s file server, workstation, and database tests, they offer much better performance than the competition.

More impressive than those IOMeter results is the fact that the Neutrons have well-balanced performance characteristics overall. They may not be the fastest options in every situation, but they tend to stick close to the front of the pack. If the drives have a weakness, it’s that their performance in some sequential transfer rate benchmarks isn’t quite as impressive as their real-world copy speeds or their ability to handle random I/O.

While I can understand Corsair’s desire to offer multiple Neutrons with different kinds of memory, the GTX is notably faster than the standard model in only a small handful of tests. Most of the time, the performance gap between the two is much narrower than the difference in price. With the Neutron 240GB selling for $210 right now, it’s hard to justify dropping another $40 on the GTX. Heck, if you look at the scatter plot on the previous page, you might also find the standard Neutron’s value proposition to be a little bit tenuous given the price.

Corsair does have an ace up its sleeve: five-year warranty coverage. Among the rivals we lined up to face the Neutrons, only the Intel 320 Series, the Intel 520 Series, and the OCZ Vertex 4 come with five-year warranties. All the other drives have three-year coverage.

A longer warranty doesn’t guarantee the integrity of your data, of course, but it does provide some peace of mind to offset the fact that the LAMD controller is a relatively unknown quantity. The last time we saw an SSD maker get first dibs on a new controller—OCZ with SandForce’s SF-2281—things didn’t work out so well. We haven’t seen any reports of BSODs or other serious issues with the Neutrons, but they’ve only been available since last month. Only time will tell whether the LAMD controller and its associated firmware are free of troublesome bugs.

In the meantime, we recommend steering clear of the Neutron GTX, which simply isn’t fast enough to justify its premium price. The standard Neutron is definitely the better deal of the two, provided you’re willing to trust a relatively new controller chip.

Comments closed
    • aim18
    • 7 years ago
    • TO11MTM
    • 7 years ago

    I don’t feel the Duke benchmark is very realistic. If you want to capture a real world scenario you should be measuring the uninstall time.

      • Chrispy_
      • 7 years ago

      I think the Duke benchmark is a best case scenario; Nobody will dare make a game with lower-quality, blurry-ass art assets like that for years to come.

    • eofpi
    • 7 years ago

    There’s a typo in the Neutron GTX flash type. Page 1 says 24nm, while page 2 says 26nm.

    • Scrotos
    • 7 years ago

    First page…

    “Corsair has a 480GB version of the GTX in the works, as well, and the controller supports up to a gig of flash.”

    You mean up to a terrorbyte of flash?

      • Firestarter
      • 7 years ago

      A gigabyte of RAM cache maybe?

    • Yeats
    • 7 years ago

    I’m disappointed at the lack of a Pointer Sisters reference.

    • thedosbox
    • 7 years ago

    I assume you meant “Crucial m4 256GB with 010G firmware” instead of “Corsair m4 256GB with 010G firmware”:

    [url<]https://techreport.com/review/23693/review-corsair-neutron-and-neutron-gtx-solid-state-drives/2[/url<]

      • Dissonance
      • 7 years ago

      Fixed. Good catch, thanks.

    • Shambles
    • 7 years ago

    Time to bring on SATA 4.0 (More like time to dump the crawling SATA spec altogether.)

      • StashTheVampede
      • 7 years ago

      SATA is fine for optical drives and platter based hard drives. SSDs are getting their own type of spec, soon. There are a few different ones out there, but they will stop being tied to SATA in the “near” future.

      • Bauxite
      • 7 years ago

      Sata express, think ssd on a card that is native boot and not a cobbled together mess like [i<]some[/i<] of the current types. On the enterprise front SAS is still advancing, 12Gbps/port shipping soon, and it already lets you do stuff like dual port drives and multipath.

      • Vasilyfav
      • 7 years ago

      What mainstream consumer apps need more than 10,000 I/O/s and 550 MB/s reads and writes?

      I understand that PCI-E is the cheapest and most logical way forward for SSDs, but until SSDs capture at least 30% of all storage share by volume, there’s no reason a typical joe user needs it or will even comprehend the difference between 6 and 12 gbit of throughput.

      And if you really need more than that, there are PCI-E solutions from fusionIO.

      SATA is fine for mainstream consumers for the forseeable future. It’s cheap to make and doesn’t require changing interfaces on all of your storage devices.

        • indeego
        • 7 years ago

        The very one everyone uses. The core OS. Boot times in <3 seconds would be very very nice.

          • Firestarter
          • 7 years ago

          I’m guessing that with fast SSDs, it’s mostly limited by device initialisation.

          edit: I just measured the boot time of my computer, it seems to hang a bit at finding disks and such in the EFI, after that booting into Windows 7 takes about 10 seconds to login. That’s with a 2500K at 4.3ghz and a Samsung 830. Even though it used to be faster, it’s still soundly in the ‘whatever’ territory, especially considering that it’s immediately usable after logging in.

          I guess I could use sleep if I wanted it to be available faster, but then I’d still have to wait a second or two for the screen to wake up.

    • ptsant
    • 7 years ago

    Just for info, Plextor drives also give a 5y guarantee.

    • Forge
    • 7 years ago

    [url<]https://techreport.com/review/23693/review-corsair-neutron-and-neutron-gtx-solid-state-drives/7[/url<] Maybe my brain is getting old and slow, but aren't the labels there inverted? The key at the top seems to say that the solid color is the slowest result, the solid bars (longest) are 100+ms, the speckled/halftone are 1-100ms, and the striped are sub-1ms. All the drives have very long solid bars, meaning most requests took over 100ms to service? Edit: Already asked and addressed. That'll teach me to get called away for over an hour between typing and hitting submit. The legends still look wrong on a force reload on page 7.

    • CampinCarl
    • 7 years ago

    Okay, so, maybe I’m stupid, but aren’t the final graphs from DriveBench 2.0 reversed? The legends suggest that the 512GB Vertex 4 actually had 87.4% of its writes complete in MORE THAN 100ms, and so forth.

      • Dissonance
      • 7 years ago

      The data’s correct, but the legends were backwards. Fixed.

    • sschaem
    • 7 years ago

    isn’t using iops as a benchmark metric absolutely meaningless??
    (Edit: for compiling, transcoding, virus checker, etc..)

    ex: From those graph you have the impression that transcoding with an SSD is 10x faster. bollocks.

    You also have the impression that upgrading from a vertex4 to an intel 520 will double you encoding performance. When I bet you might only get single digit % improvement, if any at all.

    What is the reasoning for doing this ??

    Same goes with compiling firefox… really, do we care that iops is 10 time higher on SSD if the compile time is nearly the ‘same’ ?

      • Elsoze
      • 7 years ago

      No- it is a situational metric. It won’t affect *you* but it will affect others.

      Case in point: Databases

        • sschaem
        • 7 years ago

        So you look at firefox & transcoding iops to buy an SSD for your database needs???

      • Deanjo
      • 7 years ago

      [quote<]You also have the impression that a upgrading from a vertex4 to an intel 520 will double you encoding performance. When I bet you might only get single digit % improvement is any at all.[/quote<] Chances are that there would be absolutely no difference as the bottleneck in transcoding has been the the encoding process and not the I/O throughput of the media.

      • Dissonance
      • 7 years ago

      [i<]isn't using iops as a benchmark metric absolutely meaningless??[/i<] Not if you're interested in seeing how fast a drive can crunch the I/O stream associated with a given workload, which we are. If you follow the link at the end of the DriveBench intro, we provide additional details about the test and why we use it. Here's the snippet that's relevant to your question: [i<]During playback, any idle time recorded in the original session is ignored—IOs are fed to the disk as fast as it can process them. This approach doesn't give us a perfect indicator of real-world behavior, but it does illustrate how each drive might perform if it were attached to an infinitely fast system.[/i<] DriveBench gives us a look at disk throughput sans bottlenecks in different situations. That's meaningful information.

        • sschaem
        • 7 years ago

        Computers are not infinitely fast.
        But most importantly since Transcoding & Compiling use async IO, using iops graphs with those operation is a futile effort to measure the benefit of a drive for those tasks.
        Sure you can extrapolate and get some idea, but measuring iops this way seem useless.

        Because it doesn’t reflect server base workload and its doesn’t give usable metric for desktop/workstation workload. As a consumer on both end, those numbers are just not useful.

        Now if you had both numbers (iops + task completion time with resource usage) I could see this being useful, and get a sense of scaling. (task being io limited or cpu limted)

        So, anyone here can tell me if I should upgrade from a vertex4 to an intel 520 if I do allot of transcoding ?
        (going from 2700 to 4300 iops)
        How much will it help with my compile time (3100 iops vs 5300) ?

        And BTW, your iops in your test have little relation to the iometer results.
        Your iops show the intel 520 crushing the vertex4, yet the vertex4 crush the 520 in all iometers test.

        As a netbook/desktop/laptop/workstation owner, I dont see how any of this help with my SSD purchase. And for server wouldn’t I just look at the iometer results?

        Am I the only one that would be interested in Task time , not just iops for transcoding and compiling?

          • jwilliams
          • 7 years ago

          I think what you are looking for is something like PCMark 7, which includes idle times in its storage benchmarks. The result is that all the most recent SSDs come within a few percent of each other in scores. It is worth having a benchmark that includes idle times to compare with the other benchmarks, since it gives a much better idea of how little difference is noticeable between SSDs when running typical consumer workloads.

      • Chrispy_
      • 7 years ago

      I tend to use IOPS as a rough guide to 4K performance, which is the magic thing that makes Windows feel snappier in everyday workstations.

      They don’t always correlate but low IOPS usually means poor 4K read/write rates. Since everyone quotes IOPS in their sales blurb it’s useful to see if a drive can actually produce IOPS as claimed.

      • rrr
      • 7 years ago

      IOPS are clear indicator to why HDDs feel so horribly slow vs SSDs.

      Difference between specific SSDs (excluding old JMicron stuff and alike) is never worth it, pick cheapest reliable one.

      • Bensam123
      • 7 years ago

      Server benchmark. It’s not something a end user will be looking for.

    • DPete27
    • 7 years ago

    Seems like many manufacturers are steering away from Sandforce with their next-gen SSDs. (I don’t blame them) Perhaps we’ll see better reliability now that manufacturers have more access inside the controller rather than having to work with Sandforce’s black box.

      • Jason181
      • 7 years ago

      I think they’re trying to get away from the Jeckyl and Hyde with compressible/uncompressible data exhibited by the Sandforce controllers. They are still fast enough it doesn’t really cause an issue on a home machine imho (I have a Vertex 3), but then most compressed data is going to be large by nature or nobody would’ve bothered with compression, and the size of affordable ssds makes it unlikely that you’re going to be transferring large amounts of already compressed data to the ssd on a regular basis.

      • MadManOriginal
      • 7 years ago

      There’s also the fact that Sandforce hasn’t had a new controller in ages, so there’s nothing new to make.

        • anotherengineer
        • 7 years ago

        Well their first controller maxed out sata2 spec and the current SF controller pretty much maxes out sata3 spec.

        We will have to wait for the sata4 spec for sandforce to make a new controller to max out that.

          • Chrispy_
          • 7 years ago

          Not on compressed data, far from it in fact.

      • Chrispy_
      • 7 years ago

      Yep, I bought a bunch of Sandforce drives based on recommendations and I’ve been happy with them, but doing the odd benchmark I’ve been seeing 200 and 300MB/s speeds rather than the 500+ that are common with a better controller and compressed data.

      It’s the 4K reads that really make the biggest difference to real-world performance, but most end-users deal in compressed data – jpgs, cabfiles, game textures etc and I guess there is a small but tangible difference caused by the lower Sandforce compressed data speeds.

      • HisDivineOrder
      • 7 years ago

      Sandforce is owned by LSI now. Perhaps they don’t like the new owners or perhaps they want to have fast drives that don’t have a problem with compressible data. Or perhaps they are starting to buy their own controllers to use instead.

      Or perhaps they’re tired of all the firmware updates. Hard to say, really.

Pin It on Pinterest

Share This