Tom's Hardware's SSD Hierarchy Chart
We understand that SSD prices make it difficult to adopt the latest technology, which is why many enthusiasts are hesitant to blow several hundred dollars on solid-state storage (especially when they can get a quartet of 2 TB hard drives or a high-performance processor for the same price). That's why it's important to put things into perspective.
Over the past five years, CPU performance has hit new and unforeseen heights, and processors are increasingly spending time waiting on data from hard drives. This is what makes storage today's most glaring bottleneck. Overcoming it requires an SSD.
At the end of the day, the real-world differences between SSDs in a desktop environment aren't altogether very large. The most important jump happens when you go from a hard drive to (almost) any SSD. With that said, there are measurable attributes that separate one SSD from another. However, they have to be digested as a sum of many parts. Within individual apps, you'll hardly notice the difference between a Vertex 2 and Samsung's 830. But if you look at performance over an entire month, you will find the 830 to be a better performer.
The hierarchy chart below relies on information provided by our Storage Bench v1.0, as it ranks performance in a way that reflects average daily use for a consumer workload. This applies to gamers and home office users. The chart has been structured so that each tier represents a 10% difference in performance. Some rankings are educated guesses based on information from testing a model at a different capacity or a drive of similar architecture. As such, it is possible that an SSD may shift one tier once we actually get a chance to test it. Furthermore, SSDs within a tier are listed alphabetically.
There are several drives that we're going to intentionally leave out of our hierarchy list. Enterprise-oriented SLC- and 512 GB MLC-based SSDs are ignored due to the extreme price they command (and the difficult we have getting samples in from vendors). Furthermore, SSDs with a capacity lower than 60 GB are left off because of the budget nature of that price range.
In order to simplify the landscape, we're going to omit brand names for those vendors leveraging SandForce. There are simply too many to list. At a given capacity, performance breaks down based on memory type, and this is their order of performance, from highest to lowest.
We're making an exception with Intel's SSD 330 because it's special case of a SandForce-based SSD that runs with reduced performance specs. The 60 GB SSD 520 is also being called out on its own because it offers performance somewhat higher than the norm.
- SandForce controller with Toggle DDR NAND (Mushkin Chronos Deluxe, Patriot Wildfire, OCZ Vertex 3 Max IOPS, OWC Mercury Extreme Pro 6G, Corsair Force GS)
- SandForce controller with Synchronous ONFi NAND (OCZ Vertex 3, Corsair Force GT, Kingston HyperX/HyperX 3K, Intel SSD 520)
- SandForce controller with Asynchronous ONFi NAND (OCZ Agility 3, Corsair Force 3, Mushkin Chronos, Patriot Pyro, OWC Mercury Electra 6G)
| SSD Performance Hierarchy Chart | |
|---|---|
| Tier 1 | 240 GB second-gen SandForce SSDs with Toggle NAND Samsung 840 250 GB Samsung 830 SSD 256 GB Plextor M3 Pro 128 GB/256 GB OCZ Vertex 4 512/256 GB OCZ Vector 512/256 GB Samsung 840 Pro 128 GB/256 GB |
| Tier 2 | 240 GB second-gen SandForce SSDs with Sync ONFi NAND Corsair Neutron GTX 256 GB Plextor M3 128 GB/256 GB |
| Tier 3 | Crucial m4 256 GB Corsair Neutron 256 GB Samsung 840 SSD 120 GB Samsung 830 SSD 128 GB 120 GB second-gen SandForce SSDs with Toggle NAND 240 GB second-gen SandForce SSDs with Async ONFi NAND Corsair Performance Pro 128 GB OCZ Agility 4 256 GB |
| Tier 4 | - |
| Tier 5 | Crucial m4 128 GB Intel SSD 330 180 GB Samsung 830 SSD 64 GB 120 GB second-gen SandForce SSDs with Sync ONFi NAND |
| Tier 6 | Intel SSD 330 120 GB Samsung 840 120 GB Samsung 470 SSD 256 GB |
| Tier 7 | 240 GB first-gen SandForce SSDs Intel SSD 320 300 GB Samsung 470 SSD 128 GB 120 GB & 180 GB second-gen SandForce SSDs with Async ONFi NAND OCZ Agility 4 120 GB |
| Tier 8 | - |
| Tier 9 | Crucial m4 64 GB Intel SSD 320 160 GB Intel SSD 520 60 GB |
| Tier 10 | Intel SSD 320 80 GB Intel SSD 330 60 GB 60 GB second-gen SandForce SSDs (with Sync or Async ONFi NAND) 120 GB first-gen SandForce SSDs OCZ Agility 4 64 GB |
there was as typo with the Vertex 4, there is no 120GB, it's 128GB
Mushkin Enhanced Chronos has an awful lot one-star reviews on Newegg.
Just sayin'...
Seriously? No mention of the Samsung 830 at ALL? The 128GB model is only $90 on Amazon/Newegg!
what about bootable pci-e ssd ??? regadless of the Enterprise designation often give, they are price competative with these and we are talking PERFORMANCE...
Reliability is my primary concern, so my quick and dirty rule for SSDs is "No Sandfarce, and no OCZ."
I use a Crucial m4 mSATA (238GB formatted) as my system drive in my primary rig, and it is certainly fast enough, even though the mSATA slot is "only" 3Gb/s.
My other rig uses a 256GB Samsung 830 on a 6Gb/s port. I can not tell any performance difference based on the drive; the two rigs are too different to compare them directly.
I've never had a problem with my Vertex 4. Read/write are both around the 500 mark and everything loads instantly.
Reliability is my primary concern, so my quick and dirty rule for SSDs is "No Sandfarce, and no OCZ."I use a Crucial m4 mSATA (238GB formatted) as my system drive in my primary rig, and it is certainly fast enough, even though the mSATA slot is "only" 3Gb/s.My other rig uses a 256GB Samsung 830 on a 6Gb/s port. I can not tell any performance difference based on the drive; the two rigs are too different to compare them directly.
OCZ Vertex/Agility 4 is just as reliable as Crucial M4 with current firmware, so a no OCZ rule doesn't seem fair. Going by individual line rather than company is a better way of going about this.
I don't see why OCZ gets recommended at ANY price point...Since when is unreliability a benefit?
I run 2 x 256 GB of Vertex 4 and they've ran flawlessly for about 3 months now. I guess it's easy to regurgetate what you read from outdated articles referencing the original firmware.
yeah how come no mention of Samsung SSD drives?
yeah how come no mention of Samsung SSD drives?
The first page has a bug picture of a Samsung SSD along with a four paragraph *essay* mostly about the drive in the picture. There are several references to Samsung SSDs in this article.
It is true that the OCZ "4" series drives aren't Sandfarce, however I don't care to be an unpaid beta tester for their firmware. OCZ has lost my trust. There's competent competition (e.g. Samsung) with no such issues, so there's no reason to choose OCZ.
It is true that the OCZ "4" series drives aren't Sandfarce, however I don't care to be an unpaid beta tester for their firmware. OCZ has lost my trust. There's competent competition (e.g. Samsung) with no such issues, so there's no reason to choose OCZ.
Generally better pricing and at least compared to the Samsung 830, better performance. There's plenty of reason to go with OCZ. Furthermore, how OCZ did the launch isn't really bad. They could release it as it was at the time and improve it along the way, or simply wait until they've already improved it (which would probably take longer and probably also mean higher launch prices).
At that point, you're not an unpaid beta tester because you got the drive cheaper than you would have if OCZ had to wait months to a year or more before getting it *ready*. Besides, it's not like Vertex 4 wasn't a good drive for the money even when it launched and the support that OCZ has offered to Vertex 4 owners is superb, to say the least.
Furthermore, Samsung has had firmware issues. No SSD company has been truly free of them.
In actual use, the performance difference between a couple of Marvell drives, or even a Marvell and a Sandfarce, will be virtually invisible. I'll happily pay a few dollars more for a drive I can plug in and use, vs. one I have to plug in and update first. Updates should be optional (even if they provide a visible improvement), not required to get the drive to work properly at all. OCZ's latest drives may be fine, but like I said OCZ has lost my trust. I simply find no compelling reason (a few dollars isn't it) to take the risk. As far as whether or not that's "fair," I'm sure they weighed the risks of releasing unfinished / untested products, and decided it was a risk they could take; at least in my case it looks like they guessed wrong.
I would love it if you guys would include a chart, like your transfer speed chart, that also factors Price and Capacity in the ranking.
I took the first price from Amazon to build this and weighted the Capacity and Speed equally, normalized to Price (enhanced GB/$)... Not sure that's the most realistic, but you get the idea, and it helps identify good value/capacity/speed compromise points.
Thanks,
Ben
Sorry, the formatting on this sucks
M3Pro 128 214 115 3.0
830 64 142 70 2.9
V3 120 149 100 2.7
Agility3 120 114 90 2.6
M3 128 182 120 2.6
V3IOPS 120 179 120 2.5
830 256 201 185 2.5
SSD330 120 127 100 2.5
Agility4 128 104 95 2.4
V4 256 191 195 2.3
V3 240 194 190 2.3
Agility4 256 130 170 2.3
M4 256 168 188 2.3
SSD330 180 141 150 2.1
Agility3 180 122 150 2.0
SSD520 240 183 235 1.8
Perf.Pro 128 171 190 1.6
SSD320 300 121 450 0.9
470 256 124 500 0.8
In actual use, the performance difference between a couple of Marvell drives, or even a Marvell and a Sandfarce, will be virtually invisible. I'll happily pay a few dollars more for a drive I can plug in and use, vs. one I have to plug in and update first. Updates should be optional (even if they provide a visible improvement), not required to get the drive to work properly at all. OCZ's latest drives may be fine, but like I said OCZ has lost my trust. I simply find no compelling reason (a few dollars isn't it) to take the risk. As far as whether or not that's "fair," I'm sure they weighed the risks of releasing unfinished / untested products, and decided it was a risk they could take; at least in my case it looks like they guessed wrong.
There can be significant performance differences between different Marvell-based SSDs, especially where the M4 is concerned because it's simply not that fast compared to some of these other drives and of the current Marvell drives, the Agility 4 is probably the only line that is generally slower than M4. SandForce versus Marvell can have huge performance differences in different scenarios. The firmware updates for Vertex 4 are optional (if you don't want them, then don't get them, the drive isn't not going to work just because you're not on the latest version). There isn't a risk.
OCZ lost your trust because they released products when they were "in beta" rather than "finished? Sorry, but that's ridiculous. OCZ simply gave us the choice of buying them early. There's nothing wrong with that. If you wanted more mature firmware, then whether or not you have the option to buy it before it has firmware that you like doesn't hurt you whatsoever.
It helps you because not only will it take less time to get to that point, not only will you pay less money when the firmware is at that point, but you also get nearly constant news about the progress and thus an understanding of how the drive/firmware is doing at any given time as well as how well customer support and such deal with issues. If anything, I'd trust OCZ more for doing things this way because it better serves intelligent consumers than simply waiting in the dark of what's going on and not knowing what to expect.
Perhaps you'll disagree anyway, but I do not understand your point of view at all if you still disagree.
I use: OCZ Vertex 4 as my system/boot drive. under 15 seconds from "Loading Windows" (Win7pro) to a responsive desktop and that includes typing in my password.
In the preface to your article you write, "we've heard rumors as to why SSDs are so cheap, though, and we're not sure this is sustainable." If I'm not mistaken, you never follow up on that. What's the point of making such a big one-off statement like that if you don't elaborate?
There can be significant performance differences...
OCZ lost your trust because they released products when they were "in beta" rather than "finished?
Performance differences on benchmarks, certainly, and I'm sure on certain types of workloads. In "typical" uses, not really, especially when compared to a mechanical HDD. A double-digit performance difference that amounts to a fraction of a second simply doesn't matter. I do agree that for those certain workloads where it matters, where time is money, a professional will need to focus more on those benchmarks.
OCZ has lost my trust for a pattern of decisions, from PSUs that review well on day 1 but fail early due to substandard caps, to the way they handled the change to 25nm, and more recently the fact that a lot of their [Sandforce] drives had major firmware issues that were bricking drives. What else do they have that might fail, and how? When I can buy a Seasonic, or a Crucial, or a Mushkin, or a Samsung, or ... and be certain it is going to work (given that anyone can have an occasional DOA), I just don't see a need to risk OCZ.
Please don't misunderstand, I'm not lumping them with Crappermaster and others guilty of willful dishonesty; OCZ might earn my trust again someday, it's just that they've been relying on their premium name but have stopped delivering premium products.
I don't see why OCZ gets recommended at ANY price point...Since when is unreliability a benefit?
This is the one thing that gets me. Every site talks about OCZ because of their read/write numbers. I wouldn't buy any of their products. I would pay the premium for intel or crucial reliability. (which i do/have)