Intel 520 vs Neutron GTX vs OCZ Vector vs Plextor M5P (256GB version)

which is the best in term of realibility? i don't really need a benchmark speed. i need realibility which i can trust storing my crucial data on it.

any suggestion anyone? Intel 520 and neutron GTX at $250~
Vector and M5P about $270~

all with 5 year warranty. I live in indonesia so Samsung is not an option.

 

Reliability? Well, you're mainly looking at Crucial, Plextor, Samsung (though you don't have this option), and Intel (non-SandForce... Intel used to make their own, better controllers) if you want that. OCZ's customer service is probably the worst I've ever read about here on NBR, or Newegg, or wherever. As for LAMD, I can't find any articles detailing its reliability from a Google search, so I'd take a grain of salt when considering the GTX.

I'd also check out Crucial's M4 or M500 and compare/contrast the prices of that to the Intel and Plextor you're looking at, for completeness.

But as disclaimer, nothing is 100% reliable. Doesn't matter what drive you get, I'd suggest that you keep multiple backups on different drives/media if you don't already, just in case .

 

The Neutron GTX has the smoothest and most consistent random write performance of all the consumer SSDs (due to the use of the Enterprise LAMD controller). This is the ideal drive to use as a scratchdisk or for operations which are latency sensitive. The main disadvantage is the idle power consumption which is significantly higher than the others.

The Plextor M5p is a fairly solid performer all round but it suffers from predictable but significant latency spikes with heavy write operations. The main advantage of this drive is the price, it gives features of other SSDs for slightly less but at $270 I don't recommend.

The Intel 520 is probably one of the most reliable drives out there due to a combination of Intel engineering, premium NAND and long time on market. The usual caveats of Sandforce applies but within those limits, this drive offers very good all round performance with special attention to consistency (better than competitors but not quite on the same level as the Corsair Neutron GTX). Basically, fast, reliable and power efficient given the limitations of Sandforce.

OCZ Vector is probably the most expensive of the lot. However, this is the fastest drive when it comes to pure write operations with some of the lowest latencies and access times. The read performance is competitive with the others too. Additionally, the vector is almost immune to performance degradation due to its 75% system (when you use 75% of the NAND, it triggers a special garbage collection mode for a few minutes then performance recovers). I would say reliability should be good (note quite Intel but definitely trustworthy) simply because OCZ have tested the hell out of it for a whole year before release (similar to what Intel did with the 520). Reliability is not quite on Intel's level because OCZ doesn't have access to the ultra premium NAND flash that intel does but it has control over just about all other aspects of the drive.

Honorable mention: Crucial M4, not the fastest but probably regarded as the most reliable SATAIII SSD simply because it has been on the market for the longest. Pretty much all the problems have been found and ironed out at this stage.

If you want absolutely brute reliability: go with the m4
If you want slightly more reliability than speed: Intel 520
If you want slightly better speed than reliability: OCZ Vector
If you want a scratchdisk for heavy IOPS: Neutron GTX

 

ths.kuroi-tsubasa, i forgot to mention that M4 prices in my country is a bit high on par with Neutron GTX and Intel 520 which turns me down. of course, i always make backups. not everyday, maybe once a week.

Indonesian Distributors of OCZ brand for SSD, is one of the best distributors of computer parts in my country, as they always provide a good solution and speedy RMA.

i've bought crucial M4 128 GB in the past and in a year the speed crawls down.. i tried to contact support but, they didn't provide any firmware to fix these issues. RMA my old drive is not an option as i bought it in different country when sandforce controllers bug haven't been ironed out. no other alternative than sandforce. and Intel was very expensive that i can't buy it..

low 4K-64Thrd with AS SSD - Crucial Community

so OCZ is a safe option? is premium price of OCZ worth for $20 than intel?

 

With the OCZ vector you will be paying for a very sturdy SSD frame, high quality 25nm Intel NAND Flash (which is getting harder and harder to find) and an extremely fast Controller. For the Intel drive, you will be paying the premium for top of the line Intel 25nm NAND flash and the validation procedures to ensure reliability. I think if you are extremely concerned with reliability then go with the Intel 520 whereas, I'm more inclined to go with the OCZ Vector simply because I hate having to work around the limitations of Sandforce, I want top performance without caveats like worrying about what kind of data i'm storing or if my drive will degrade because Sandforce can't process TRIM efficiently.

 

I wouldn't pay for an OC, drive simply because of their bad customer support. Their pervious SSD drives were crap (I'm thinking 1st-gen Sandforce and their service for that, plus stealing customers 32nm NAND and replacing them with 25nm NAND when people RMA'ed their drives). So even if their new drives are any good (and that's a big if), I doubt that all the bad CS reps were retrained and/or removed.

 

Choose whichever you want-Intel or OCZ. But of them use the smart controller SF-2281. I am a big fan of SandForce for years now & have used most of the SSDs containing SandForce. My latest addiction is Intel 520 SSD due to the advantages it not only showcases on papers but is proven as well. Take a look at this tech brief of this product & I am sure many of you will have your perceptions changed - http://www.intel.com/content/dam/www/public/us/en/documents/technology-briefs/ssd-520-tech-brief.pdf

 

It's pretty well-known that Sandforce uses compression tricks to boost apparent speeds. However, if you have read OP's post, you'd know that OP is looking for reliability, and that is something that SandForce historically has a very difficult time providing.

If we're only going to consider Intel drives, the most reliable Intel SSDs on the market are the 320 (uses an Intel-made controller) and (possibly) the DC S3700 (which also uses an Intel-made controller, though is a newer product).

 

Is Intel still selling the 320? I thought they EoL'd it a long time ago and replaced it with the Sandforce-based 330.

 

Intel themselves don't sell anything to end-users like you or I. But resellers such as Newegg, TigerDirect, Amazon, etc. can still sell it if they wish, so long as they have stock. For example, http://www.amazon.com/Intel-Series-...UTF8&qid=1367901256&sr=8-4&keywords=intel+320

 

The OCZ Vector would be my choice.

I've owned 6 OCZ SSDs and never had a problem with any of them.

I did have problems with a few different Intel SSDs that had Intel controllers,a G-Skill and a Crucial C300.

A lot of people that post negativity about OCZ SSDs never owned one.

"anything that can go wrong will go wrong"

 

I wouldn't buy any of those. Get a Samsung 840 pro!

 

It is not that simple as "anything that can go wrong will go wrong"
People went away from OCZ mainly because of media effects, when one people says that the company sucks you'd doubt it, when reviews says the company sucks you'd run away from it.

In OCZ's case it is reviews making it look bad, which is quite a while ago ie. a year
Now I'm not talking about reliability anymore because majority of SSDs last longer than they were first in the market for consumers (imagine all of the 1st gens breakdown due to end of write cycle, thats when we can get the average life data from each SSD brand)

I'm pretty sure OCZ is having some progress on making SSDs more reliable, the only problem to solve is the lost of trust due to past experience on previous products, heres my explaination on why its got so many hates on OCZ: it takes 1 mistake to ruin 1's reputation and a lot of hard work and time to get them back

 

and some people just don't learn to forgive, when they had bad experience with something or someone, they would run away from it rather than giving it another chance

 

For the record, OCZ as a company still sucks. If you need to RMA an OCZ SSD you still have to send your drive in and wait 3-4 weeks, during which time their CSRs will completely ignore your requests for status updates.

 

That in itself is a good reason to skip over OCZ, imo.

 

OCZ doesn't deserve any type of forgiveness. What they deserve is to go bankrupt.

 

its not about forgiveness, its about delivering reliable products. AFAIK vertex 4 is a good and reliable SSD.

One notable fact is that OCZ sold at lower prices SF SSDs, guess how many bought into that SF promise of a good product? For me its a simple measure of SF not being a good product and with the intersection of OCZ selling most of them.

 

Even if OCZ had stayed away from Sandforce and exclusively sold Marvell-based drives with Crucial M4 levels of reliability, we would still hate OCZ just the same.

The 25nm bait and switch scandal and the terrible customer service are entirely of their own making - Sandforce has nothing to do with it.

 

Sounds to me like there are a lot of hate indeed, but do note that OCZ im HK actually replaces your SSD immediately when you visit its distributor directly, and well, for someone like me who accidentally breaks the OS every 3 days with backup solution, brands don't matter anymore but numbers...

 

Exactly. Through my own experiences so do Supertalent!

I'am afraid how can people forgive a company that killed your data numerous times. I certainly can't cause it was the biggest irritation I have experienced with computers.

OCZ went too far with their appalling track record. Getting a vertex 2 back in the day equaled flipping the coin and hoping you got a drive that lasted out the week. Not exactly confidence inspiring!

If you want reliablility. Sammy and Intel are the way to go. Period.

 

Last time I read up on the subject even intel was having problems with their sandforce drive.

I stand by what I said. Get an 840 if you're a general user or an 840 pro if you do lots of writing. Those are the best consumer SSD's money can buy right now.

 

Just my two cents, but my SandForce-based Intel 330 has been running beautifully since buying it last year. Though I was a little worried at first simply because it was SandForce, but the 2281 isn't half-bad with good firmware.

 

..No, sandforce has pretty much all to do with it. It's also worth pointing out, I think, that intel sold many, many ssds with the same controllers (and the same fairly serious faults) that OCZ disks had. All being issues with how the controller interface dealt with memory addressing on the firmware level.

So if you're going into the actual problem, both OCZ and Intel had exactly the same issues. The critical faults and hangs and so on. And.. old Intel and ocz disks still have firmware that creates very well known slow-down issues when overwriting already written sectors.

The difference, I think, is that the first Intel disks probably were sold with oem setups, with support deals provided by various different other retailers. While the OCZ disks were sold to private enthusiasts, with promises of ridiculous speeds and so on. So when these drives failed, caused faults, needed firmware updates to even work, that was a big problem in the first place. And saying, like OCZ essentially did, that private users "just had to" flash their own drives to use them. That was a mistake. Obviously they should have recalled all of them, flashed the new firmware, and then put them out again. Instead of keeping a faulty product on the shelves, so it would fester for as long as anyone had the devices in stock.

But technically speaking, the drives - all the drives that used the first sandforce controllers - had the same issues. Corsair, who are extremely open with their construction and the way they issue firmware.. who also have the best drives now.. they had recalls on top of the sandforce firmware issues from ram failures or something like that. Intel and OCZ never had that, and arguably never rolled out actual unsalvageable products..

Anyway. So if you're looking for something in terms of whether it's:
1. going to work.
2. without hangs.
3. and no unpredictable writing speeds.
4. no crazy issues with manual trim.
5. and no problems with driver-based trim that can erase your entire drive.. theoretically..
6. and which will periodically cause slow-downs, since they're intended only for "periodic" personal computer type use..
7. and that won't cut you down on write speeds, which is often important to make sure heavy jobs don't actually lose read throughput (what happens is the controller often has to wait for certain writes to complete before the next batch can be read, etc. So average reads might suddenly be cut down).

Then what you want is basically any reasonably new drive with 500Mb/s read and write. Practically every controller on the market now works. The only real difference on the drives produced lately is that some use smaller nm production to cut down cost. And the sandforce 2 series controllers are usually among the best, because they run the "trim" type preparations and reallocation routines transparently to the user. That.. was the point with the firmware that failed really badly as well.

And personally, I would not pick an ssd - for any use - that has a manual user-space program that periodically runs trim. This is asking for trouble, frankly. Imagine having a hdd where essentially random pieces of information is periodically erased and moved around by commands issued by a windows program. Microsoft for example specifically forbids you to even use your drive when defragging... you know, that's the level of trust they give a program they write themselves of that type..

 

nipsen,

manually running TRIM is no different from it being run automatically. Windows (or the O/S in use) still issues the TRIM command... The SSD decides what to do with that 'suggestion' that is was offered.

I've never seen MS 'forbid' you to use your system when defragging - and I have not seen any consequences from doing so (even with SSD's...). As a matter of fact, the MS 'approved' way to defrag is still head and shoulder above any other proprietary method (and why I prefer PerfectDisk as they fully adhere 100% to the MS method for moving files on a 'live' file system - after all; who knows the file system better than the O/S manufacturer...).

 

..someone who actually knows how to program a filestream?

No, seriously though, they lock files that are open. They disallow moving files that may be written back from memory. Because the user-space program doesn't know if the file-system is current after the calls have been made, before they are reported as complete. Before that happens, there's a possibility that other routines could take place, and corruption would happen.

So since Windows has no real differentiation between monitor calls and user-space calls, there's really no safe way to do what the trim commands do from the user-space, with "smart" logic, etc. And you end up with the lock problems that also can cause pretty bad corruption. You potentially have this when the OS "requests" TRIM(aka. "Engineers have no imagination when it comes to acronyms") commands as well, because other programs (which also have monitor status), may write back to "sectors" that are not current. This is a problem unique to windows, but it's a problem nevertheless.

If you on the other hand make sure this happens on the controller itself, the controller software can report "sectors" as present, even while doing small move-operations, and always be guaranteed that these operations will complete without causing inconsistencies. And on the concept-level, this removes a potential problem. Where.. basically.. it normally "should" work as long as you don't start the percolator at the same time as plugging in the toaster-iron. Which.. happens.

Like I said, this is a reason why automatic trim has been in focus, because it removes a reliability issue.

(By the way.. apologies for how my replies to you seem a bit.. testy for some reason, tiller lol I don't mean to do that).

 

nipsen,

your info is a little behind the times:

See:
PerfectDisk 12 Help

See:
http://www.raxco.com/user_data/whit..._disk_defragmentation_solutions_5_31_2011.pdf




Furthermore, contrary to what you believe - the hardware (SSD/controller/nand) does not know what is valid data or what is invalid data - only the O/S does.

That is why whether we send a TRIM command manually or automatically the result is the same (Windows tells the storage device which data it CAN purge - if/when it feels like it - it is not 'forced' to do this depending on many other factors).

 

Why are my posts now moderated?

Edit: of course, this one gets through... sigh.


nipsen,


See:
http://www.raxco.com/user_data/whit..._disk_defragmentation_solutions_5_31_2011.pdf


Check out the Safety Considerations section of the above file (almost at the end: pg:5).

Clearly states the opposite of what you are suggesting with regards to defragging...



As for TRIM - a device does not know by itself what is valid/real data and what is invalid/un-needed data - the O/S provides that information to the nand based storage device.

See:
http://www.anandtech.com/show/2738/10


Automatic TRIM has been a focus because it keeps the performance and the health of the nand at it's highest - not because it is more or less safe than manually TRIM'ing an SSD storage device (or, having to have an overly aggressive GC routine in place which only tries to keep the performance high but the power usage and nand's health as secondary).

 

No, it copies the data physically, verifies it, before "moving" anything, and updating the reference. It goes faster than it used to, but the principle is the same as before. And you run into.. performance drops if you run that operation while working. Normally not a problem, but..

Yes, but windows doesn't see actual sectors on the ssd. It pages the controller of the disk, and addresses it essentially like an external ram module. So what you can run into is that one program locks resources, another runs trim, external function completes, reports perfect results. And then the first program writes back data and causes either errors or protection faults. This shouldn't be possible, but it is because there are filewriters in windows that still map out data based on "different", old methods that Windows still have to support..

Scope problem. Or.. something..

 

So between the Neutron GTX and the Intel 520...someone said earlier the Neutron GTX will probably only be very good as a scratch disk not as a main drive for a laptop?