SSD Thread (Benchmarks, Brands, News, and Advice)

This is what they say on their website: "This warranty is valid only for the original purchaser of products purchased from an authorized OCZ dealer. Except where prohibited by law the warranty is non-transferable. An original sales receipt or valid copy the receipt is required to establish purchase date and original purchaser."

In real life it may work very different. I know in the Netherlands they'll give warranty to anyone, even without receipt.

 

Hmm. I'll look up Crucial's terms of use, then. I know I managed to get my CT255 replaced even though I got it from eBay. If Crucial has the same official stance as OCZ, there might be some hope.

And, in any case, I could always help the new owner with it. Going to have to make it a NBR sale for sure, though. Thanks, Phil.

 

Same as when You don't leave any unpartitioned space (which will be used sooner or later) and leave some free space in Your partition.

SSDs erase files same as HDDs (only if TRIM is not working), they just delete addresses to "file".

And those cells which is in unpratitioned space will be filled after wear-leveling kicks in.

It's "read-move-erase-write" only if TRIM and GC isn't working.
If TRIM is working "read-move-erase" will happen instantly after deleting file, but if GC is working then "read-move-erase" will happen when drive is idling.

Not 100% sure does it need spare area, but this isn't wear-leveling algorithm.

I agree!

All available space on drive is used for wear leveling (no mater if it's unpartitioned or just unused).

It IS done there and it won't wear out faster because wear-leveling takes care that all cells are equally worn.

Wear leveling DOES work on partitioned space, because it moves around user data that isn't deleted frequently so that all cells can be worn equally.

Lets say Your partition is from cell 0 to 79 (lets say it's 80 GB) and You have cells from 80 to 89 (10 GB) in spare area. For OS partition always is form cell 0 to 79, but for controller on SSD (and physically on SSD) today it can be from 0 to 69 plus form 80 to 89, tomorrow form 10 to 89...

So today Your movie is on cell 2 and all other cells gets used ones except cell 60 which is used three times, then wear-leveling kicks in and move your movie to cell 60 so it doesn't wear out to fast.

Just say if You still doesn't know how wear-leveling algorithm work and I'll do my best to explain it to You.

BTW: Your preconception that I don't know how SSD deletes files is wrong.

 

This is what wear-leveling algorithm does.

 

Ok, I narrowed it down. I'm getting an OCZ Vertex 2 or Agility 2 120gb to tide me over til next year. They both cost the same. I'm thinking the Vertex 2 is better or is it agility?

 

Vertex 2 is theoretically better. If they costs the same you should always get Vertex 2 over Agility 2.

 

That's what I thought. Thanks, Phil. You've been a great help.

 

@tilleroftheearth:
Thanks for that clarification. But I'm not entirely convinced that partitioned space is used for wear-leveling. Why would they say that only "never-written-to-cells" (unallocated space) can be used for this? That's what was stated in the Intel video. So, you're saying that additional space within partitions that hasn't been written to is reserved for this use? If that were true, then why would they have an inaccessible hidden area at all? It makes no sense (to me). They could just program the controller to reserve space in any partition for that purpose. Perhaps the hidden area is just a safeguard... to ensure decent longevity.

In any event, I do notice a slight slowdown in the speed of the drive after using it for a week or so. This speed decrease is not reflected in benchmarks... maybe it's just my imagination.

The online retailer I bought my Intel 80GB from dropped the price by $20 just 5 days after I received mine from them. These G2 drives might drop a lot more in price after the G3 drives come out.

 

@ raydabruce: Are You talking about third presentation from here (IDF 2009 presentation: Enterprise Data Integrity and Increasing the Endurance of Your Solid-State Drive)?

I'm asking because in that presentation the isn't any single word about wear-leveling.

 

That video is (mostly) about endurance (wear-leveling). What else would they be talking about? What do you think they are talking about?

 

They are talking about endurance if You increase unused space, this isn't wear-leveling.

What do You think wear-leveling algorithm is?

 

Yes, that's exactly what it (wear-leveling) is.

 

I replaced my Crucial C300 64GB SSD with a Kingston V+ 64GB.

Batterylife has gone up by 30 minutes. Battery Bar reports 6:33 now.

I'm quite happy with it, I don't notice any performance difference.

 

What's the price difference between those two?

 

I really don't care will You or will You not, but if You want You can read this and this.

It really bothers me in some way that guys like You and raydabruce are here saying nonsense.

 

On Newegg the V+ is about $15 more, for me the V+ was cheaper because I got it second hand.

 

Ok, I've been following this for a while. I've watched the 3rd video, and I've just read the two articles you linked. I'm concerned that you seem more interested in arguing "wear levelling" and being right, than actually being factual and logical.

Both of the articles that you presented are short and lack any more detail than what was presented in the Intel presentation. Neither do they conflict. The only difference is that the Intel presenation never mentiones "wear levelling", yet the entire premise of the presentation is wear levelling.

If you have that much a better grasp over the concept than we do, please cite much more specific sources.

 

Thank You!

I don't have any more specific source about wear-leveling algorithm, but I'm in this tread from the beginning, used to own 1st gen Samsung SSD (Samsung 470 series SSD is 4th generation), reading allot and I know what I'm talking about.

Maybe I should just step aside, read and judge by my own which information is relevant and which isn't. To me it looks like all "old guys" are doing that.

 

In the thread that I spend most of my time in I've had to learn to do just that. Step aside and not argue every point. Take what I like from it and be happy. Only correct the grossest of misinformations. Otherwise, I'll never be happy here.

 

He he he!

I'm not sure what makes me more sad, sitting and not reacting or react and see no effect.

 

It's sometimes best not to react and just let the facts speak for themselves, no matter how hard that is to do. I for one, certainly appreciate all your contributions, especially your alignment posts. I saved myself a lot of hassles following your advice...

 

That's true, but if I didn't react some facts (for example: in third presentation from IDF 2009 about SSDs wasn't any single word about wear-leveling; this one; that some people doesn't know what wear-leveling algorithm is... ) won't come up and lots of people would think that:

Did You try it with XP, Vista or Win 7?

 

That is true, better to make it clear so others don't get mixed up with false info.

I did it using Win7. Worked fine. I've since installed two SSDs the correct way, after reading everything on this thread from Phil, Eye, JJB, yourself, and others, the old guys as you put it...

 

Update for SSD owners who use LINUX:
The upcoming (Oct. 10th, 2010) new release of Ubuntu (Maverick Meerkat) has the new kernel with support built into the mass storage driver for TRIM.

I'm using the beta right now -- works great so far, lots of updated drivers-- i didn't have to manually install any drivers at all.

Current kernel is 2.6.35-22 (comes with the beta).

SSD:
Just saw this in the Fry's ad in the paper:
Corsair Force CSSD-F120GB2-BRKT 2.5" 120GB SATA II MLC

Only $229 after rebate. That's a really fast SSD! I wonder how it holds up over time.

 

Hey all,

given a choice between Corsair Nova V128 and OCZ Vertex 120GB at the exact same price, which one should I pick and why? I'm a mainstream user.

thanks

 

Nova V128 over Vertex 1. See techreport.com for explanation.

 

dang, that's a pretty nice price. i'm about to order my Corsair Force 160GB SSD and the cheapest place i could find was amazon, which had it for $435 + free shipping. do u guys know of a cheaper place to get the 160GB one from? from what i can see, not very many ppl have the corsair force. is there any particular reason u guys are going with other brands/models? i have no experience in this area, so i don't want to get something that won't be worth my time & money.

p.s. - all of these posts about SSD wear levels has got me really paranoid about them. i have never owned a SSD or ever really had any experience with one, so i don't know what to do. if i'm spending $435 on 160GB of space, i really want all 160 of those GB's to be reliable.

p.s.s. - $229 for the 120GB version of the corsair force is a really nice price. my sager NP8850 will be able to house 2 SSD's in raid. would u guys recommend me getting 2 120GB corsair forces (roughly $460 for both) and running them in raid 0, or just stick with 1 160GB corsair force? would 2 120GB SSD's, running in RAID 0, have worst battery life than my OEM 7.2k HDD that comes with the laptop? what are the pros and cons of a RAID-equipped system? i've never owned a RAID-equipped setup, so i was wondering if RAID effects my performance or any of that stuff.

 

@ houstoned, OCZ seems to sell more SSDs than Corsair and the level of expertise & support on their forums is higher.

You can get a 180GB Vertex 2 for $424
OCZ Technology OCZSSD2-2VTXE180G 180GB Vertex 2 SATAII 2.5 SSD

A 240GB Agility 2 for $480
OCZ Agility 2 Series 240 GB Internal hard drive - 300 MBps

I do not recommend RAID. TRIM isn't working yet.

_________________________________________________________

The latest Intel RST Beta driver (*43) boots 1 second faster than the previous stable version on my Acer 1830T.
Be careful though as this is a Beta driver there is the possibility for problems.

 

thanks for clearing the RAID stuff up for me. does that make the OCZ "better"? the main reason why i decided to go with the corsair force was because u guys recommended it. i just want the most reliable, worth-while, one out of the bunch. performance all seems to be relatively the same, so i guess it comes down to reliability and customer service/support.

the vertex 2 is faster than the agility 2 correct? $424 for 180GB is a much better deal also compared to $435 for corsair's 160GB model.

 

Without having hard data I'd put my money on Vertex 2 being the most reliable. Tony from OCZ has said that the Vertex uses the same NANDs as Intel G2. In combination with the write protection of Sandforce that should be very reliable.

I don't know what memory Corsair uses though.

According to the benchmarks of Storagereview the Corsair Force is slightly faster because of firmware but you won't notice these differences in real life.

 

@ houstoned: if You don't know what wear-leveling algorithm is and how it works just don't bother, won't make any difference if You know. Just enjoy Your SSD when You get it and I hope it would be sooner then later.

BTW: between Corsair and OCZ I personally would go with OCZ

 

that's exactly what i want to do. i want to purchase something reliable and performs well so i can just enjoy my laptop. i know that the SSD's performance will degrade with time. that's why, if i'm spending my money, i'd rather get something that won't turn to crap in a few months. u know what i mean?

thanks for all the help guys. i'm probably going to order the OCZ vertex 2 (better than the agility 2 correct?) later today or tomorrow.

 

Both Techreport and Anand have said that you won't notice the performance difference between Agility 2 and Vertex 2 in real life. I agree with them.

If the Agility 2 is significantly cheaper I would get it.

By the way unless you write unreasonable amounts to it it will not become slower over time.

 

Hasn't it been shown though that the performance over time isn't degraded significantly? I mean maybe 10%?

 

Unreasonable by whose measure?

The manufacturer's or our own specific usage pattern?

In just over a month ('work time') I have managed to write almost 3TB of data to my Inferno (100GB) SSD.

If I continued to use it like this (I have stopped using it as I intended because of this issue for over 2 weeks now), I would have under 15 months of use out of it - all the time with performance degrading further and further (31 days divided by 7%).

Is this unreasonable?

I would expect much more out of current technology.

 

According to Intel the average user writes about 5GB per day.

Demystifying SSD Endurance | StorageReview.com

Tiller you write 100GB per day, that's what I call out of the ordinary (unreasonable might be the wrong choice of word).

But didn't your benchmark results in the SSD tweak threads look fine? How much is the degradation?

Who says so?

But yeah if you write 100GB per day it would be advisable to take a Seagate Momentus XT.

 

Actually, being a SandForce drive, it is more than 100GB per day (when the data is compressable - it only writes the smaller amount).

The benchmarks may look fine - but in actual use the SSD performs worse than an XT Hybrid.

Also, don't forget that I have stopped using it at that level for almost a third of its life.

The numbers tell me that if 7% degradation happened in around a month (31 days, for sake of argument) then 100% degradation will occur at just under 15 months - if I had continued to use it at the same level.

When I was using it 'normally' for me - the posts I made were heavily moderated/deleted.

Suffice it to say that the SSD was put to use in a simple 'digital notebook' role where email, websurfing and (to me) very, very light PS'ing is all that is demanded of it (for now).


I also have to comment on the fact that 100GB per day seems 'out of the ordinary' for a high performance storage subsystem. Why else would I want to use the fastest available storage options if I wasn't going to use it fully?

 

If you don't have data to support the degradation how can you say it's 7%?

Your comparison between XT and SSD is between different laptops. That's not an accurate way of comparing nor benchmarking.

You were assuming your SSD was degrading before, then it turned out later to be caused by the PM55 chipset bug.

So please come up with real data or real comparisons.

Your posts have been deleted because you were spreading wrong information without backing it up with adequate proof.

If you really have performance degradation, which is not unthinkable after writing 3TB, it should be extremely easy to prove. Just run the AS-SSD File Copy benchmark.

 

Okay, if I have the opportunity, I'll try again.

Given:
My installs are identical between all my systems. Including IRST drivers.
Systems have 8GB RAM installed.
The systems I compared both run at 2.26GHz one an i3 350M the other a P8400.

If anything; the i3, SSD based system should be faster, right? With identical O/S and apps installed, and the identical data used for the 'tests', it should be faster.

If anything it should be the same speed/time. It was not and well outside the margin of error too (3 to 4 times longer than the P8400 based XT Hybrid I was comparing it to).


Addressing the 7% degradation - it is what SSDlife and CDI reports - the SSD with just a month (740 hrs) of usage has 93% life remaining. Is this not adequate proof?

 

@ tilleroftheearth: Do You mind telling me what do You do with Your notebook so that You need to write so much to SSD?

 

No. Different chipsets, different laptops. Any comparison based on that is not accurate.

How much of a negative impact the PM55 chipset can have we have seen lately. The Photoshop updates benchmark numbers you supplied were caused by the PM55 problem.

It doesn't say much. There are a couple of uncertain assumptions in what you're saying. One is that the software can predict it accurately, two is that it will degrade linear.

I do believe that any SSD could degrade if you write 3TB to it in a month. But if there is performance degradation it would be quantifiable. If it's not quantifiable, there isn't any performance degradation. Quantifying it would be as easy as running AS-SSD File copy benchmark.

 

Download, duplicate, edit, convert (in 4+ RAW image converters) and produce final output of RAW digital images.

 

Intel use 10k PE/C NAND on their non enterprise class drives. OCZ uses 5k PE/C NAND for all of their non enterprise class Sandforce drives. They also use 5k NAND on some of their Indilinx drives but the later firmware revisions 'took increased measures to look after the longevity of 5k NAND drives,' whatever thats supposed to mean.

The Sandforce MLC drives have Duraclass and whilst nobody but Sandforce knows exactly how it works, everyone selling Sandforce drives has been nice enough to give us all 3 to 5 year warranties. So just backup your data like you are supposed to do anyway and use your SSD. You do not need to worry about bricking your drive with write wear.

What exactly are you doing to write 100gb a day to your drive?

Some SMART values are reporting incorrectly. Drive life left is one of them and this applies to any SMART reader like SSDLife and CrystalDiskInfo. While we are here, you can also ignore 'Retired Block Count' since that is also incorrect. The day I got my Vertex 2 and immediately after installing windows 7, CrystalDiskInfo reported 72% drive life left and 96 bad blocks.

Some folks on Indilinx drives also managed to reset their drive life left counters by flashing their firmware from 1.5 to 1.6. Go figure.

 

Well, Phil - I can agree that this was not a scientific comparison - but, at the same time, I did not buy a new system (at a higher overall price) to be slower than what I have been running from almost two years ago and what is now described as not only being 'over the hill', but ready to retire.

See:
Sony VGN-AW110 expert reviews


I also agree that the degradation should be quantifiable. To me, my tests were enough proof. But you seem to think that file copy benchmarks show how the SSD 'felt' during use or would show how/why PS updates took so long.

We can agree to disagree here, no prob!

Just that a 3TB write cycle in a month (actually was more like 10 days to two weeks) should be easily handled for a storage subsystem that is marketed as 'fast'.

Why do I need fast if I can only do it for a few seconds/minutes of its maximum speed?


Another distinction I see is that you seem to relate degradation to 'itself' (the SSD in question). While I relate it to 'my previous working setup'.

We can both be right about these definitions, but the one that is most important to me is if the new tech gets my work done faster - and it clearly doesn't.

At least, the SandForce technology doesn't (for me).

 

Hayte, see the post above yours for my usage.

I totally agree with you about the fickleness of the 'Smart Reporting'. No argument there.

I'm just worried 'what if it's right'?

 

I fixed this for you.

If its right you have a 3+ year warranty. What do you care? You'll get a free replacement drive equal to the one you bought or better if it dies inside 3 years. But its not right, you are covered, and if you worry about things that may or may not affect you then you got bigger problems than your drive dying in 15 months.

I mean you may or may not die tommorow.

 

No the reason I said AS-SSD Filecopy benchmark is because:
- it's real world
- easily repeated
- easily quantified
- quick

It's by no means perfect, I understand that. But performance degradation tends to show in write performance, so that would easily be shown in the FC benchmark.

Installing CS5 would work too, but that's a lot less easy to do, repeat and quantify.

The PS updates benchmarks you posted were before you found out about the PM55 problem.

That's the only reason a platter drive could outperform a SSD.

I was going by what Tony said: Intel uses 5K P/Ec in the G2. Here: http://forum.notebookreview.com/har...-full-advantage-fast-ssds-20.html#post6658771

 

There was a whole article on the reporting of the SMART attribute here somewhere and it clearly explained why SMART values (E7 in particular) wouldn't always report properly because SMART wasn't designed with SSD's in mind.

My brand new OCZ Vertex 2 shows 93% Good (7% wear). That is not accurate.

 

No, this has a 5 year warranty. But, I have never returned a HD for 'warranty' work with mine or any client's data on it. Ever.

So, to me the 5 year warranty may as well be 15 minutes - because that is how long it takes me to put personal data on a new drive.


And I care because I want to depend on my hardware for longer than a depreciation cycle.

 

Well if it showed severe performance degradation you could wipe it and send it for RMA.

I'm just not sure what they'll say if they see there was written a very large amount of data to it.