Best SSD for SATA II Notebooks?

TRIM doesn't "kick in"; it's either on, or off. TRIM is simply a message sent to a SSD when a file is deleted which lets the SSD know that the file is deleted. This is required because of the way that file systems and storage media typically interact.

Before TRIM (well actually, more like before SSDs), when you deleted a file, the file system marked it as deleted (within itself), but typically didn't bother to inform the hard drive/storage medium. This is why and how data recovery systems work; because the file is only deleted within the file system, all of the data is still there on your hard drive/storage medium, and all that's required is figuring out again which parts of the data on the hard drive/storage medium make up that file. This data remains on the HDD until the location where the file "used to be" is overwritten with a new file or data. This is why they always say that if you want to truly delete a file, you needed to full-format a drive, which actually would overwrite all the old data.

Now, for an SSD, this doesn't work out as well because of the way SSDs also have to manage wear leveling, which means they have to move around data all the time. Without TRIM, this means that a lot of "old data" that's safe to delete, ends up sticking around for a while and gets rewritten as the data is moved around for wear leveling because the SSD simply doesn't realize that it's data that can be deleted, thus contributing to write amplification, which increases wear on the drive, and also slows down writes (because the drive has to "write more" when it moves things around, due to the nature of page writes and block deletes).

So TRIM lets a drive know when data has been deleted, and thus can be cleared from the storage medium, instead of waiting until the file system says "oh, by the way, put this new data where that old data used to be". So in that respect, TRIM happens "immediately", as soon as the file/data is deleted. However, that simply tells the SSD that the data iin question can be removed; it's up to the drive itself when it bothers to do that. This would be the job of the garbage collection routines of the SSD.

Now from what I understand, Intel will get the TRIM command and "immediately" clear the appropriate data from the NAND, to try to keep as many open pages as possible to be written to. The downside of this approach, naturally, is that it'll involve a lot more writes to the drive, since every time you delete data, it involves taking the information from a block, figuring out what you can keep and what you can delete, and then writing the information you want to keep to new pages, before you completely wipe the block you just cleaned up (this is due to how flash works; you can write in pages, but you have to delete in entire blocks). Thus you can see how if you delete a file, and then delete another file a little bit afterwards, this could result in using up 2 blocks of writes, since you rewrote after the first deletion, and then had to rewrite again after the second (this assumes that the files fit inside a single block, the numbers aren't quite as bad as the number of blocks goes up).

Sandforce, on the other hand, while it doesn't exactly ignore TRIM, it delays using that information until it does a regular garbage collection (GC) cycle, so, in the previous example of 2 deletions, if it doesn't run a GC cycle in between the 2 deletions, it'll run the GC cycle after the deletions, notice that the deleted files are no longer valid, and then clean up and write a single page as opposed to the 2 that would have happened if GC happened "immediately". This has obvious benefits in terms of saving program/erase (P/E) cycles.

Anandtech touches on this in his Vertex 3 Pro Preview. So if my explanation isn't quite clear enough, you can try that link as well, which explains largely the same thing from the Sandforce point of view. As well, there's the SSD Anthology here, here, and here which explain page writes, block deletes, and TRIM (you can read the pages in between, too. They expand a bit more, but aren't strictly necessary in terms of what I'm trying to explain here).

As for the left idle at the login screen thing, that's really only necessary if you have a background processes that won't let the drive be idle long enough for garbage collection to be run. I don't usually bother to leave my Vertex 2 at the login screen, and I've seen no performance degradation. This may be different depending on how you use your computer. If you just login to the computer, and immediately do things that involve constant disk i/o, and then shut it off immediately, then you're never giving garbage collection a chance to run, and thus performance would degrade (unless Sandforce forces GC runs... which by the Anandtech link it does, but will obviously have limited speeds at that point, which is the 80 MB/s clean/recycle path limit). So basically, if you leave your computer idle occasionally, then you should be fine. The login idle is just to "guarantee" GC runs.

 

This is incorrect. Sandforce drives don't require maintenance under Windows 7.

Like I've said a couple of times, the only real issue of Sandforce drives is that sequential writes of incompressible are limilted to 80MB/sec. There is no solution for that. A secure erase only helps temporarily, normally a few hours.

This issue affects mainly large file copies of incompressible data. The rest of normal usage isn't affected.

I've seen examples of performance degradation on X25m in the SSD thread.

If your priority is reliability I'd go with X25m or Samsung 470.
If your priority is speed under multi tasking Sandforce has benefits over X25m.

 

I have my doubt of your guess about Intel controller. The main reason the x25m(since G1) was so successful comparing with their counter parts of the time(JMicron/Indilinux etc.) is that they have an algo that significantly reduced WA(they said close to 1.0, i.e. no wasted write). What you described sounds more like the approach taken before it.

The different between Intel and Sandforce seems to be that Intel would do the erase when a block is 'free'(i.e. if a TRIM makes a 1M block completely free, it would be erased and go back to the writable list). Sandforce however would only erase where there is actual write request or batch it up during idle time.

 

Well, I don't know that we'll ever get confirmation on exactly how it works, since that's probably considered proprietary information. BTW, since the G1 didn't have TRIM, then everything I've said obviously doesn't apply to it. As well, part of what reduces WA for the Intels is their write cache; Sandforce drives don't have a write cache.

Let's see, other differences. Intel has stated that their WA is 1.1, which is excellent. Sandforce of course, will usually average less than 1 due to compression, unless, of course, we're dealing with incompressible files. Other than that, well, I'm fairly certain that Intel uses higher grade NAND than most of the Sandforce vendors do (which contributes to their high price), and given how Intel states they have no problems with guaranteeing equivalent (or higher) life even with newer, lower P/E NAND, I don't see why they wouldn't do what I've stated, if they've been as conservative with their life figures as they seem to have been. After all, I was referring to a worst case situation; in most conditions, it won't be nearly the double WA that I was positing (not least of which because most files aren't less than the 512 KB of a single block). The write cache helps with this too, obviously, since anything stored in the cache and modified there obviously won't be an "actual" write until the cache gets written.

And really, as has been pointed out multiple times here (and is still worth repeating), the differences that your average user will notice performance-wise is close to nil. The closest your average user will get to noticing these issues is probably if they buy a small drive (like 30-60 GB), and filling it just about completely. Then, as they write/modify/delete files on the small drive, they're most likely to notice any speed issues, as we'll be dealing with a massive number of read/delete/write operations as they try to shuffle data.

 

If they do they way you describe, it is highly unlikely to have such a low WA.

 

Who has already tried Intel 320? How is it?

 

I'm sure it would be pretty comparable to the X-25m in everyday use. Higher write speed of course

 

There is one person on the sticky thread that already has its hands on one, in case you interested, link.

 

Actually, looking a little further into it, it could still work (similarly) to the way I described it, because I get the feeling that a TRIM operation isn't considered when they calculate WA. After all, WA is defined by how much needs to be written to the NAND after the host requests a write, and TRIM happens when a file is deleted, not when anything is written. Thus Intel can claim the 1.1 WA because their algorithm is good at not having to "rewrite" already existing data in a block, just writing down new data in a new page/block, with the downside of leaving the invalid material in the old block, just marked off so that it can be gotten rid of later in a GC/wear leveling cycle. Of course, as we've covered before, Intel GC is probably a lot more aggressive than Sandforce, so it'll happen more often than Sandforce does (or maybe it doesn't, which is why the G1/G2 had such low sequential writes?). That way TRIM doesn't figure into WA at all, since it's not a host requested write, it's an internal "cleanup" write. A bit disingenuous, I know, but that's nothing new, and the claimed 1.1 WA seems to be for both the G1 and the G2.

Anyway, this is all getting into semi-educated speculation here; I don't think we'll ever get any of the details anytime in the near future. Even the claim that Intel utilizes TRIM "immediately" is somewhat suspect; most tests of TRIM done to date involve filling the entire drive, and then deleting the entire drive and running benchmarks to see if performance is the same before, and after, the test. This doesn't leave "partial" blocks as would happen in a more normal situation, and a controller could be smart enough to just realize that the entire drive has been invalidated, and thus everything is a simple erase with no rewrites.

 

Is there any way to be sure that GC is being done? I mean aside from going to the log-in screen, how will I know if my notebook is "idle enough" for GC to occur?

 

I don't really think so. I mean, you can tell when it's not being done, because you'll notice significant slowdowns in your writes, but it's one of those negative response things; as long as you don't notice anything, it's probably working. It's kind of like an airbag in a car; you kinda have to assume it works, because actually finding out that it works is... not terribly useful unless you're in the middle of an accident.

 

Makes sense..Nice analogy btw..Sorry, it's just I'm an obsessive compulsive type of guy so stuff like this "bothers" me.. So I tend to ask a lot of questions..

 

What do you mean by disingenuous ?

 

When you get an SSD and you run Windows 7 you just enjoy it. There's no need to worry about all that.

(In Windows XP it's a different story)

 

The standard dictionary definition of it, which is "lacking in candor". Basically, while technically true, it's not the whole picture.

 

And you know that as a fact that they do a PE then copy when they receive a TRIM and don't count that in their WA factor calculation ?

 

The more I think about it, I don't see how it could be, since TRIM is not a requested write. Whether or not they do a P/E cycle is somewhat irrelevant at that point; the issue is more to the nature of the definition of WA which is, rather specifically, "Amount actually written to the disk" divided by "amount requested to be written to the disk". So unless Intel has decided to go by a different definition of WA than everyone else, TRIM has nothing at all to do with calculating WA. Now, TRIM could have an effect on WA, as a drive that keeps up with TRIM will have more open blocks available, which would reduce WA (because the more open blocks/pages you have, the lower the WA since you don't have to do extra cleanup for that specific write request), but for the actual calculation of WA, TRIM has no effect. Basically, TRIM can "setup" a lower WA, because a cleaner, emptier drive will necessarily have a lower WA than a nearly full one, but it's not actually part of WA.

 

To me, WA is simple. If I write 1K byte to the drive and the drive needs to write 2K due to whatever(be it delayed or now) in order to fulfill my request. that is WA 2.0. IOW If over time I have written 100G to the drive and during this same period of time it has to write to the NAND for 110G, that is WA 1.1

We all know what TRIM is I am just saying that your description(guess) of how Intel does it is unlikely to have their claimed WA factor unless you are saying they have do some 'internal write' and don't count that. Which IF IS TRUE(that I need a bit of evidence) means they are 'disingenuous'. But without such evidence ....

Intel so far is the most opened in terms of doing endurance test. They published the number on their site as product addendum. I failed to find any such information for other vendors. So if Intel is 'cheating' with their number, the likelihood of other vendors cheat even more is even higher.

 

Not wanting to be rude, but why dont you guys carry on this discussion in private and get back to the topic in question......best ssd for sata II notebooks. I dont wanna put anyones nose outta joint but its gettin kinda longwinded.....

 

This is technically ON TOPIC which is 'best SSD for SATA II'. What we are discussing are how Intel fare in terms of over time performance degrade. He made a guess of how Intel does it(which implies unnesseary write to the NAND). If there is no such implication, I won't even bother to reply as I have made my vote already.

 

Technically, I guess your right. It doesnt make for a cohesive read though....its turned into more of a discussion about TRIM and GC and not which ssd is the best for Sata II.

I aprreciate that both are a very important factor in SSD's

Now if there were valid points that were proven with data rather than specualtive viewpoints, that would be constructive. I fail to see how "guessing" how a drive handles TRIM or GC is important or even relevant. If there is data or proof to back it up, fine.

Sorry ... I just hate wading through pages of "conjecture" without any real world proof either way.....there is just no positive outcome of it at the end of the day if there isnt such info.

 

128gb C300 for $200 at newegg using code EMCKFKJ22.

 

Actually, I don't think anyone includes TRIM in their WA figures, when they bother to publish them. The only 2 people where I've heard actual numbers from their own companies (as opposed to numbers given to them by competitors) is Intel with their 1.1, and Sandforce with their "less than 1" (which they probably can't pin down more than that due to the nature of varying compression). And TRIM or not, Intel's WA of 1.1 is only for a "typical client workload", whatever they consider that to be.

By the way, here's an interesting link from Anandtech that covers how the max latency of a drive is often related to real-time garbage collection (and all of the drives involved support TRIM). By those charts, it seems that the new 320s delay garbage collection more than the G2s do/did. I wonder if that's part of how Intel manages to retain life in the 320s despite the lower P/E cycles.

As to the original post, "best" depends on what you intend to install on the drive, and what you're looking to get out of it. For, say, tilleroftheearth, the best for him out of the pure SATA II drives might be a Samsung 470, because of his extreme writes (so 34 nm NAND versus 25 nm NAND is better), and the relatively high sequential speeds. An Intel 320 probably runs a close second for him. For the actual OP, I think we covered that for him, "best" would be a Sandforce, Intel, or Samsung; since he's in it mostly for the gaming, any of the 3 would be almost equally good for him (especially since he was looking for a drive for his M11x). If you have any specific questions about what's best for your uses, we'd need to know a bit more about what specifically you want.

Oh, and for most of us, price is included in our considerations of "best", so then we also have to deal with location. For example, Intel tends to be a lot more expensive than Sandforce in the US, while in Europe, the difference is much smaller or non-existent. This is why for Europeans, it's a lot easier to recommend Intel or Crucial over Sandforce; Sandforce just doesn't have the price benefit that it often does in the States. Although with the release of the M4, C300 prices are dropping, as pointed out by MaX_PL above.

 

Well..personally, I am looking for reliability (Intel of course) but I also want speed (as do we all when buying a SSD).

Value for money is also a key issue. Although I have the means to buy more or less any drive I want, I dont see it like that. I would like to have the best performance/reliability available but, for me, the fact that very few new drives come out and inform you direclty of sata II figures its hard to judge how well a new drive would perform on my platform.

Example being an Intel 510 on sata II or a G3 on sata II....one is 34nm the other 25nm. From what i have read (and understood to a degree) is that the 320/G3 is not that much better than the G2 and with reduced nand, may have less longevity....so how does one balance the two?

The same question could be asked about C300 vs C400 on sata II.....manufacturers just dont seem to care about advertising speeds etc for sata II , its all about "top show"....i am fairly green when it comes to ssd's but i would imagine alot of people may see vertex 3, for example, and think...oooooo...great.

Also C300 is a great performer on Sata II, a strong contender for me but how do the other newer drives fair in comparison. I believe the C400/M4 is not a significant improvement over the C300 and dependant upon price, may be ruled out (unless its really cheap which i doubt)

OCZ I have more or less ruled out due to recent events....although the vertex 3 seems to be performance king of sata III, reliability is unproven and upon track record, i would be pushed to opt for anything OCZ based.

So...as you can see....quite in a muddle over which one to get....if any at this point in time...I may just settle for a nice IPS panel external monitor instead LOL

 

Well, if you're avoiding Sandforce (note that while OCZ uses Sandforce, it's not the only maker of Sandforce drives; G.Skill and Corsair also make Sandforce drives), then the C300, especially with price drops, is probably the best on a SATA II interface when price/performance is considered (I still like the Samsung 470, but it still tends to the expensive side). And the truth of the matter is, unless you're a fairly extreme user, the chances you'll notice a real difference between any of the top drives is pretty slim. Of course, if you like to chase the pretty numbers, go right ahead.

 

how are firmware updates on the crucial drives?

are updates a common thing or are there few of them over the course of a year?

 

Nah..im not a number chaser. Im more of a reliability guy. I did ponder briefly od raided G2's but then fairly quickly decided against it with good advice from here.

I have been so close to buying a 256gb C300 but have read the crucial forums and problems about stuttering on the latest firmware....now that crucial are releasing the C400/M4 my main concern with buying that drive would be being stuck with a ssd without any firmware updates and no cure should the problem arise.

The Intels seem to have not had an issue (G2) and although performance may not be on the blazing side, like you say, its probably unnoticable to the average user.

Maybe my best bet (and one that i have strongly considered) is waiting for the prices of G2 drives to drop after G3 and go for one of them....reliability, proven performance....stable....probably one of the best sata II drives available.

What dya think?

(sorry for my comment earlier, I have trawled the forum and although good advice has been found, I still find myself "tentative" to taking the plunge on any drive.....maybe i over think things!)

 

Totally agree. For the average user (or even somewhat enthusiastic, power user), if you're coming from a HDD ANY SSD will be a HUGE upgrade!! Stop looking at Benchmarks...real world, you won't be able to tell any difference unless you go for those really top of the line.

 

Thats what I was wondering...see above ^^^^^

From what i have read on the Crucial forums, they released 006 from 002 but you cant flash back to 002 if you have 006 or you will end up with a paperweight. Seems that from those forums, loads of people have stuttering issues on 006 firmware - one of my main concerns in buying one.

Given that manufacturers tend to "abandon" their "offspring" when selling a new "baby", that concerns me - to what extent would the old drive be supported in means of firmware. I dont want to be stuck with a brick!

 

Well, the G2s _had_ issues, but those are all in the past (there were some bad firmware releases that would brick drives). I can't say that waiting for G2 prices to drop is a bad choice... as long as they actually do so. The problem being, of course, that this is not guaranteed. And since the G3s are pretty much out now, we should be finding out over the next few weeks if G2 prices are really going to drop.

As I've stated elsewhere, if G2 prices don't drop appreciably, I feel there are better drives for the money, like the Samsung 470 (which is arguably as reliable as Intel). And as for the earlier comment, I'm not too worried about it. Part of the problem with any long thread like this one is that once the OP's question is answered, it's almost guaranteed to tangent, unless there's no real way to answer the OP's question. In this particular case, the OP actually started asking questions about TRIM and GC, which is what started the digression.

 

Well, I appreciate your words of wisdom no matter what. I am not in a hurry, I will wait and see over the next month or so and keep an eye on the reviews etc....

I did look too at the 470 but in europe, (UK) the 256gb version is almost too pricey....way overpriced for a sata II SSD in my opinion....around £450gbp

 

hey stevie,

found this in phil's post in the main ssd thread up top.

New firmware for C300 update - Crucial Community

edit:

and wow just read something i thought was pretty lame.

apparently people who downgraded from 0006 to 0002 voided their warranties in doing so.

 

If that is not a supported scenario and not something recommended by the vendor, you can't blame them.

Basically, I don't want any drive that 'NEED' firmware flash, no matter what vendor. An SSD/HDD is supposed to be plug and use, not something that requires TLC.

 

@steviejones - I've been using the Crucial RealSSD C300 since Oct. '10. No stuttering or other any anomalies to report. Note, I did feel the drive was a bit slow when I received my laptop, but once I flashed to 0006, it felt more responsive. Note, I couldn't flash the drive's [FW] on my x7200 as the flash program didn't support AHCI, so the flash program couldn't see the drive. I had to use another machine to update the flash. That was a bit of a pain.

Boot up times are around 13 seconds. AS SSD


In any case, I don't know what you are looking for in a drive, but I consider myself a satsified C300 customer.

 

The problem is the G2 seems to be going up in price more than going down, with the MIR you were seeing $315 or around, now the MIR is gone and prices between $385-430. I was hopping to catch a 160gb sub $300, but seems its either going to take a lot of time, or they simply wont come down before they are sold out. At this point the Intel 320 160gb seems more viable in price.

Im going to take my chances with Intel 320s, same controller so maybe they already have it nailed down stable and reliable, but no one really knows.

 

I would not pay more for the G2 even though it is a more time proven line. G2 and 320 is a bit like Windows XP and Server 2003 to me. XP was very good, 2003 is just a little bit better in most aspect.

 

I've read crucial C300 doesn't run as well on SATA 2, of course, an average user probably won't notice but if we're considering the "best" SSD, we have to look at "some" numbers..Or else it'll be a pretty subjective thread with people having different experiences due to different systems being used..I believe the C300 performs at it's best on SATA 3 but when you install it in a SATA 2 drive, the performance drops below par with the sandforce and intels..

If reliability is what you're looking for, I'd say go with the G2, but if it's more expensive than the 320s, that just doesn't make sense and you'd be better off going with the 320..They all have warranties anyway. IMO, you'll never get much out of "waiting" since tech is pretty unpredictable IMO..

If you're a "regular" user (like me) meaning you don't really do benchmarks and just want an SSD loading times, which means booting your PC in around 12-15 secs and shutting down in 5 secs, intel is actually (usually) the leader in real-world benchmarks..Like grabbing a stopwatch and timing. We're talking about milliseconds and probably 2 seconds at most though, so it's again really up to you..

I read somewhere that the difference between intel and the sandforce drives can "only" be felt when handling numerous files in photoshop or movie maker..Not sure if this is true..Regarding firmware, all sandforce drives are equal except for OCZ which has "better" firmware..Intel is of course always updated, the intel tool that comes with the package is pretty cool though..

Cheers

 

Where did you read that? The only difference for the C300 on SATA II I can see is that sequential read is capped from 355MB to 250MB (SATA II speed).

 

I saw a review a while back (can't remember where but I'll try to look for it) showing it was below the pack, not by much though, with the sandforce drives..But seeing the benchmarks above, it doesn't seem to be the case..I remember the review stating in the conclusion that the C300 works best on SATA 3 and that the SATA 2 cap hurts the performance..Was something about relying heavily on the SATA 3 port or something..Could be a case to case basis though..But hey, if you can grab one at a really low price, I wouldn't not grab it..I'm just saying you know..

These SSDs are pretty equal, it'll be quite difficult to notice differences between them anyway..

Sorry for the double post, I found the review..Power consumption seems to be pretty bad at sequential writes too..

Crucial RealSSD C300 Review (256GB) | StorageReview.com

 

Multi tasking performance takes a slight hit, shown in the bottom review in my signature and the review on Storagereview.com

It wouldn't stop me from buying a C300 though.. in real life it will be hard to notice.

 

Good find! - thanks. +rep

 

The review didn't specify which [FW] was on the C300. As I mentioned above, once I flashed to '0006', speeds definitely were better.

At the time of purchase, for a SATA II controller, I felt the C300 couldn't be beat for size (256GB)/price.

 

i agree, the review is pretty old, so the new firmware could've fixed this issue..If the price is right, I wouldn't not get a C300..I mean it's a good drive overall..

 

Does the intel toolbox really help all that much for Win 7? I heard it's only useful for xp/vista..I mean is it useful? What does it do anyway? I mean what can be found in the software CD?

 

From the Intel toolbox page.

Intel SSD Optimizer, System Configuration Tuner, Secure Erase.
Check SMART values.
Run diagnostic scans.

 

For normal usage under Windows 7 you don't need it.

 

i just gave in. After to much though, i reach the conclusion that i was buying the 320 for the wrong reasons,

Reliability - very important for me, buts unproven with the new gen, although i would still have much more confidence into it being more than any other drive, still no one knows.

Size - For me 300gb was more for the future than for now, but probably by the time i need that much there will be faster and better ssds, and probably cheaper.

Interface - i was already used to the idea of G3 being sata II, so not much to blame here, but its going into a sata III interface soon, so kept bothering me.

Price - I kinda wished they were priced more like crucial went with their C400, $450 would seal the deal for me, but at $545....

So i went with x25m 120gb, i been happy with 160, performance should be the same, i wish the 160gb were down in price, but doesn't seem like it. At the time i order my 120gb there were 25 in stock after couple of hours 13, seems like people still going for them, even with all this new ssds already out.

Im going to wait to see how Intel 700 series does, the 120gb should hold me till then, just hopping the 200gb doesn't pass $600, but with Intel extreme editions you never know.

 

the 160gb G3 is $295 at newegg after a $125 promo code if anyone cares.

 

Post in the deals section?

 

@ Abula, I think you made a good decision. The x25m has proven reliability and slightly better read speeds. The 320 doesn't seem worth the premium over X25m imo.