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

you would love them if you have them, too.

no, but the mtrons are no worse than ordinary harddrives in your notebook, so if you don't plan to gain in batterylife but have a silent and hellfast disk, then the mtron is way-to-go. most people i know don't really care much about batterylife, don't use their notebooks much on the go. an mtron is a god-send for them.

but yes, the samsungs battery savings are still great and have to be honoured.

 

Also, the Mtrons run hotter, dave. I know, I had one for a while.

 

i have three. none of them run hot at all. on my notebook, the cooler turns on much less even while cpu usage got higher due to faster data accesses.

 

Not sure what you have but the Mobi 3K I had ran way hotter than my Gen 1 Samsung. So off it went. I can live with the battery drain but that was a real issue in my XPS M1330 which was overheating to begin with. No doubt the chassis design for the M1330 is flawed though.

 

you have a dell?

don't tell me i have much space for my hdd and it's cooling. it's a 12" tablet with 1.8" drive in there. still, my mtron is cold and you can touch it. same for the two mtrons in raid0 in my pc. you can easily touch them (and throw them around like in the 24sammy video i just have only 2).

they are absolutely cold.

they are mobi 3500 2x, and a mobi 3000 1.8" zif.

battery is great on my tablet anyways (around 6h), no difference there.

still, sammies use less power (just as intels do). but i guess you just had bad luck with yours.

 

Davepermen, have you ever thought your usage could be really "light"?? Maybe that's why you don't notice things like extra heat and slowdowns.

As you know, experiences are all dependent of usage.

X25-M gets 44MB/s with a fresh drive which equals to 11,000 IOPS. The figures shown by the Supertalent looks something like for a fresh drive.

But don't look at any IOPS higher than 100 if you are a regular PC user. Sequential access is more important at that point.

 

em, no, definitely not. my usage is not light at all.

and, well, experiences may differ. all i want to say is, not all mtrons "get hot and thus shouldn't be suggested as an option". they may work perfect depending on your needs. some may get hot, replace them if they do. or replace the notebook if it's that stupid to not be able to handle a harddisk.

if i would only have light usage needs, i could buy ocz

 

You could be right, but you never know. Couple of people suggested that the Mtrons ran hotter and they replaced it.

Like I mean internet surfing isn't exactly light. Firefox for example does tons of it to save the state in case the browser crashes. Instant messaging programs constantly write data to disk in case the user wants to bring it up, and same with MMORPG games.

Things people normally call heavy usage like blue-ray ripping and video encoding, or first person shooter games doesn't do a lot of random writes.

OCZ MLC is light in the sense of netbook-light.

 

lets see. i do internet stuff like surfing, downloading (no torrents on the ssd, though..), livemessenger. homeserverbackups. movie editing. programming. light gaming. reinstalling to test oses, restoring back from homeserver. music production with ableton live. and more.

and yes, sometimes about all of that at the same time.

no stutter, no hotness, nothing.

it gets hotter than other ssd's, yes. but it's still similar at max to an ordinary harddrive. it should never be worse. if it is, the disk is faulty (maybe a firmware update could fix it? i've done updates to the most actual state).

and ocz for me is not even netbook-light maybe the vertex..

but the sammy is so much less snappy than the mtron that i'm thinking about getting an intel 1.8" one laters. the 160gb ones for 1.8" are not planned anytime soon, are they?

 

Any actual temperatures recorded with the various SSDs?
The Samsung based SSDs report 0 degrees, while the dual JMicron SSDs report 44 degrees; neither of which are accurate, but certainly affect the BIOS setting the fans to turn on/off.
What do the Intel & Mtron SSDs report via HDTune as opposed to actual operating temps recorded?

 

+1 , I agree

 

3 Year warranty They really want you to buy these SSD's..

 

Where the sensors? I was under the impression SSD's have yet to ship with temp sensors...

 

does the samsung support NCQ ? I just put it in my uni macbook, did a resh install, current NCQ status is off in system profiler.

 

There are no sensors, not as of yet anyway. The moment they start getting hot enough, you can be sure OEMs will be putting sensors in 'em

 

I don't think Samsung support NCQ

 

Apparently OCZ is coming out with yet another firmware revision for the Vertex to make it even faster. Will this firmware make it over to competitors, such as Super Talent and G.Skill? I'd love to see this faster firmware in the upcoming Indilix based SLC drives from Super Talent.

Here's a thread with Tony posting some info:
http://www.ocztechnologyforum.com/forum/showthread.php?t=53212

In post 23 Tony claims the following stats about the new firmware, with comparisons being made on the same mobo:
single drive FW1199 (second firmware release) 150 writes 225 reads
single drive FW1275 (forthcoming third firmware release) 200 writes 250 reads

Take it with a grain of salt since it's Tony posting those numbers, but the 225/150 figures are consistent with what people are getting from FW1199, if FW1275 really delivers, and becomes available on other manufacturer's Indilix drives (namely, ST's SLC ones), I'll be very excited.

 

A 128GB SLC for around $600??? Let me know! I am very interested. Interested, not buying yet, but darned interested

 

Hey folks, I finally got around to reading that fatty AnandTech SSD Anthology. I don't really have any reason to distrust AnandTech, but there's so much info in that article; can anyone confirm or deny that it's all correct? Like is the absence of virgin blocks really the one big reason solid state drives (especially in the case of the X25-M) mature over time?

That being said, I hope this article pushes Samsung SLC prices even further down; then I can finally pick up two more and complete my army. Vertex is also solid in my book now; I'm glad they made that firmware revision before release.

I'm kind of disappointed that they didn't test power consumption.

 

well, I really don't know what to say, I mean, I must have drove like a 5 or 6 time more data over my 64gb drive, and, like anand said, my drive should deliver slower performance by now, but, drive is still fast as it was first time...
I mean, wear leveling just don't fit into that theory for me

 

Anandtech's article has me really tempted to pick up a 30GB OCZ Vertex drive for $99 from ZipZoomFly.

Thoughts? It'd be going in my desktop.

 

I think it'd be a solid choice. The Vertexes are priced competitively with Samsung MLC drives of similar capacity... and they're pretty much all around faster, even in a used state. My only concern would be power consumption, but being a single controller drive and all, I don't think it'll be too high.

 

If you are using it for boot drive and some apps, I dont see why not.

Get 2 and Raid 0, then get a 1 TB for Data.

 

I've already got a 500GB and 400 GB SATA drives in the system plus a 200 GB external, so storage space isn't my concern. As long as I can fit Windows + core applications on it, I'll game from the 7200 RPM drive. My entire Windows 7 installation, programs, and day to day setup without games is only about 20 GB. Anybody know the formatted size?

Now to determine if I have enough money, I'm supposed to be making an $8000 student loan payment at the end of this month and I don't want to undercut my own plans.

 

Make sure not to push the vertex too hard until the firmware update came out..
http://www.ocztechnologyforum.com/forum/showthread.php?t=53290

 

$83 on Newegg with promo code EMCLPMX37 ($25 off) and $20 rebate. I'm sold, at that price even with the issues they're describing, worst case scenario I just backup image with Acronis, flash firmware, restore image, and I'm good to go.

This image from Anandtech's article is what finally convinced me to buy an SSD:

 

Daverpermen: The 160GB X18-M are out there. I just checked NCIX Canada for it(cause I live in Canada). I think its more rare than the conventional 2.5 inch ones though, for obvious reasons.

Vertex is really not that less pricey than the Intel X25-Ms here. The X25-M 80GB is going for $550 Cdn and the Vertex 120GB is $615 Cdn, which makes $/GB difference only 1.3x.

ofelas:

It shows the temperature of the X25-M as 0 Celsius.

Commanderwolf:

Virgin blocks, good one LOL.

Yes it is right. To me though they are still only scratching the surface. There are more to be explained.

And don't let people fool you, you don't need anything more than 0.5 MB/s in random writes as a PC user. Those are what 4200RPM platter HDDs achieve and they don't stutter. JMicrons are a different story because they are at 0.02 MB/s.

 

newegg.com has a promo EMCLPMX37 $25 off and free shipping making 30gb vertex $83 you have to be a subscriber to their email deals though.

promo may still be current. can't test it as I already used it X 2 for $50 off.

 

Yep that's what I bought.

To me, $83 to make my PC run tangibly faster and snappier is more important than the $115 I spent net on upgrading to a 4870 and selling my 4850. (Something I will only notice in brand new games)

 

PC Cyber in Ottawa (where I am) is having a sale this weekend on the Vertex 120GB, $349 - $20 MIR. So very tempting!

 

Hmm, I don't think I phrased my question too well... I think I'm trying to ask whether or not the fundamental "flaws" of flash memory (aka virgin blocks, etc) are the biggest contributors to the degraded performance of "used" drives, as opposed to any sort of problems with the controller(s).

Like given that you lose X amount of performance in steady state, does the nature of flash contribute to 100% of X, or does the nature of flash contribute to only 50% of X, and how the controller decides to handle written blocks contribute to the other 50%?

Bleh, I'm not even sure what I'm trying to ask anymore.

Though concerning your last point; for me it's not really about the stuttering, it's a matter of speeding up the slowest part of the machine. In an entire computer, your HDD is the slowest element, so that's why you want an SSD. In an SSD, random writes are the slowest elements, so that's why you want something with high random writes.

 

Well for me it was a matter of there being no SSD's under $350 that didn't stutter. I have wanted to upgrade to one for a while to deal with the "slowest part of the machine" thing, but all the stuttering problems always stopped me from buying a lower priced part. This Vertex drive appears to be the only "affordable" SSD that doesn't suffer from being JMICRON'D.

 

Hmm, thanks for clearing up the question I guess. The reason degradation happens is entirely because of how flash works. How the controller works around reliability problems and random write problems will affect how much the degradation is.

For random writes, look at it this way. If certain SSD achieves 100MB/s but it runs into a brick wall(say it ran out of buffer) then what's going to happen? You'll notice stutter! But another SSD only achieves 2MB/s but can maintain that, you won't notice it.

Bad news on X25-E for database: http://www.mysqlperformanceblog.com...nc-write-cache-barrier-and-lost-transactions/

Basically he's saying RAID 10 with 8 drives is cheaper than X25-E. The X25-E can achieve 5,000 IOPS, but due to losing data with write cache they'll have to turn it off and it'll only achieve 1,200 IOPS and the regular RAID 10 setup with 8 drives are more cost effective for similar performance.

 

I have the Vertex SSD with the FW1199 firmware on my Macbook Pro and I tested it and got 228-240 reads along with 128-138 writes.

 

i just install Samsung 256G in my E6400.
however, the reading is only 160MB/s, and my intel can reach 230MB/s via eSata.
What s wrong!!!

 

I'm really looking forward to what "zfs" could bring to mac in Snow Leopard 10.6...

 

I'd like to point something out and ask for opinions on this and other graphs like it from the Anandtech review...

You will notice for this test - a very good indicator of regular day-to-day experienced performance - that the JMicron drives are *way* faster than not only the fast notebook HDD, but also the powerful raptor HDD. Furthermore, in absolute terms, the JMicron drives are nearly as fast as the high-end Samsung / OCZ / Intel drives.

The story is the same for most of the other tests; the JMicron drives have excellent performance compared to the HDDs. When the JMicrons are better than the HDDs they are usually *much* better, and when they are worse they are usually still very comparable with the sole exception of random write performance where the JMicron drives really hump the dog.

Now I know that most SSD freaks are really only concerned about the top performers and don't have much interest in wasting time discussing the inferior products. But I think this is a major weakness in Anand's review. He claims that the JMicron drives are total crap, yet his own charts indicate that one would usually be way better off with a "POS" JMicron drive instead of a HDD for productivity tasks. The only place where it is obvious that the JMicron drives are terrible is in dedicated small random write performance - something that many (most?) users are unlikely to encounter in regular use. In other words, his conclusion that one should stay far away from the JMicron drives does not jive with his data.

After using a JMicron drive (OCZ Solid Series) for the last 2 weeks I can definitely say that the performance issues are most definitely blown way out of proportion. In day-to-day tasks the difference in performance with my JMicron SSD is so much better than my old notebook HDD is isn't even funny. However, I have noticed that some tasks are definitely much worse. The only times I've been able to induce noticeably poor performance is by editing a large uncompressed audio file with Audacity (time to perform a normalize effect with my HDD is about 60 seconds, SSD time is about 4x that) and when I download new RSS feeds in Windows Live Mail (clicking on a new feed topic takes several seconds to load text but only for the ~15 seconds while new feeds are downloading). The latter issue is mostly a non-issue since I am basically never trying to read new feed topics at the same time that new feeds are downloading since feed downloads normally occur in the background. Anyway, I guess this is what other users have called "stuttering", but my experience has been that my system is still totally usable - and fast - for other tasks that are read-only (loading an IE8 window, using messenger, etc) so personally I wouldn't call this "stuttering" but rather "slowdown" for the specific write-intensive task. I should also mention that I am by no means giving my system a light load: I am constantly running several live services (messenger, mail, mesh), Hostsman/HostsServer, uTorrent (saving to an external drive), Logitech SetPoint, and MS OneNote.

Anyway, none of this really matters. We can be certain that any SSD controllers released from this point forward will not be plagued with the awful random write performance of the JMicron 602A or 602B. This means that essentially any new SSDs brought to market from now on will exceed HDD performance by a wide margin. This is in part due to the coverage of the issue by Anand, so big ups to that nerd.

 

stuttering should be much better on the solid series than on the core series, which where the ones anand tested first. there, the os would block for half a second to second sometimes during non-complex usage. that was terrible. since then, all ocz should have improved, and drives like the solid should be better, but not perfect.

 

I thought the same thing and it really shows the bias in his review. While I understand him and think it is fair for him to say he cannot give his recommendation for those drives, it is too extreme and unfair to say those drives are total crap. There are many happy users of these drives. They have a weakness but it is really not that bad for many (probably most) people.

His calling them "total crap" means it's much harder for me to trust his review/article because I know they are not "total crap".

 

problem is, the weakness does not deserve you to pay more for it.

it's less quality than an ordinary fast drive (say, a 7200rpm one), still you pay much more. you should not have to accept any weakness then.

 

If you look at his graphs, the "total crap" drives beat the HDs (which includes a 10K VelociRaptor) in many of the benchmarks. So how can you say a 7200 RPM is faster? And in access time, an HD will never beat it... and access time is very important.

 

it's more constant. it's like you spend much money on a expensive fast race-car that, when you drive around in your village, turns off from time to time. better get something slower that you can trust in as it always works.

for paying more money than the highest high end of harddrives, you want something elite, don't you? those aren't. those are quite nice, but not always.

 

*Edit: sorry davepermen posted just as I did.

Speaking of small random writes - what is that? Writes to logs? Or what kind of activity would translate to small random writes?

 

any activity which is not about big files. any setting that gets updated, any log, any tiny thing that gets saved when anything on your os updates. every tiny thing in your internet history, in your chat history, in your programs history. every write to the registry etc.

 

I kind of wonder if Anand is being polemical in this regard. He understands how influential his reviews are. By completely disregarding the JMicron drives due to the horrid small random write performance he has basically guaranteed that no manufacturer will dare release a new controller that has the same flaw. The interviewee at JMicron (below) even specifically acknowledges Anand's fall 2008 article and the impact it had on their direction. Also note that apparently JMicron has plans for a new controller in Q3 that uses DRAM cache. Full interview is below, source unknown.

 

It's probably important to note that these activities are not a problem on their own, at least with the latest JMicron controller (JMF602B). They only become a problem when there is a number of these activities that are occurring simultaneously which overwhelms the controller's abilities.

Also note that disabling some logs is one easy way to solve the problem. For example, I don't log my WL Messenger conversations and I have not once had messenger pause while using it (but then again maybe even if I did have logs turned on I wouldn't have a problem). I haven't experienced IE8 having any pausing/slowdowns either which I am kind of surprised/confused about. I would think that the number of small writes performed by IE8 would cause me problems...

 

Sorry, one last point

The "total crap" solid series 60GB goes for around $100 - $130. This is barely more expensive than a 7200rpm HDD. While it is true that the 7200rpm HDD will have much more capacity, that is a non-issue if performance and not storage is your goal.

 

Thanks for clarifying.

 

sure, but for a that much higher pricepergb you get something that doesn't always outperform your ordinary system.

instead of getting 320gb for 70$ you get 60gb for 100$+, and may get stuttering and thus have to reduce your workload to be not that much multitasking.

i don't care about the storage myself much as you may know. but a lot do.

i do care about a stable system that i can rely on. ocz don't give that to me.

 

I prefer the speed of an SSD (even the Solid series if I want something on a budget) over a 7200 RPM HD. I'm sure that a Solid is faster in normal use than a good notebook spinner, and in many cases, even the VelociRaptor. This "stuttering" that people talk about is way overblown - at least with the Solid series and what I would consider normal usage. And of course an SSD is dead silent.