Damn it, Anand. For the past month I've been trying to decide which drive is going to be my next SSD, but every time I reach a semi-solid conclusion, you pull something better out of your hat. Stop that!
So much good stuff coming out lately; it really sucks.
I have read many of the SSD articles, mainly because I have one main bottle neck in any of my systems. That being of slow spinning disks in my many drives. The one thing i do not understand though, is why the drives are only in 2.5 inch form factor. Most all of us have 3.5 inch drive bays in every system we own except our laptops. It would seem that if we want higher storage size put these in larger form factor and viola more GB per package. Drive could end up triple or quadruple in size over night. Imagine that.
What? Are you saying that simply changing form factors will result in a magical price drop in NAND? If you put music or photos on an SSD, shame on your affluence/stupidity. On the other hand, I dislike having to buy a conversion kit to 3.5" - I actually have my SSD sitting in my floppy bay with nothing holding it down. On the OTHER hand, making the same form factor for notebooks and desktops means cheaper manufacturing which means cheaper SSDs, so...
as a software developer myself, I would like to give Sandforce the benefit of the doubt or at least cut them some slack.
If I give a customer a beta version of my software and say this is not for official release, there will be changes but you can use it for your own internal testing and then a client starts using that in a production environment only to find out later that a feature is removed by the time the official release comes out. I would like to say to that client, you should not have put beta software in a production environment. However, I would have made sure that the client knows as soon as I know that the said feature would be removed (this is where I think things fell apart on Sandforce's part although as far as I know there is no evidence that they did not tell Corsair, it could be that Corsair was told and simply did not understand the implications of the change).
I know this example is not exactly the same as what happened here but I would not put all the blame on Sandforce. I think they have always said from the beginning that the 1200 would have the lower specs than the 1500.
However, all that being said...I think the only way out of this debacle for Sandforce and all the rest of the people involved is simply to make the updated firmware run at the 1500 specs instead of putting the limiter in there for the 1200's. I bought a OWC LE so it doesn't affect me either way but really who will it hurt by taking out the limiter for the 1200's??? I'm sure they'll have to do something nice to OWC to make up for this but I'm sure they can work something out to OWC's advantage...
That's you problem as a software developer. Read what you wrote again and i am sure you will find your mistake. It's a golden rule in SW development: you never take features out in the final release from the beta release. It looks like you might be lacking a business analyst in your team mate. No hard feelings but the world has gone through serious development years so far, so best practices exists...
I will agree with you that it is not ideal to remove things between beta and release but I would also point out that both Microsoft and Apple have removed features between the beta and RTM of their products. It is not as uncommon as you may think (or perhaps you are just venting). Sometimes, there are performance reasons for things being removed until later when they can be addressed and are re-included in a future update. In an ideal world, this would never happen but that is not the world we live in.
I want to agree, but the difference here is that they didn't just run out of time, couldn't polish it off, interacted badly with something else, etc. They knew from the get-go that this feature would not be available. When you know something isn't going to be in MP, you keep it out of the beta, especially when it's as artificial as this, AND you've promised exclusive access to someone else. That's just bad. I doubt it's malicious, but that still doesn't make it good.
They probably wanted to test the firmware without having to keep track of all those limitations they'd have to add to the 1200 fw. You want to get something as complicated at this fw probably to run, before starting with those probably easy to implement ahm "features".
As long as the firms knew that those limitations would be in the MP fw I really can't see anything wrong with that - as pointed out many other businesses do the same and I don't think MS really needs more business consultants or whatsoever.
Sorry for being a little off-topic (still SSD though)...
Are there any news on when is the Indilinx JetStream going to be released in some drive from any vendor?
Cause as far as I know, this was initially going to be out in end of 2009, then there was a delay... but no further info that I could find more recently.
I applaud those of you willing and wealthy enough to be using SSDs at this point in time. As I've said many times already, this platform isn't stable yet (IMO of course) and if you can wait, just wait it out. All these fiddling around with firmware should worry you as a consumer. It worries me as an engineer! When software/firmware is constantly changing it's usually not a good sign; but that's not 100% true all the time. I pray it's not true with your SSDs.
False. Even Anand himself has had reliability problems with latest-gen SSD technology. There have been widespread reports of high failure rates of non-Intel SSD's rolled out to corporate laptops.
Yeah and HDDs have been without any problems for the last few years! It's not as if Seagate killed a whole bunch of drives with faulty fir.. oh damn. So you're using tapes for mass storage? Those are pretty reliable, not that fast or useable, but who cares about that.
As with most products, do not buy the 1st versions coming out - there is no amount of QA testing in any company that replaces thousands of users with different setups/configs using a product for a few months (as Anand, points out, buy these new very fast SandForce ones, at your own risk).
So, with SSDs, and specially these with new architectures, double that precaution, but any Indilinx or Intel can be considered stable enough, now.
As for money, most people can buy a 30Gb SSD - I will simply ignore people that say that is not enough space. It is enough to have the OS and couple more things - and the OS alone makes a difference! If someone does not want to see that, it is their problem.
Also, I've read of people (at some companies) that upgraded laptops with SSDs instead of buying new ones. Since any 60Gb SSD is enough for work purposes (95% of the cases at least), so $180 (OCZ Vertex @ Newegg) instead of a new laptop, seems like a _bargain_.
In other words, it all depends on your starting point and your expectancies. Of course I would agree in saying someone trying to make a new computer with 1Tb intel SLC SSDs for instance needs to be loaded! ;)
Anand please tell me that your going to have atleast one benchmark with the new Intel® Rapid Storage Technology 9.6. I've heard it makes a big difference.
Yes SSD's are fast, but they are unproven long term. I've read all of the SSD articles on Anandtech and others and I've come to the conclusion that at this time that the technology is not mature enough to warrant the high premium for them. I'll give them another 2 years.
In two years' time how many of the current batch of SSDs will have uncorrectable read errors on data written a week before ? The effective write-read cell lifetime of a MLC flash memory is a joke, and as the processes shrink and the density gets higher, there are fewer electrons/holes available to waste in leakage. And there is no practical way of measuring the residual life of a SSD. Unlike a hard-disk, where a non-correctable block of bad bits can usually be found by some simple write-read tests, a non-correctable block caused by end-of-life cell-leakage cannot be found by a normal write-read cycle - unless you want to wait a couple of weeks between the write and the read..... BTW, just to make you happier, an end-of-cell-life SSD in a computer will have user-symptoms similar to those caused by flaky ram -----mysteriously erratic program-crashes and data-loss.Rewriting the corrupted data might keep things going for a couple of weeks until the cell-leakage again overcomes the SSD error-correction and the PC dies again. Unfortunately the physics-reality of the current generations of flash-memory may upsat all these rosy dreams of the ideal long-term replacement for hard-disks.
SSDs do have their place in a PC.... storing programs or data with low rate of change. Ideal for the OS core of a PC; not AT ALL ideal for ANY use as "virtual memory". Better make sure that the PC's 'virtual memory' is always constrained to occupy the available system ram if a conventional hard-disk is not available for that task instead of the SSD.
but is this the first time in history that everything has gotten better while nothing has gotten cheap. Usually poor dudes like me benefit when all this new fancy stuff comes out, but I'm not seeing the affordable old last-gen stuff....
Progress for the rich. The rest of us are waiting.
Your comment doesn't align with reality. Sales on Kingston's 30GB last-gen have been $70CAD and Intel's 40GB just over $100CAD. US sale prices surely about the same. You are not looking hard enough.
but I really do think they are already more reliable that a spinner. Could you provide more information on why you think there should be no swap file on an SSD?
I mean if the current crop of SSD's all of a sudden fail then that would set back sales by a decade or so. And since variations of these drives, but I'll admit not these exact drives, have been around in military and other uses for a really long time now.
Bottom line, my Corsair better be good to my data or I will be wicked pissed!
It looks like the SSD has shed it's 'awkward teenage years' and is now coming of age. The only real bummer is that short supply is driving the price of nand memory up...thanx a lot Apple. Anyways...for those who have been sitting on the fence waiting for SSD technology to mature, now is your time. Sure SSDs are expensive but now that a single drive can saturate a sata 3Gb/s channel and can sustain the high write speeds, it shouldn't take much convincing. I was pretty impressed with the real world difference that my first gen Apex made (even with the crap JMicron controller.. >< ) over a pair of raptors in raid0. Then came the Inilinx drives which improved 4k writes and throughput. Now, sata2 bw is maxed, 4k writes are no longer a problem and sustainable write speeds aren't far from matching the read speeds. Now that the technology has matured, all we need is for apple to quit hogging all the nand for their crap wear.
I'm usualy a pretty early adopter (I've owned 3 x 58 boards so far :)) but I waited a long time to do the SSD thing. Partially because I already had a smoking fast RAID 0 setup with VR's, and partly because of the reliability issue. I've been running an Intel 160GB G2 for the past few weeks and have no regrets whatsoever. Read times are faster and write times are slower, but they certainly don't feel slower. Boot-up times and program loading loading times are non-existent. Games load faster. I don't hear the annoying churning of the VR's.
Bottom line is that if you have the money and the know-how, there is no reason to buy a mechanical hard drive any more unless you want large capacities/RAID or you just can't afford the SSD. My mechanical hard drive now functions solely for storage and backup images of my Win7 install on the SSD.
Well, Sandforce is the hot topic in SSD's, and since OCZ has 3 separate lines of Sandforce SSD's out, it follow that it would be a busy month for OCZ SSD's. He's just reporting them as they come in. I don't think anyone would accuse Anand of being biased towards a particular company for any reason other than its performance and reliability.
OCZ was the first to support SandForce and thus it's one of the first to bring product to market. We just recently reviewed Corsair's Force drive and I'm awaiting new drives from OWC and other companies as well.
I have two more OCZ drives as well as a new Patriot drive that I'm working on right now, I'm just publishing whatever I've got in the order that I get it :)
You should perform that 100% random data test on an intel drive. I had always assumed that intel was doing some rudimentary data compression to make their random write performance so good. For example, if you create a new text file and then type "hello world" and save the file, you've just consumed 4 kilobytes but in reality most of that 4KB is just a bunch of zeros. When windows is doing a flurry of small random writes in normal daily usage, I bet most of that data is either *unchanged* or is mostly 0's.
It seems like a no brainer to me that a controller would count the number of consecutive 0's or 1's in the data as it comes in. And if there are a lot of 1's or 0's in a row then it would replace all those bits with a flag in the header that says there are X amount of 0's or 1's at location Y.
Of course there is nothing they can do to speed up the amount of time it takes to write a page, even if they can fill a page faster.
Certainly, a smart controller would calculate a checksum of each flash page, and not rewrite that page unless the data has actually changed? I know the vast majority of the time when windows changes a file, it is only a small fraction of the file that actually is changing. Again, I had assumed that Intel was capitalizing on that to get better performance than simply "dumb writing" every single bit of every single file whether it changes or not. It would be nice to have an actual test to see who is doing these kinds of things.
It's not gonna be cheap...but it will be blazing fast. After that, I guess the only way to top it would be a Vertex 2 EX (SF1500 w/SLC nand) with an SAS/Sata6 interface (insert drool here). I'm just crossing my fingers that they'll release a sata6 version of the Vertex 2 and/or Agility 2. And hats off to OCZ for investing the time, money and resources that it took to bring the SandForce controllers to market...they MORE than deserve to have a leg up on the competition when it comes to exclusive stuff such as the special SF1200 firmware.
I am a Swedish anesthesiologist with 2 kids, a research program and far too many hobbies. I am interested in computers but I just don´t have the time to be any good at it. I have just ordered a new laptop (see below) and want to equip this with a SSD. The upcoming Vertex 2 is the one I am thinking about. I have been surfing around during the limited free time I have and Anand and this forum seems to be about as good as things get. I therefore would like to ask a few questions and would be very happy if I could also get some decent answers:
I have ordered a 15" Macbook pro, core i7, 8Gb RAM, crappy 500Gb harddisk. I chose this because i needed a computer which would last a few years and with very good battery life. It also looks pretty cool.
I almost ordered the Apple 256Gb SSD upgrade until I read on the net that they are rubbish. Intel seem to make good stuff but OCZ Vertex are "compatible with Mac". The upcoming Vertex 2 seems to be a nice piece of hardware: My questions:
1. Mac does not have TRIM support. What is TRIM? 2. Since there is no TRIM, Garbage Collection seems to do the job. Is this correct? 3. Does the Vertex 2 have Garbage collection? If so, does it work? 4. The whole "limited lifetime due to limited read/write episodes" is confusing to me. Is this really an issue with new SSD´s? The Vertex 2? Doesn´t garbage collection sort this out? 5. Is the Vertex 2 "compatible with Mac"? Is this even necessary?
Will be REALLY happy if Anand or anyone else could help me out here. I know I sound like a loser, but I just don´t have the time to check these things out myself (and I lack alot of the basic computer know-how that I think is needed).
Agility 2 random 4K (aligned): 10000 Vertex LE random write 4K: 15000 Vertex LE max 4k IOPS: 50,000 Vertex 2 random write 4KB (aligned): 50,000
Since Vertex LE uses Sandforce SF-1500 instead of Sandforce SF-1200, it raises the question on why Vertex LE has lower IOPS at random write 4KB. Is OCZ try to hold cripple Vertex LE (which are bound to sold out at any moment) to boost Vertex 2 sales? Thanks to the unholy exclusivity deal between Sandforce and OCZ, whatever 'solution' Sandforce is working on is going to be bad for consumers. For a bigger picture, what's gonna stop Intel or any other Flash controller manufacturers from making anticompetitve deals similar to the one between Sandforce and OCZ? After all, OCZ also has already used custom firmware on Vertex[1] and Vertex Turbo[2].
The expression "through the wringer" harks back to the days of wringer-washer machines, where the wringer would squeeze out almost all of the water from the laundry before it was hung out to dry on the clothesline. So, to put something "through the wringer" would be to squeeze out all the information you can about it. The expression used in the article "through the ringer" is a ringer, in that it uses the homophone "ringer" for the proper term "wringer". Sorry to put your article "through the wringer"!
Be it 'ringer' or 'wringer', one thing is for sure...you have waaay too much time on your hands. oO
Also will peeps stop whining about Ocz's exclusive deal with SandForce! There's a reason why Ocz are getting such favors....they invested the time, money and resources (not to mention the risk) to help SF to finish developing, testing and prototyping the controllers and firmware and get them to market. SF will be making most of their money from the high end enterprise products (SF1500 + SLC + SAS) and Ocz (and others) will take care of the mainstream and enthusiast market (SF1200+MLC & SF1500 + MLC). Props to Ocz for getting their hands dirty and helping SF to bring this great new controller to market in a timely fashion.
Very serious. I deplore the decay of the English language in the media, and cringe whenever I see egregrious errors appearing in print or on-line. People who publish anything should at least know the proper usage of their own language. Sadly, one-third of high-school graduates don't even have basic language skills. If I and others like me don''t try to stem the tide, pretty soon we'll be reduced to the grunts, coughs, and clicks used by our cave-man ancestors for verbal communication. And don't even get me started on the even more deplorable state of math skills in North America. I used to be a college professor, so I know. The fact that we have this amazing technology at all is due to the valiant efforts of a few heroic "geeks" and "nerds" who are ridiculed by the "cool" and "in" crowd, who like to try to make themselves feel superior by trying to making those much more capable than themselves feel inferior, yet rush out to buy the latest and greatest technology produced by the labors of the very people they scorn. I would have hoped for more respect from of a devotee of AnandTech such as yourself.
Hi Anand, I don't think you have ever discussed the error correcting capabilities of these controllers. I believe it is an important aspect of any controller , be it HDD or SSD. Thanks for all the informative SSD related analysis.
Firstly, thanks for a great review and great site.
From looking at your photographs of the Vertex 2 and Agilty 2 I don't see any difference other than the branding of the memory chips (Micron/Intel). Both use the SandForce 1200 (SF-1222TA3-SBH). Both have the same board revision PCB-0025-X01. Check it out in the gallery.
In theory you should be able to flash the Agility 2 with Vertex 2 firmware and unlock the 4k write limits.
Anyone know where that firmware can be found? No sign of it on the OCZ site.
Just installed an Agility 2 for my OS. Remaining apps and files are on a RAID0. Running Windows 7 on Agility 2.
Now when I do a disc operation such as a resize or anything associated with the RAID0, the system hangs in a never ending reboot of the OS. Goes like this - after initial instruction to resize, it reboots - resizes the partitions properly - finishs and reboots - starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7 - starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7 - starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7 - etc.
These types of operations worked fine when the OS set on the RAID0. I performed many of them. Issues appear to be associated with the Agility 2.
Hi, I have one of these SSD HDD and is dead. I cannot access it. The BIOS cannot see it. The only thing is that green LED is ON and the red LED is flashing ON for one sec. at the moment of powering on the SSD HDD. I would like to repair this hard drive if possible. Any help or suggestions are more than welcome. If anybody knows the meaning of J1 (that internal connector Vcc, Tx, Rx, GND) please help me. Is some kind of serial port? thank you.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
47 Comments
Back to Article
Earthmonger - Thursday, April 22, 2010 - link
Damn it, Anand. For the past month I've been trying to decide which drive is going to be my next SSD, but every time I reach a semi-solid conclusion, you pull something better out of your hat. Stop that!So much good stuff coming out lately; it really sucks.
Confusador - Friday, April 23, 2010 - link
You have a really interesting definition of 'sucks'. ;)abnderby - Sunday, April 25, 2010 - link
I have read many of the SSD articles, mainly because I have one main bottle neck in any of my systems. That being of slow spinning disks in my many drives. The one thing i do not understand though, is why the drives are only in 2.5 inch form factor. Most all of us have 3.5 inch drive bays in every system we own except our laptops. It would seem that if we want higher storage size put these in larger form factor and viola more GB per package. Drive could end up triple or quadruple in size over night. Imagine that.clarketelecom - Monday, April 26, 2010 - link
What? Are you saying that simply changing form factors will result in a magical price drop in NAND? If you put music or photos on an SSD, shame on your affluence/stupidity. On the other hand, I dislike having to buy a conversion kit to 3.5" - I actually have my SSD sitting in my floppy bay with nothing holding it down. On the OTHER hand, making the same form factor for notebooks and desktops means cheaper manufacturing which means cheaper SSDs, so...BlackDragon24 - Thursday, April 22, 2010 - link
This whole SF1200 mix-up says at least one of these things, if not more:Sandforce has no ethics
Corsair has no ethics
OCZ has no brains or balls
I'm trying to figure out which is true.
TonyB - Thursday, April 22, 2010 - link
all the aboveBarshj - Thursday, April 22, 2010 - link
as a software developer myself, I would like to give Sandforce the benefit of the doubt or at least cut them some slack.
If I give a customer a beta version of my software and say this is not for official release, there will be changes but you can use it for your own internal testing and then a client starts using that in a production environment only to find out later that a feature is removed by the time the official release comes out. I would like to say to that client, you should not have put beta software in a production environment. However, I would have made sure that the client knows as soon as I know that the said feature would be removed (this is where I think things fell apart on Sandforce's part although as far as I know there is no evidence that they did not tell Corsair, it could be that Corsair was told and simply did not understand the implications of the change).
I know this example is not exactly the same as what happened here but I would not put all the blame on Sandforce. I think they have always said from the beginning that the 1200 would have the lower specs than the 1500.
However, all that being said...I think the only way out of this debacle for Sandforce and all the rest of the people involved is simply to make the updated firmware run at the 1500 specs instead of putting the limiter in there for the 1200's. I bought a OWC LE so it doesn't affect me either way but really who will it hurt by taking out the limiter for the 1200's??? I'm sure they'll have to do something nice to OWC to make up for this but I'm sure they can work something out to OWC's advantage...
AstroGuardian - Friday, April 23, 2010 - link
That's you problem as a software developer. Read what you wrote again and i am sure you will find your mistake. It's a golden rule in SW development: you never take features out in the final release from the beta release. It looks like you might be lacking a business analyst in your team mate. No hard feelings but the world has gone through serious development years so far, so best practices exists...Barshj - Friday, April 23, 2010 - link
I will agree with you that it is not ideal to remove things between beta and release but I would also point out that both Microsoft and Apple have removed features between the beta and RTM of their products. It is not as uncommon as you may think (or perhaps you are just venting). Sometimes, there are performance reasons for things being removed until later when they can be addressed and are re-included in a future update. In an ideal world, this would never happen but that is not the world we live in.
icrf - Friday, April 23, 2010 - link
I want to agree, but the difference here is that they didn't just run out of time, couldn't polish it off, interacted badly with something else, etc. They knew from the get-go that this feature would not be available. When you know something isn't going to be in MP, you keep it out of the beta, especially when it's as artificial as this, AND you've promised exclusive access to someone else. That's just bad. I doubt it's malicious, but that still doesn't make it good.Voo - Saturday, April 24, 2010 - link
They probably wanted to test the firmware without having to keep track of all those limitations they'd have to add to the 1200 fw. You want to get something as complicated at this fw probably to run, before starting with those probably easy to implement ahm "features".As long as the firms knew that those limitations would be in the MP fw I really can't see anything wrong with that - as pointed out many other businesses do the same and I don't think MS really needs more business consultants or whatsoever.
JohnMD1022 - Friday, April 23, 2010 - link
Tell that to MS :)Voo - Friday, April 23, 2010 - link
Yeah I had to grin too when I read that. So obviously MS is missing business analysts - interesting analysis ;)_Q_ - Thursday, April 22, 2010 - link
Sorry for being a little off-topic (still SSD though)...Are there any news on when is the Indilinx JetStream going to be released in some drive from any vendor?
Cause as far as I know, this was initially going to be out in end of 2009, then there was a delay... but no further info that I could find more recently.
Thanks for any help.
PS: Anand, thanks for the reviews!
The0ne - Thursday, April 22, 2010 - link
I applaud those of you willing and wealthy enough to be using SSDs at this point in time. As I've said many times already, this platform isn't stable yet (IMO of course) and if you can wait, just wait it out. All these fiddling around with firmware should worry you as a consumer. It worries me as an engineer! When software/firmware is constantly changing it's usually not a good sign; but that's not 100% true all the time. I pray it's not true with your SSDs.irev210 - Thursday, April 22, 2010 - link
Maybe at first, you could make the "not yet stable" argument, but times have changed - and they changed some time ago.Taft12 - Friday, April 23, 2010 - link
False. Even Anand himself has had reliability problems with latest-gen SSD technology. There have been widespread reports of high failure rates of non-Intel SSD's rolled out to corporate laptops.Voo - Friday, April 23, 2010 - link
Yeah and HDDs have been without any problems for the last few years! It's not as if Seagate killed a whole bunch of drives with faulty fir.. oh damn. So you're using tapes for mass storage? Those are pretty reliable, not that fast or useable, but who cares about that._Q_ - Thursday, April 22, 2010 - link
As with most products, do not buy the 1st versions coming out - there is no amount of QA testing in any company that replaces thousands of users with different setups/configs using a product for a few months (as Anand, points out, buy these new very fast SandForce ones, at your own risk).So, with SSDs, and specially these with new architectures, double that precaution, but any Indilinx or Intel can be considered stable enough, now.
As for money, most people can buy a 30Gb SSD - I will simply ignore people that say that is not enough space. It is enough to have the OS and couple more things - and the OS alone makes a difference! If someone does not want to see that, it is their problem.
Also, I've read of people (at some companies) that upgraded laptops with SSDs instead of buying new ones. Since any 60Gb SSD is enough for work purposes (95% of the cases at least), so $180 (OCZ Vertex @ Newegg) instead of a new laptop, seems like a _bargain_.
In other words, it all depends on your starting point and your expectancies. Of course I would agree in saying someone trying to make a new computer with 1Tb intel SLC SSDs for instance needs to be loaded! ;)
dmayes - Thursday, April 22, 2010 - link
Anand please tell me that your going to have atleast one benchmark with the new Intel® Rapid Storage Technology 9.6. I've heard it makes a big difference.Rob94hawk - Thursday, April 22, 2010 - link
Yes SSD's are fast, but they are unproven long term. I've read all of the SSD articles on Anandtech and others and I've come to the conclusion that at this time that the technology is not mature enough to warrant the high premium for them. I'll give them another 2 years.kilkennycat - Friday, April 23, 2010 - link
In two years' time how many of the current batch of SSDs will have uncorrectable read errors on data written a week before ? The effective write-read cell lifetime of a MLC flash memory is a joke, and as the processes shrink and the density gets higher, there are fewer electrons/holes available to waste in leakage. And there is no practical way of measuring the residual life of a SSD. Unlike a hard-disk, where a non-correctable block of bad bits can usually be found by some simple write-read tests, a non-correctable block caused by end-of-life cell-leakage cannot be found by a normal write-read cycle - unless you want to wait a couple of weeks between the write and the read..... BTW, just to make you happier, an end-of-cell-life SSD in a computer will have user-symptoms similar to those caused by flaky ram -----mysteriously erratic program-crashes and data-loss.Rewriting the corrupted data might keep things going for a couple of weeks until the cell-leakage again overcomes the SSD error-correction and the PC dies again. Unfortunately the physics-reality of the current generations of flash-memory may upsat all these rosy dreams of the ideal long-term replacement for hard-disks.SSDs do have their place in a PC.... storing programs or data with low rate of change. Ideal for the OS core of a PC; not AT ALL ideal for ANY use as "virtual memory". Better make sure that the PC's 'virtual memory' is always constrained to occupy the available system ram if a conventional hard-disk is not available for that task instead of the SSD.
philosofool - Thursday, April 22, 2010 - link
but is this the first time in history that everything has gotten better while nothing has gotten cheap. Usually poor dudes like me benefit when all this new fancy stuff comes out, but I'm not seeing the affordable old last-gen stuff....Progress for the rich. The rest of us are waiting.
Taft12 - Friday, April 23, 2010 - link
Your comment doesn't align with reality. Sales on Kingston's 30GB last-gen have been $70CAD and Intel's 40GB just over $100CAD. US sale prices surely about the same. You are not looking hard enough.capeconsultant - Friday, April 23, 2010 - link
but I really do think they are already more reliable that a spinner. Could you provide more information on why you think there should be no swap file on an SSD?I mean if the current crop of SSD's all of a sudden fail then that would set back sales by a decade or so. And since variations of these drives, but I'll admit not these exact drives, have been around in military and other uses for a really long time now.
Bottom line, my Corsair better be good to my data or I will be wicked pissed!
Chloiber - Friday, April 23, 2010 - link
I only read OCZ in the past days/weeks...what about other SSD vendors?Taft12 - Friday, April 23, 2010 - link
I was just coming to make the same comment - OCZ sure is buying a lot of front page coverage these past couple weeks!Movieman420 - Friday, April 23, 2010 - link
It looks like the SSD has shed it's 'awkward teenage years' and is now coming of age. The only real bummer is that short supply is driving the price of nand memory up...thanx a lot Apple. Anyways...for those who have been sitting on the fence waiting for SSD technology to mature, now is your time. Sure SSDs are expensive but now that a single drive can saturate a sata 3Gb/s channel and can sustain the high write speeds, it shouldn't take much convincing. I was pretty impressed with the real world difference that my first gen Apex made (even with the crap JMicron controller.. >< ) over a pair of raptors in raid0. Then came the Inilinx drives which improved 4k writes and throughput. Now, sata2 bw is maxed, 4k writes are no longer a problem and sustainable write speeds aren't far from matching the read speeds. Now that the technology has matured, all we need is for apple to quit hogging all the nand for their crap wear.BlackDragon24 - Friday, April 23, 2010 - link
I'm usualy a pretty early adopter (I've owned 3 x 58 boards so far :)) but I waited a long time to do the SSD thing. Partially because I already had a smoking fast RAID 0 setup with VR's, and partly because of the reliability issue. I've been running an Intel 160GB G2 for the past few weeks and have no regrets whatsoever. Read times are faster and write times are slower, but they certainly don't feel slower. Boot-up times and program loading loading times are non-existent. Games load faster. I don't hear the annoying churning of the VR's.Bottom line is that if you have the money and the know-how, there is no reason to buy a mechanical hard drive any more unless you want large capacities/RAID or you just can't afford the SSD. My mechanical hard drive now functions solely for storage and backup images of my Win7 install on the SSD.
Sabbathian - Friday, April 23, 2010 - link
Look, an article about OCZ... cool, interasting... wait, here is another one.... and another, and another!Anand, don`t get to OCZish, cause your great site is starting to look like one big OCZ commercial....
BlackDragon24 - Friday, April 23, 2010 - link
Well, Sandforce is the hot topic in SSD's, and since OCZ has 3 separate lines of Sandforce SSD's out, it follow that it would be a busy month for OCZ SSD's. He's just reporting them as they come in. I don't think anyone would accuse Anand of being biased towards a particular company for any reason other than its performance and reliability.Anand Lal Shimpi - Friday, April 23, 2010 - link
OCZ was the first to support SandForce and thus it's one of the first to bring product to market. We just recently reviewed Corsair's Force drive and I'm awaiting new drives from OWC and other companies as well.I have two more OCZ drives as well as a new Patriot drive that I'm working on right now, I'm just publishing whatever I've got in the order that I get it :)
Take care,
Anand
Shadowmaster625 - Friday, April 23, 2010 - link
You should perform that 100% random data test on an intel drive. I had always assumed that intel was doing some rudimentary data compression to make their random write performance so good. For example, if you create a new text file and then type "hello world" and save the file, you've just consumed 4 kilobytes but in reality most of that 4KB is just a bunch of zeros. When windows is doing a flurry of small random writes in normal daily usage, I bet most of that data is either *unchanged* or is mostly 0's.It seems like a no brainer to me that a controller would count the number of consecutive 0's or 1's in the data as it comes in. And if there are a lot of 1's or 0's in a row then it would replace all those bits with a flag in the header that says there are X amount of 0's or 1's at location Y.
Of course there is nothing they can do to speed up the amount of time it takes to write a page, even if they can fill a page faster.
Certainly, a smart controller would calculate a checksum of each flash page, and not rewrite that page unless the data has actually changed? I know the vast majority of the time when windows changes a file, it is only a small fraction of the file that actually is changing. Again, I had assumed that Intel was capitalizing on that to get better performance than simply "dumb writing" every single bit of every single file whether it changes or not. It would be nice to have an actual test to see who is doing these kinds of things.
Movieman420 - Friday, April 23, 2010 - link
I know for sure that Ocz will be releasing a Vertex 2 Pro (SF1500) as well with both sata and SAS interfaces to choose from.http://www.ocztechnologyforum.com/forum/showthread...
It's not gonna be cheap...but it will be blazing fast. After that, I guess the only way to top it would be a Vertex 2 EX (SF1500 w/SLC nand) with an SAS/Sata6 interface (insert drool here). I'm just crossing my fingers that they'll release a sata6 version of the Vertex 2 and/or Agility 2.
And hats off to OCZ for investing the time, money and resources that it took to bring the SandForce controllers to market...they MORE than deserve to have a leg up on the competition when it comes to exclusive stuff such as the special SF1200 firmware.
DrJohan - Friday, April 23, 2010 - link
Hello!I am a Swedish anesthesiologist with 2 kids, a research program and far too many hobbies. I am interested in computers but I just don´t have the time to be any good at it. I have just ordered a new laptop (see below) and want to equip this with a SSD. The upcoming Vertex 2 is the one I am thinking about. I have been surfing around during the limited free time I have and Anand and this forum seems to be about as good as things get. I therefore would like to ask a few questions and would be very happy if I could also get some decent answers:
I have ordered a 15" Macbook pro, core i7, 8Gb RAM, crappy 500Gb harddisk. I chose this because i needed a computer which would last a few years and with very good battery life. It also looks pretty cool.
I almost ordered the Apple 256Gb SSD upgrade until I read on the net that they are rubbish. Intel seem to make good stuff but OCZ Vertex are "compatible with Mac". The upcoming Vertex 2 seems to be a nice piece of hardware: My questions:
1. Mac does not have TRIM support. What is TRIM?
2. Since there is no TRIM, Garbage Collection seems to do the job. Is this correct?
3. Does the Vertex 2 have Garbage collection? If so, does it work?
4. The whole "limited lifetime due to limited read/write episodes" is confusing to me. Is this really an issue with new SSD´s? The Vertex 2? Doesn´t garbage collection sort this out?
5. Is the Vertex 2 "compatible with Mac"? Is this even necessary?
Will be REALLY happy if Anand or anyone else could help me out here. I know I sound like a loser, but I just don´t have the time to check these things out myself (and I lack alot of the basic computer know-how that I think is needed).
Cheers from Sweden!
Johan
DrJohan - Saturday, April 24, 2010 - link
Please some answers.......!Moonstarr - Saturday, May 8, 2010 - link
Read up on SSD terms and tech here... http://www.storagesearch.comAlso don't hate on your 500g HDD. As long as you keep it under 50% capacity it will work well.
And like Anand said in his article... It's too early for him to endorse these drives.
quanta - Friday, April 23, 2010 - link
Here are the quotes from OCZ's page over IOPS:Agility 2 random 4K (aligned): 10000
Vertex LE random write 4K: 15000
Vertex LE max 4k IOPS: 50,000
Vertex 2 random write 4KB (aligned): 50,000
Since Vertex LE uses Sandforce SF-1500 instead of Sandforce SF-1200, it raises the question on why Vertex LE has lower IOPS at random write 4KB. Is OCZ try to hold cripple Vertex LE (which are bound to sold out at any moment) to boost Vertex 2 sales? Thanks to the unholy exclusivity deal between Sandforce and OCZ, whatever 'solution' Sandforce is working on is going to be bad for consumers. For a bigger picture, what's gonna stop Intel or any other Flash controller manufacturers from making anticompetitve deals similar to the one between Sandforce and OCZ? After all, OCZ also has already used custom firmware on Vertex[1] and Vertex Turbo[2].
[1] http://www.pcper.com/article.php?aid=760
[2] http://www.desktopreview.com/default.asp?newsID=80...
Dr.Neale - Friday, April 23, 2010 - link
The expression "through the wringer" harks back to the days of wringer-washer machines, where the wringer would squeeze out almost all of the water from the laundry before it was hung out to dry on the clothesline. So, to put something "through the wringer" would be to squeeze out all the information you can about it. The expression used in the article "through the ringer" is a ringer, in that it uses the homophone "ringer" for the proper term "wringer". Sorry to put your article "through the wringer"!Movieman420 - Friday, April 23, 2010 - link
Be it 'ringer' or 'wringer', one thing is for sure...you have waaay too much time on your hands. oOAlso will peeps stop whining about Ocz's exclusive deal with SandForce! There's a reason why Ocz are getting such favors....they invested the time, money and resources (not to mention the risk) to help SF to finish developing, testing and prototyping the controllers and firmware and get them to market. SF will be making most of their money from the high end enterprise products (SF1500 + SLC + SAS) and Ocz (and others) will take care of the mainstream and enthusiast market (SF1200+MLC & SF1500 + MLC). Props to Ocz for getting their hands dirty and helping SF to bring this great new controller to market in a timely fashion.
Dr.Neale - Saturday, April 24, 2010 - link
Very serious. I deplore the decay of the English language in the media, and cringe whenever I see egregrious errors appearing in print or on-line. People who publish anything should at least know the proper usage of their own language. Sadly, one-third of high-school graduates don't even have basic language skills. If I and others like me don''t try to stem the tide, pretty soon we'll be reduced to the grunts, coughs, and clicks used by our cave-man ancestors for verbal communication. And don't even get me started on the even more deplorable state of math skills in North America. I used to be a college professor, so I know. The fact that we have this amazing technology at all is due to the valiant efforts of a few heroic "geeks" and "nerds" who are ridiculed by the "cool" and "in" crowd, who like to try to make themselves feel superior by trying to making those much more capable than themselves feel inferior, yet rush out to buy the latest and greatest technology produced by the labors of the very people they scorn. I would have hoped for more respect from of a devotee of AnandTech such as yourself.Dr.Neale - Saturday, April 24, 2010 - link
Oops...I should have said "trying to make" instead of "trying to making". Even I make mistakes!DrJohan - Saturday, April 24, 2010 - link
See comment from friday by DrJohan. The more I read, the more confused I get....Would be great if someone could take a few minutes to answer my questions, if possible.
Regards, Johan
the_encoder - Tuesday, April 27, 2010 - link
Hi Anand,I don't think you have ever discussed the error correcting capabilities of these controllers. I believe it is an important aspect of any controller , be it HDD or SSD.
Thanks for all the informative SSD related analysis.
Moonstarr - Saturday, May 8, 2010 - link
Firstly, thanks for a great review and great site.From looking at your photographs of the Vertex 2 and Agilty 2 I don't see any difference other than the branding of the memory chips (Micron/Intel). Both use the SandForce 1200 (SF-1222TA3-SBH). Both have the same board revision PCB-0025-X01. Check it out in the gallery.
In theory you should be able to flash the Agility 2 with Vertex 2 firmware and unlock the 4k write limits.
Anyone know where that firmware can be found? No sign of it on the OCZ site.
deeppow - Tuesday, June 1, 2010 - link
Just installed an Agility 2 for my OS. Remaining apps and files are on a RAID0. Running Windows 7 on Agility 2.Now when I do a disc operation such as a resize or anything associated with the RAID0, the system hangs in a never ending reboot of the OS. Goes like this
- after initial instruction to resize, it reboots
- resizes the partitions properly
- finishs and reboots
- starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7
- starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7
- starts Win7 boot and flash screen says it has complete disc operation, then reboots Win7
- etc.
These types of operations worked fine when the OS set on the RAID0. I performed many of them. Issues appear to be associated with the Agility 2.
miniature_schnauzer - Thursday, August 23, 2012 - link
Hi,I have one of these SSD HDD and is dead. I cannot access it. The BIOS cannot see it.
The only thing is that green LED is ON and the red LED is flashing ON for one sec. at the moment of powering on the SSD HDD.
I would like to repair this hard drive if possible. Any help or suggestions are more than welcome.
If anybody knows the meaning of J1 (that internal connector Vcc, Tx, Rx, GND) please help me.
Is some kind of serial port?
thank you.