RSSBad Vista-to-Windows 7 upgrade experiences #1: 'Hosed' Intel SSDs
By Scott M. Fulton, III | Published October 28, 2009, 12:05 PM
Although we had good reason to expect that most folks' experiences with Windows 7 upgrades this past week would be, as we put it, "without the crap," the exceptions are starting to show up. One of the more serious cases involves Intel, which has withdrawn its latest solid-state drive firmware update after multiple reports from disgruntled users of complete storage system failure following their Windows 7 upgrades.
The new firmware, along with Windows 7, was supposed to support a new internal file management methodology called TRIM. Its purpose was to compensate for a problem typical of memory-based storage, as opposed to traditional magnetic disks: Since memory systems must keep track of their contents even some of those contents aren't really in use, over time, SSDs' performance can lag. While traditional disks don't have to retain a memory of the contents of sectors pointing to "deleted" files, SSDs do...and they can't wipe the contents of those sectors individually. Instead, they have to wait until entire blocks become disused -- which happens less and less often as drives become more and more fragmented. TRIM was supposed to overcome that deficiency with a kind of self-optimizing mechanism, letting SSDs wipe blocks more often, thus overcoming lags and keeping performance levels high over time.
The trouble appears to be that something in the Windows 7 RTM distribution wasn't ready for TRIM after all. Though Windows 7 appeared to work fine just after installation, soon afterwards, Intel SSD owners were finding they couldn't sustain a reboot.
"I removed the drive and put it in an external case and went to the desktop. Guess what, the partition was raw!" reported one contributor to Intel's support forum early this morning. "Deleted the partitions and reinstalled Windows, and that seemed to do the trick. After an hour or two, a message came up and said that my drive was about to fail and S.M.A.R.T. reported it bad!"
That contributor was joined by a flood of similar reports all over the major hardware forums. On HardForum Monday, there was this: "I've scared myself off of SSDs for a few years and would probably never touch an Intel SSD again. I'll gladly take a 5% real-world performance hit for a drive that doesn't have a history of bricking itself when a firmware update is applied."
The firmware update was part of a comprehensive SSD Toolkit from Intel released just last Monday, which had promised to boost write speeds by as much as 40%. As company marketing director Pete Hazen said at the time, "We are encouraging our 34nm customers to download the new firmware update today. Not only will Windows 7 users receive the performance enhancements of the Trim command, but so will our Windows XP and Vista users."
Reviewers of the TRIM firmware appeared to confirm that figure and did not report problems. One of those positive reviewers was Anandtech, which was also first to update its TRIM review with Intel's comment that it was pulling the firmware update due to problems. Up to that point, commenters' complaints had focused on Intel's lack of willingness to extend TRIM support to owners of older SSD models, which cost much more in the early going than they do now. As one contributor noted, "I don't think that you should lose the wiper and TRIM support for being an early adapter, it does not make sense."
PC Perspective's Allyn Malventano believes the drive-hosing issue must not be widespread, noting it did not happen to him during his review, which involved an Asus P6T motherboard with an Intel Core i7 920 CPU.
http://www.promdressonsales.com/
http://www.weddingdressart.com/
http://www.dresses4dancing.com/
http://www.dressupinyou.com/
http://www.hidresses.com/
http://www.digdress.com/
http://www.promdress2go.com/
http://www.promdress2u.com/
I don't find the use of the motion sensor in the iPhone to be counter-immersive (and yes, the adjective exists, as of the second you posted the article, that's the way languages remain dynamic 8D), althouogh some executions of it are a bit silly (like shaking the unit while playing Tap Defense to activate Earthquake Towers, very immersive but it disrupts gameplay as you have to reestablish
Score: -2
|For what it's worth, most of the what I've read on Intel's and other forums are that those with problems reported issues not after the initial firmware flash, but after Windows 7 installs an updated driver for the "new" drive.
Reboots were successful after the flash. Then, Windows detects new hardware, installs a driver, reports that the installation was successful and requires a reboot, then poof.
Legacy / Compatibility (IDE) or AHCI BIOS settings did not seem to matter.
The fact that Intel has pulled the firmware is perhaps simply a precautionary measure, and should not be construed as definitive proof that they are solely to blame for this problem. It could be a problem with the firmware in conjunction with whatever updated driver that some affected users have reported that Windows installs after the flash... or, it could simply be the driver for that particular (series of) drive(s).
Nevertheless... a lot of speculation at this early stage.
Score: 0
|Either way it'd be a bad Intel driver or a bad Intel firmware update.
Yet it's Windows 7 problem...
Go figure.
Score: 1
|"Either way it'd be a bad Intel driver or a bad Intel firmware update."
From what I've come to understand, Intel's own IMSM drivers do not pass the TRIM commands to the drive (go figure)... only Microsoft's drivers included with Windows 7 do.
http://www.anandtech.com...doc.aspx?i=3667&p=3
Score: 0
|Microsoft got those drivers from...whom?
(hint: Microsoft doesn't write drivers for other companies hardware)
Score: 0
|Ordinarily, I would agree wholeheartedly with you. Typically, that's how it works.
However, from the article:
"There’s a major problem with TRIM today. The only Windows storage drivers to support it are written by Microsoft. The Intel Matrix Storage Manager (IMSM) driver will not pass the TRIM instruction to your SSD. This means you can't use anything but the drivers that ship with Windows 7."
That being said, I seriously doubt that it's a model-specific driver. It's merely a standards-compliant storage driver designed to pass ATA8 TRIM commands to a compatible drive... of which the 34nm X25 G2 is now, after the firmware update. The firmware adds new features to the drive, hence making the device appear new to the OS and in turn invoking the updated storage driver (again, not necessarily a specific driver meant for Intel drives).
That in and of itself seems to contradict my theory that it has anything to do with a Microsoft driver... especially if it's not a model-specific driver. The "bricking" seems to be an extremely unpleasant side effect of new drive features being enabled and accessed through the updated driver... features that a screwed-up firmware introduced. Until those features are enabled through the new storage driver, the drive appears to work properly, hence the successful initial reboot after flashing.
By no means am I trying to insult your intelligence... you should know me better than that by now. Just thinking out loud, I guess. =)
Score: 0
|No doubting you, but....
Considering the source of "the article", I'd be much more likely to take it as fact if it came from "somewhere else"...
If you get my drift...
Score: 0
|Heh, fair enough. =)
Also, for what it's worth, I've yet to find any reports of these drives being bricked on any supported Linux distribution (kernel 2.6.28 and up with EXT4 fs)... the only other OS besides Windows 7 that supports passing TRIM commands to compatible drives.
Surely, there must be at least one Linux user who has performed the firmware update that is affecting so many Windows 7 users... or maybe not. *shrug*
Score: 1
|How to install Win7 fresh with an u pgrade disc: http://www.winsupersite....stall_upgrade_media.asp
Score: 0
|microsoft has designed clean installs with upgrade media's to be good for only 30 days.
i wouldn't tell people to waste any time with the methodology above.
Score: -1
|This is a normal part of the shake-down cruse. Nobody wants to see it happen to them but it is the sort of thing that shows up at GA. Mac had their guest account issue and now 7 has its' 'Hosed' Intel SSDs. The more important point is how responsive they are in analyzing the problem, crafting an effective solution and getting it in the users hands.
Score: -1
|This isn't Microsoft's issue. Intel needs to respond and update their firmware. From the sound of it, the only possible win7 solution is to stop Win7 from using the feature....the OS uses it correctly. The drive apparently doesn't.
Score: 4
|@ PC_Tool: How did you determine that? The author seems to be suggesting the opposite conclusion.
Score: 0
|Because the author made some false assumptions?
TRIM is a technology used by many popular and arguably more widely used SSD drives than just this specific Intel SSD. None of these other drives are having these issues in Windows 7. That right there is a dead giveaway.
The "data" in the article comes from reviewers with their own assumptions. There is no "data" from either Intel or Microsoft. Funny that, eh?
Also note Intel has pulled the firmware: http://news.zdnet.co.uk/...0000091,39843949,00.htm
Why pull "good" firmware?
Score: 8
|Mod down, for making sense. =)
Score: 1
|I do believe PC_Tool would be correct on this one.
Score: 2
|The root cause here is not Win7, but Intel's firmware! To suggest otherwise is pandering to those who want Win7 to fail.
I totally agree with Thrax's assessment above. I'm using an OCZ Vertex with TRIM enabled right now, so Win7 handles it fine and some SSD vendors do as well.
Score: 4
|@ shoek: I, for one, do not want Windows 7 to fail. The author is suggesting a problem in Windows. "The trouble appears to be that something in the Windows 7 RTM distribution wasn't ready for TRIM after all."
Score: 0
|"The author is suggesting a problem in Windows. "
...and what is the author using to back up or source that suggestion?
Unverifiied comments on other forums.
This isn't citing knowledge or fact, it is assumption.
What we do know: Other SSDs that support TRIM do not seem to be having these issues. Microsoft has made no statements. Intel has pulled the firmware.
I did the math. At worst "we don't know 100% yet". At most, it seems exceedingly likely that this issue, being that it is only affecting Intel drives with this update, is an Intel issue, not a Microsoft one.
Score: 2
|Honestly, I can't figure out what the hoopla is about a broken technology. We're talking after all about a technology that only allows a rather limited number of whites before it, oh I don't know, DIES. Extend the life of the medium to about ten years of heavy I/O (with full-duplexed reads and writes) and then I'll think about paying ridiculous prices for 1Tb drives, never mind someone's buggy firmware.
Until then, it's a really good example of the "more money than brains" category.
Score: 2
|wouldn't it be a bad experience all round, Vista or Windows 7 as intel released buggy firmware? not really anything to do with upgrading Vista to 7...
if i'm wrong correct me, headline is misleading
Score: 8
|Not necessarily. If the issues with the firmware lies only in the automatic TRIM, then Windows 7 is the only OS that would show signs. Linux too, to a lesser extent. As far as I can tell, the Optimizer for XP and Vista performs a similar, but not identical, function called garbage collection (GC). GC has to be scheduled, and it's not just TRIM that must be started by a user.
Obviously there are a lot of ifs/thens here, but Intel is keeping us in the dark on this one, so I'm just speculating.
Score: 2
|I think it may be as Thrax suggests. However, the author of this article failed to make this clear at all.
Score: 2
|agreed
Score: 1
|"The trouble appears to be that something in the Windows 7 RTM distribution wasn't ready for TRIM after all."
Windows 7 is ready for TRIM. The OS contains a standard implementation of the ATA8-ACS2 draft standard (Here: http://t13.org/Documents...oposal_for_ATA-ACS2.doc). Linux since kernel rev 2.6.28 also supports this standard.
The ATA standard operates on a level beneath the file system. It describes how drive controllers and devices communicate with the OS which must in turn comply with the ATA spec to perform drive I/O in an expected manner. That's the point of a standard.
The real root of this failure lies in Intel's firmware which had some sort of error. Obviously we will never know what, but clearly there is a bug in the firmware or the flashing process.
The real proof of this assertion is in the sheer number of drives which *work just fine* with TRIM on Windows 7, with none of the issues exhibited with Intel's drive. These drives include the Crucial M225, OCZ Vertex, OCZ Vertex EX, OCZ Agility, OCZ Agility EX, SuperTalent SuperDrive and the G.SKILL Falcon. All of them use the Indilinx Barefoot SSD controller and have exhibited no file system issues.
If you want more proof, the HardForum quote suffices:
"A message came up and said that my drive was about to fail and S.M.A.R.T. reported it bad!"
SMART failure errors are indicative of failing NAND cells. These messages are invoked when the drive is approaching its write cycle limit as programmed into the drive's firmware (based on cell endurance). Obviously this dude's NAND cell's didn't go tits up over night, so something else is to blame, and that is a controller with bunk firmware not appropriately talking to the cells.
Secondly:
"Up to that point, commenters' complaints had focused on Intel's lack of willingness to extend TRIM support to owners of older SSD models, which cost much more in the early going than they do now. As one contributor noted, "I don't think that you should lose the wiper and TRIM support for being an early adapter, it does not make sense."
It's a nice complaint, but it ignores that the first-generation Intel SSD controllers do not support the ATA TRIM spec. These people can wish and want and hope until the cows come home, but X25-M G1 drives would require physical alteration for a new controller to support it.
None of these issues are an indictment of TRIM, Windows 7, or the viability of flash-based SSDs. Intel simply ****ed up, and people are looking everywhere but at the culprit. Lord knows why.
Score: 15
|Wow an informative post on bn. Will wonders never cease?
Score: 1
|It's about time someone realized that TRIM by itself does not cause data corruption at all, ever. If an SSD with TRIM support running under Windows 7 or a version of Linux/Unix that supports TRIM fails it is always a problem with the hardware in the SSD drive and never, ever a problem with the operating system.
Score: 0
|Very informative post. Thanks Thrax!
Score: 2
|Intel should have found this problem months ago. I wonder if their QA team is asleep at the wheel.
Score: 6
|This is why I always recommended the full install to anyone, regardless of the chipsets, hardware n such, full installs are always the best when upgrading to a new Operating System anyways in the long run :)
Score: -2
|From what I understand this wasn't an issue with upgrading Windows versus clean installs of Windows. Rather, it was an issue with a firmware bricking Intel's SSDs, causing SMART errors and thus not allowing either an FFR or an upgrade.
Score: 1
|I haven't experienced any problems running the RTM on my Intel SSD - but this article definately has my attention and I will be keeping a closer eye on my system. Thanks Scott.
Score: 1
|Interesting. SSD's are still relatively new-to-market, so we should probably expect more of these issues. Won't personally be ordering any for at least a few years yet.
Score: 5
|Agreed. Although for me, and mostly everyone else, the speeds of these drives are what has my eye, however, the pricing it just too steep for me to acquire any sizable drives. In a few years that will change. And now these newly cropped up issues only solidify my stance to wait. So in a few years all of my concerns will be worked out and SSDs will be smooth sailing.
Score: 3
|I agree that the price is a little steep for those of us just looking for large storage options or fast sequential writes/reads, but obviously (and as you said) this isn't where SSDs shine and those weren't the reasons I bought my Intel X25m. So far I've been enjoying instant read and write access which has in turn cut boot time to under a minute and very high shock ratings while has in turn allowed me to vigorously shake my laptop while booting/copying files, along with the added bonus of longer battery life.
However, while I respect PC_Tool's decision I also disagree. I don't think that anyone should be waiting as long as a few years to get one if a problem like this is all that's holding you back. I also don't think that problems with Intel are indicative of problems you might run into with other SSD manufacturers (if there have been any such problems - I only have experience with Intel). Sure being on the bleeding edge hurts sometimes but this was easily avoided in my case by just not installing the firmware the day it came out. I'll give it a week or maybe two and then Intel and Microsoft will have all their ducks in a row and it will be perfectly safe to reflash my SSD. Then I'll be safely enjoying the new features a few years before everyone who chose to wait.
Score: 0
|