Microsoft Suspends Windows April Update For Some Systems With Intel SSDs

[Christopher1]

Considering that this is a very common breed of SSD's out there today, there is no justifiable reason for this kind of nonsense from Windows and Microsoft.

 

[AndreasSF]

I had both of my systems crash. I have Samsung SSDs on both. The update is clearly not well designed. Seriously sloppy execution for a consumer product.

 

[genz]

Does anyone at Microsoft or Intel test anymore?

Regarding Microsoft, the answer is, "no," actually. Microsoft fired their QA staff with the release of Windows 10, and now the consumer is the tester.

As far as Intel performing testing, that's not really the issue here. Microsoft issued a patch that causes the error condition when encountering certain Intel SSDs. Whether Intel tested or not, it would have been against the last stable version of Windows 10, not against a patch they didn't have access to. Unfortunately, we no longer have long term stable releases for Windows going forward, unless you want to step up to an overpriced Enterprise version where you can actually manage the patching. Welcome to the moving security target that is Windows 10.

Who has time to worry about the truly bad players on the internet torpedoing our systems when we can get instability or more serious downtime every time Redmond decides to shoehorn another batch of patches that haven't been vetted well enough?

Lesson from this is, don't delete your previous version right away, as you are the tester, not Microsoft.

That's just not true. All windows patches are in Developer Builds weeks prior and Preview Builds days prior. Intel definitely would have. Everyone else managed to, and although it's rare on MS for an outright failure because of just how *failsafe* the existing code is, it's also can be issues.

I said Failsafe because that's exactly what it is. MS has bent over backwards for decades to protect you from bad hardware. USBs, PCI, you name it and Windows can do something to keep the device running after a partial failure... provided your drivers check out.

Also, AHCI, which software controls modern SSDs, is an Intel technology (even on AMD boards). So this bug is when an Intel interface driver connected to an Intel SSD on Windows. Windows doesn't know the difference between devices, drivers do.

Apple made an advert out of how Windows turns off RAM timing functions if they don't behave well (and 'remixed' it to say that Windows didn't use their RAM properly, when their RAM shouldn't have worked period due to a bad BIOS data table), but the end user sees his cheapo RAM still working.

Soundblaster can't do 3 streams of 44k audio (stereo out and mic), MS removes it as a requirement and redesigns their audio drivers for this bargain bucket unit (hence 'Soundblaster compatible' on the back of old video games - Soundblasters were the bottom of the barrel).

Intel (and most of the industry) can't make good enough iGPU to meet requirements for Vista/DX10 on launch, so MS hacks together a driver mode that lets the crappy GPU run the system then puts up with endless bug fixes all so you can go without a GPU in the first place. Sure they wanted to sell units, but Intel iGPUs couldn't even do full DX9 without software drawing stuff on the CPU back then.

Windows even keeps dead hard drives around for MUCH longer than Linux and Mac. I often take the Mac to a Windows box if I need to do a recovery.

 

[mnrdunck]

Is it just Intel that is effected? I have been having so many problems with my Plextor M9PeG 256gb NVMe SSD, and now my Western Digital 2TB Black HHD's are dissappearing from my PC.

 

[stdragon]

Apple made an advert out of how Windows turns off RAM timing functions if they don't behave well (and 'remixed' it to say that Windows didn't use their RAM properly, when their RAM shouldn't have worked period due to a bad BIOS data table), but the end user sees his cheapo RAM still working.

Uh, unless there's an API into a UEFI, there's no way in hell an OS can adjust RAM timings. In fact, ECC is a hardware function that occurs at the hardware level, not OS. There's no "behaved well" short of a bit-flip, and again, that's a function of ECC to correct for it. Also, you wont find ECC except in high-end workstations and servers. For consumers, ECC can be found mainly in AMD systems.

Windows even keeps dead hard drives around for MUCH longer than Linux and Mac. I often take the Mac to a Windows box if I need to do a recovery.

What does that statement even mean?? All modern file systems are journaling, so error recovery isn't necessarily in issue on dirty shutdown. If data-loss occurs at the LBA level, it's a HARDWARE problem, and has F-ALL to do with the OS. If a SMART error is going to occur, or the HDD out-right fails, that's irregardless of the underlying OS that controls it. The only value it could have is how much the user or OS hammers on read/write count; but that is it!

 

[antilycus]

The longer MS sticks with their IBM made/purchased OS, the higher the risk of them losing the market. Linux (and yes even MAC OS X) have better implementations of today's needs vs kernel offerings. Microsoft and and Apple aren't engineering O/S's they are just relying on someone else make in (MS paid IBM to make Windows and update it to windows Vista, which the win 10 kernel is still based off of and Apple is just FreeBSD/Unix with a pretty UI...that's also now outdated in feeling). The only truly engineered OS is LInux (though once could say it's ripped from Unix). MS's need for 8 to 16GB of RAM is ridiculous (go ahead, run your system at 4GB for half a year and report how well it turns after that), a RESTART to apply changes that only happen during kernel load (registry) and their constant tinkering of trying to find a GUI that isn't Windows XP is just not working.

 

[stdragon]

@antilycus

That's not true. MSDOS was based off IBM. Windows NT, and forward was completely Microsoft native in-house development.

 

[bigpinkdragon286]

That's just not true. All windows patches are in Developer Builds weeks prior and Preview Builds days prior. Intel definitely would have. Everyone else managed to, and although it's rare on MS for an outright failure because of just how *failsafe* the existing code is, it's also can be issues.

If I may borrow your phrasing, "that's just not true." Developer builds and Preview builds are in no way guaranteed to be the final code released. Even if Microsoft promised that they are, I wouldn't believe them after the nonsense they have pulled at this point with their automatic update system. There's been myriad complaints about failed installs, upgrades, and patching which even if only a percentage is true, is more than should be. You're welcome to disagree or defend their use of their tool the way they have. It is my opinion that they took things too far, and still do.

I remember the preview builds that fixed the Dolby Digital Live issues, that when the actual update was pushed for general release, just happened to be broken again. Wonder how that happened?

2nd, Developer builds and Preview builds have the patching baked in. As many people can attest to, failure modes can and often do arise from or during the actual process of applying the patch or update. These types of failures would not be present in either the developer or the preview builds, which sometimes won't even install. My personal favorite is the need to actually reinstall Windows in some instances of choosing to leave the Insider program.

So, it sounds like what you may be trying to say is, Intel knew about the problem, because they keep track of each frequent release Microsoft makes for Windows 10, and simply did nothing? Intel is already facing less than stellar PR for their behavior last year. Do you think they would let a known defect in their software cause people to start mistrusting the benefits of using an Intel SSD? If there was such a glaring deficiency or issue with Intel's drivers or firmware, why is it only now coming to light? There may be more to this.

Fail safe code doesn't always work, and if you think Microsoft's coding is that great, how is it the only company that can write an installer that takes 5 times longer to install software from an optical disc than is necessary to just do a wrote copy of the data? "Something is rotten in the town of Redmond."

They have tons of Insiders playing with new builds all the time. It's like free QA plus developers (and regular users that like cutting edge) can test and play with new features. If you read the Insider updates you'd know they fix tons of bugs before releasing to current. Second, it's probably a driver issue, and it seems it wasn't widespread enough (in terms of models affected) to crop up prior.

The 600p isn't widespread enough? If Intel wants to pay some of their employees to do QA testing for Microsoft, that's their prerogative, but it only makes limited financial sense if they're going to actually do something about it when a problem is found. Was the problem so obscure that Intel missed it or is this a case of negligence? It clearly isn't obscure enough for Microsoft to overlook.

Funny, any Insider has access to it, but Intel can't manage to get access? AMD released a driver update FOR the April Update right before it started rolling out... now how did they manage that? Weird, it's almost like they were testing with the most current builds of the April Update before the official release.

AMD and Intel may just have different priorities and practices in place. Could be another case like what happened with Windows 2000, where companies held off writing updated drivers because Microsoft was so late with their end of things.

AMD seems to be taking a few cues from their graphics division where they are trying to be preemptively ready on release day. They have a lot of good faith to build. Intel really doesn't seem to be playing the same game. They are an older, more entrenched company with far more assets to burn, still acting in a more corporate manner.

 

[jimmysmitty]

M$ has been very sloppy lately, far more than usual on the QA/QC process.

That said, I'm betting this is Optane related per the article. When it comes to storage, the enumeration and protocols are standard. How the data gets written to the NAND falls upon the drive's firmware and controller; it's a process that's abstract from the OS. As such, I don't see how this could be a vendor specific bug unless it involved a new form of storage addressing technology - and Optane is definitely one of those.

I can verify that it does not affect my Intel 520.

Boy oh boy, things haven't gone so well for Intel over the past year or so.

One day, AMD came along, almost completely stomped Intel out the ballpark with Ryzen, but making their ultimate mark with infinity fabric. Sales dropped rather dramatically for Intel as people started flocking back to AMD.

Then, the Spectre/Meltdown bombs fell, Intel taking another bloody nose in another round.

Intel decides to beat AMD with their HEDT platform by date. The platform limitations and notorious inclusion of i5 quad cores on a HEDT platform had everyone asking "WTFFFFF???? WHY???"

Then Threadripper is released smashing Intel in downright price/performance ratios, today enjoying far higher rankings (amazon etc) with Intel's offerings just about nowhere to be seen. Just recently, Intel chucked their KBL quads into the bin. Yes, they were that useless rubbish - quickly forgotten.

Then Intel shoots themselves in the foot by brute forcing an 18core as a flagship, marred by a ridiculous price and ridiculous thermals to deal with. A 16core TR costs more than twice less.

Then a while later, Intel decides "Hey, lets get stoned out of our minds! Lets release an i3 for our HEDT platform.... we'll call it the 7360X". Luckily after sobering up, they cancelled all the madness. Forgive em oh dear lord, for they not know what they're doing....

Then coffee lake comes along, reminding everyone about the fun and joy about backwards compatibility. Oh wait!

Then they released more Optane drives, reminding everyone that its features are only supported on Intel's own latest platforms. That's right, you can forget about it if your system is generations older. AMD of course also totally locked out. Don't you folks just love living on a planet where proprietary tech is preferred over open technologies? (BTW I refuse to pay Nvidia their free $200 Gsync tech. They can shove it as well).

Then Intel runs over to AMD, kindly begging a solution to their stone age GPU shortcomings. Enter Vega of course, specifically catered to Intel's needs with their NUC's.

Then more news about their 10nm process delays.

Then they hire a whole bunch of AMD professionals to help put out the fires that are coming from the first floor.

Then more news about Spectre NG flaws that could further impact performance.

Sorry Intel, lately you've screwed up so much, yet you gave too little and overcharged for it. My Ivy bridge is having its final days. Soon I will yank it out, toss it over the rooftops like a frisbee, and embrace a spankingly new 8core Ryzen with a smile!

They hired the prior AMD employees for their GPU division which they do for HTPC means, not for their CPU division.

As well so far the majority of CPUs from every vendor, AMD included, have been affected by Spectre and Meltdown.

The last update also seems to affect Nvidia video drivers, interfering with PUBG ability to see the driver as Direct X Version 10. ( I know 10 people having this issue including myself, this is happening on 1060 and 1070 that I know off), also affects another 3d world program called; "Second Life" that I know of.

I had to manually install the latest driver (use the "Clean Install" option from the driver install) and that fixed it for me. Final Fantasy XV would not load until I did that giving me a Direct X 11 error as well.

 

[redteamrules2015]

Thanks Microsoft, you screwed my pooch, I had to reformat my SSD and lost ALL MY SHIT, and I put Win 7 Pro on. wIN 10 BLOWS ANYWAY.!

 

[psiboy]

Pretty sure no one with a Brain cell bought the 600p Most reviews universally panned it!

 

[glaz.miroslaw]

Not only with Intel Optane series problem. With 2500 Pro 256GB too. Zbook 17 G3 Xeon: system freezes after 1803 update. In task manager ssd disk at 100% transer, 5 seconds delay, and blank screen or BSOD. Frustrating, 4 times in a every day.

Replacement SSD (clone system/disk) with Samsung 960 Pro resolves problem, but with anther cost: boot (UEFI) time increases to 60 seconds :-/ But this is HP incompatibility.

Intel, Microsoft, HP big thx for hard work :/

 

[boydo579]

Where's the best place to track a fix for the bug?

 

[stdragon]

Where's the best place to track a fix for the bug?

Issue resolved in Microsoft KB4100403

https://support.microsoft.com/en-us/help/4100403/windows-10-update-kb4100403