Microsoft quietly cuts off Win7 support for older Intel computers

If you have a Pentium III, for example, you may no longer be able to install Win7 Monthly Rollups or Security-only patches, in spite of Microsoft's promise to support you until January 2020. It’s all about SSE2 and some retroactively fudged documentation. Will anybody notice?

laptop displaying binary code with bandages on its screen

If your PC doesn’t run Streaming Single Instructions Multiple Data (SIMD) Extensions 2, you apparently won’t be getting any more Win7 patches. At least, that’s what I infer from some clandestine Knowledge Base documentation changes made in the past few days.

Even though Microsoft says it’s supporting Win7 until January 14, 2020, if you have an older machine — including any Pentium III — you’ve been blocked, and there’s nothing you can do about it.

Here’s how it happened. Back in March, the Win7 Monthly Rollup, KB 4088875, included a warning about SSE2 problems:

A Stop error occurs on computers that don't support Streaming Single Instructions Multiple Data (SIMD) Extensions 2 (SSE2).

I talked about the bugs in KB 4088875 — one of the buggiest Win7 patches in recent memory — shortly after it was released. At the time, the KB article said:

Microsoft is working on a resolution and will provide an update in an upcoming release.

We saw the same bug crop up in the March Security-only patch, KB 4088878, and then in the Monthly Rollups and Security-only patches for April and May. By May, the KB article had been modified to say:

Microsoft is aware that some customers have reported that network drivers are intentionally uninstalled, then fail to reinstall after applying the May 8, 2018 update. This can result in the loss of network connectivity.

Microsoft is presently investigating and will provide a status update when the investigation is complete.

Then, suddenly, on May 25, the KB article was modified to drop that explanation. The patch didn’t change. But the documentation did. It was changed to the old language:

Microsoft is working on a resolution and will provide an update in an upcoming release.

Which, it turns out, is a lie.

An eagle-eyed AskWoody denizen, DAVe3283, wrote to me yesterday with an interesting observation. Prior to June 15, the June Monthly Rollup article KB 4284826 and the Security-only article KB 4284867 included the old verbiage about “Microsoft is working on a resolution and will provide an update in an upcoming release.”  

Sometime on or after June 15, the June KB articles for both the Monthly Rollup and the Security-only update were modified to remove the “Known issue” and its resolution — and all of the prior KB articles were changed to include this bit of advice:

Upgrade your machines with a processor that supports SSE2 or virtualize those machines.

In other words, Microsoft isn't going to fix the problem. It doesn't even mention the problem, starting with the June patches. And the changes were made without warning or explanation on June 15.

Sure enough, DAVe3283 confirms:

I got the chance to test the new patches on my Pentium III. After installing just KB4093108 (2018-04 Security Only update), it wouldn't boot-- though I don't get the BSOD mentioned in the KB article. This could just be a BSOD before the GPU driver loaded fully? This was to determine if the BSOD was intermittent, or reproducible. It is very reproducible.

After a quick restore from backup, I tried installing all the updates April through June (up to KB4284867)) in one go, but got the same no-boot.

Restore from backup, and installed only KB4284867 (2018-06 Security Only update), but once again, no-boot. This is despite the KB page claiming "Microsoft is not currently aware of any issues with this update." If I can figure out how/where to report it to them (since they shut down the forums), I will, though I don't expect any action taken.

I think us Pentium III owners (is it just me at this point? ;) are getting thrown under the bus with Windows 7 now. Which is kind of interesting, because my dual Pentium III 1.4GHz system meets the minimum requirements, and should be fully supported. On the other hand, my Ryzen 1700 isn't supported, but runs Windows 7 just fine (with the wufuc patch even enabling updates). I can't upgrade the P3 to Windows 8, let alone 10, for the same reason: SSE2. Oh well, tally up one more PC on Group W.

To recap: Up until June 15, Microsoft was promising that it would fix the bug that prevented Win7 Monthly Rollups and Security-only updates from installing on older pre-SSE2 machines. After June 15, Microsoft wrote off the pre-SSE2 population, without notice or fanfare, and retroactively changed the documentation to cover its tracks.

Although you can’t retrieve older versions of KB 4284867 from the Google cache (it seems that Microsoft holds sway over Google, at least in terms of cached support documents), the June 15 version of KB 4284867 — the June Win7 Monthly Rollup, snapshot taken at 20:49 — is available on the Internet Archive. It clearly promises a solution.

Susan Bradley notes:

IMHO it goes to showcase that these Spectre/Meltdown patches (I think that's the root cause) are impacting performance and supportability more than what Microsoft and Intel are willing to admit publicly.

Sure, the Pentium III is an old workhorse — the last ones were sold in 2002 — but Win7 is supposed to be good until January 2020, no matter which old horse gets whipped. If Microsoft unilaterally and with no announcement pulls Win7 support on Pentium IIIs, what’s next? And why the fancy footwork to alter the old KB articles?

Thx to DAVe3283.

We’re holding a wake for older Intels on the AskWoody Lounge.

First look: Office 2019’s likeliest new features
Shop Tech Products at Amazon