Spectre - Meltdown Security Vunerabilities

[wevsspot]

So as most everyone knows by now, earlier this month it became public knowledge that there are a couple of potential exploit risks that exist in processors released in the last 20 years or so.  Initially it was reported that only Intel processors were at risk, but as the investigation has broadened it appears that some AMD and ARM processors could be vulnerable as well.  I'm placing this here in the OS:Microsoft section since we have a large group of enthusiasts that are running Intel and/or AMD processors and also happen to be on some version of the Windows operating system.

 

As it applies to reports of performance degradation, I've installed the updates on five different Windows 10 Pro (64) machines running various Intel processors from an Intel E6550 up to an Intel I7 3770K.

In day to day use all of the machines appear to be functioning exactly as they were with the exception of my main gaming machine (which I'll get to in a minute).  The four remaining machines are used in fairly standard fare, web browsing, emails etc.

 

My primary gaming rig presented an interesting case study however.  The release of the security updates coincided with my installation of two new SSDs to replace an aging SSD Raid0 array.  The new SSD array was installed and the PC was bench marked prior to the release of the security patches.  Initial disk benchmarks were exactly where I had expected them to be.  Sequential read / write, random read / write 4K, que depths etc. were all on par with claimed OEM specifications and the small set of data I was able to find online from reviews and other end users.  Subsequent to my fresh install and bench marking activities, Microsoft released the January, 2018 security patches.  Since I had anecdotal evidence that there could be some system slowdowns based on what I was reading, I decided to re-run my entire suite of bench mark tests.  On my gaming rig all of the synthetic graphics and productivity bench marks were in line with what they were prior to the security patches.  In some cases the benchmark results were slightly better, in others they were slightly worse, but each within the margin of error or repeatability.  The one anomaly I observed was in the disk tests.  Regardless of the bench mark suite used (AS SSD, ATTO, Crystal Disk) my random read speed and random write speeds decreased after the security patches were installed.  This was especially evident on the AS SSD 4K test where my speed dropped from 38 MB/s to 27 MB/s.  Sequential performance, higher que depths and access times remained within the margin of error.  Scratching my head on these results, I broke the array, secure erased each SSD, rebuilt the array and installed Windows again, but left my network cable unplugged so Windows couldn't automatically download and apply the updates.  I re-ran the tests again.  SSD Raid0 array at 4K returned to the same as it was after my initial install.  Then I connected my network cable and Windows downloaded and applied the updates.  I then ran the tests again.  My first observation wasn't an anomaly, 4K random performance dropped again after applying the updates.  I don't have any explanation, but it appears that the patches do have an impact on random disk performance (at least on SSDs and/or RAID arrays).  Fortunately, other than the synthetic disk benchmark goes, nothing else seems amiss. 

 

Now, my older AMD laptop with an nVidia MCP77 chipset and AMD Athlon X2 processor was a completely different animal all together.  The January security patches completely broke that machine and put it into a "no-boot" condition.  After my 3rd reinstall with the same results I finally found a Microsoft patch that fixed that problem.  For those of you suffering the same situation with older AMD based gear, find this update; 

 

KB4073290

 

https://support.microsoft.com/en-us/help/4073290/unbootable-state-for-amd-devices-in-windows-10-version-1709