Following upward to our initial testing of the Meltdown patch for Windows 10, today we're looking deeper into the matter by testing a patched desktop system, past addressing the two at present famous security flaws, Meltdown and Spectre, past applying the Os-level patch and a firmware update, more precisely a motherboard BIOS update.

If you read our previous commodity on the affair, information technology came within 24 hours of the emergency Windows x patch release intended to address the Meltdown vulnerability. We ran tests that made sense from the perspective of a desktop user and we found at that place was virtually no impact on gaming performance and no bear on for content creators. At that place were all the same a few troubling results for NVMe storage devices, mostly impacting 4K read performance. Since so other swain tech media outlets have published similar findings.

However the Windows patch but addressed Meltdown, and by now you've no doubt go familiar of the second vulnerability called Spectre. Considering Spectre is the consequence of a fundamental CPU design flaw, it can't exist fixed, at to the lowest degree not entirely. The firmware update needed to right it mitigates the problem, but doesn't completely address the vulnerability.

This is nonetheless primarily an Intel CPU flaw. AMD's official discussion is that one of the ii Spectre variants doesn't impact them at all, while the i that does is hands resolved by a software update that shouldn't impact performance in any meaningful manner. Variant "three" which is Meltdown, doesn't impact AMD at all. We've yet to properly test whatsoever AMD CPU ourselves, merely this is based on the official information nosotros have so far.

Since publishing the Meltdown benchmark we now have access to BIOS updates that deliver a microcode update which will mitigate the Spectre flaw on Intel'south latest Z370 platform. The update changes the behavior of Intel's branch prediction to be less ambitious. This will probable mean less effective branch prediction and that ways reduced IPC as the execution pipelines wait for retention access more often.

Of form, we'll get to the benchmarks in a moment but before nosotros do here are a few additional notes. As of writing, the just motherboard manufacturer to release an update is Asus and so far they've only addressed their Z370 series of motherboards. In order to complete this test we rushed out the door and purchased an Asus TUF Z370-Plus Gaming.

Once nosotros had that on hand, we benchmarked the Cadre i3-8100 without the Windows update, then tested a 2nd time with the Windows patch applied, and then a third fourth dimension with the Windows update plus the latest BIOS which includes the microcode update. We've included some updated Core i7-8700K benchmarks as well.

For verification, one time you lot've installed the Windows update you can install a PowerShell script chosen Speculation Control which will allow y'all to check if the update has been properly applied by running the command "Go-SpeculationControlSettings". With just the Windows update which addresses Meltdown this is what you lot should run into, all three requirements for the Meltdown, a.k.a. Rogue data cache load are greenlit and prepare to 'True':

For the co-operative target injection (Spectre) vulnerability only Bone back up is present, but not yet enabled as we however require the microcode update. Once the BIOS has been updated with the required version this is what you should see:

With that brief update, it's time to run some tests, starting with the Core i3-8100 results start. Note that all results are based on an average of at least three runs.

Benchmarks

First upward we have the Cadre i3 Cinebench R15 results and very footling has changed hither, from the pre-update nosotros see less than a 2% reduction in multi-thread score and 1% for the single thread test, so that'due south pretty well within the margin of mistake.

At present this is a little more interesting, the Windows patch plus BIOS update was consistently 3% slower than the previously tested configurations. Delight note lower is ameliorate for this test as we're measuring the fourth dimension it takes to complete a render. And then the BIOS update cost u.s.a. 9 seconds but as I said overall a very small reduction in performance.

Once over again the Excel workload goes unchanged, we see the aforementioned 6 second completion time, so nothing to study here.

Moving on nosotros detect much the aforementioned with the Blender render exam, all configurations took 58 seconds to complete the test.

We likewise find no existent performance deviation when testing with VeraCrypt, the AES encryption and decryption results are all much the same.

Next upwards we have 7-zip and here we see no noticeable refuse in performance with the windows and BIOS updates applied.