Can we have different Windows versions tested with and without Meltdown/Spectre patches applied?

[birdie]

@W1zzard

Microsoft has just announced the performance impact of its patches which fix Meltdown/Spectre vulnerabilities: they say that pre-Haswell systems running Windows 7/8.1 are the most affected.

So, it would be extremely interesting to see the usual battery of tests comparing the performance of Windows 7/8.1/10 with and without the security patches applied on systems running, say, Coffee Lake, Haswell and Sandy Bridge. Personally, I'd test some Bloomfield/Lynnfield CPUs as well for good measure.

Thank you!

 

[R-T-B]

My brother is running a Nehalem (well, technically Westmere) system. If you happen to have one of those beasts somewhere, I'd be curious about that as well (provided you have the time).

 

[Flaky]

Are we talking about the OS updates only, or firmware updates (microcode) also?
I haven't tested it yet, but have read that the updated haswell microcode causes general stability issues.

 

[P4-630]

Updated my Windows 8.1 rig yesterday, Asus AI Suite 3 doesn't open anymore for some reason but it does apply my fan profiles on startup.

 

[IceScreamer]

If I ever manage the time (gonna be tough, the finals are starting soon), I'll try to test my Sandy i3, on Win7.

 

[Deleted member 163934]

The tests should be with the windows unpatched, patched and patched + microcode update (this one done by bios update because at least atm the vmware microcode update doesn't make windows enable part of the protection, see #292 ; as I already said in another topic there will be an army of mb that will not see an bios update just to update the microcode because the manufacturer won't bother to do it).

 

[birdie]

The tests should be with the windows unpatched, patched and patched + microcode update (this one done by bios update because at least atm the vmware microcode update doesn't make windows enable part of the protection, see #292 ; as I already said in another topic there will be an army of mb that will not see an bios update just to update the microcode because the manufacturer won't bother to do it).

I'm pretty sure BIOS updates won't be available for old (>3 years old) systems, so old CPUs can only be tested with Windows patches applied.

 

[Flaky]

You can integrate the updated microcode yourself.

 

[londiste]

This is a bit of a strange problem. Meltdown patch is out and should be complete. Spectre patches is what includes BIOS/microcode updates and Microsoft seems to have couple more things for this in the pipeline on software side.

Microsoft has published this on topic of performance impact:
Understanding the performance impact of Spectre and Meltdown mitigations on Windows systems (Microsoft Secure Blog)

Here is the summary of what we have found so far:
- With Windows 10 on newer silicon (2016-era PCs with Skylake, Kabylake or newer CPU), benchmarks show single-digit slowdowns, but we don’t expect most users to notice a change because these percentages are reflected in milliseconds.
- With Windows 10 on older silicon (2015-era PCs with Haswell or older CPU), some benchmarks show more significant slowdowns, and we expect that some users will notice a decrease in system performance.
- With Windows 8 and Windows 7 on older silicon (2015-era PCs with Haswell or older CPU), we expect most users to notice a decrease in system performance.
- Windows Server on any silicon, especially in any IO-intensive application, shows a more significant performance impact when you enable the mitigations to isolate untrusted code within a Windows Server instance. This is why you want to be careful to evaluate the risk of untrusted code for each Windows Server instance, and balance the security versus performance tradeoff for your environment.

 

[R0H1T]

This is a bit of a strange problem. Meltdown patch is out and should be complete. Spectre patches is what includes BIOS/microcode updates and Microsoft seems to have couple more things for this in the pipeline on software side.

Microsoft has published this on topic of performance impact:
Understanding the performance impact of Spectre and Meltdown mitigations on Windows systems (Microsoft Secure Blog)

So it's worse on Haswell or older systems, great now where do I throw my z97 board or do I get a (partial?) refund from Intel for this loss in performance

 

[Deleted member 163934]

You can integrate the updated microcode yourself.

Mind to share how is possible to do it for an UEFI bios?
I know it's possible for not UEFI bioses.

 

[R-T-B]

Mind to share how is possible to do it for an UEFI bios?
I know it's possible for not UEFI bioses.

Look up Universal Bios Updater.

 

[Mussels]

A lot of people misunderstand what these fixes actually do and the ways they'll effect home users - a reputable tech website showing before and after results would help clear up confusion and FUD, even if the results are very boring.

 

[CAPSLOCKSTUCK]

I have Westmere and Bloomfield PCs on today and the PC i am actually using is a Sandy G540. I can run any tests you want provided i am told exactly what to do...like a set of rules that all testers can follow.

 

[hat]

Like I said in another thread, I believe the issue is overblown, unless you are managing a big server farm that actually makes use of virtual machines. Then you can worry about performance loss, otherwise your rig will still run crysis just fine.

 

[Vayra86]

A lot of people misunderstand what these fixes actually do and the ways they'll effect home users - a reputable tech website showing before and after results would help clear up confusion and FUD, even if the results are very boring.

Absolutely, but I would like to add: timing is key here. I think doing it now will not show the full impact.

Give it a few more weeks at least. Spectre is still in full swing. Or: focus on a smaller Meltdown-only article.