Intel/Windows 30% performance reduction

[ComSimPilot 623]

3 minutes ago, vortex681 said:

Why not just try it yourself and see how it performs for you? Because of the almost infinite number of possible PC configurations, even if someone else sees no effect (or a big effect - much less likely), that doesn't mean that it will be the same for you. You only need to look through these forums to see that users with almost identical systems often get very different performance.

Yes, unfortunately I didn't do a flight before and after, so that's why I ask if someone already did that. I have the 7700K. 

[vortex681 806]

1 minute ago, Daedalus said:

Yes, unfortunately I didn't do a flight before and after, so that's why I ask if someone already did that. I have the 7700K. 

Just run it on your system and if it doesn't look any different, why worry? It's unlikely that you'd notice a few FPS either way, anyway. Because of the way that frame rates fluctuate continuously in flight sims, other than for big changes (unlikely with your CPU) it's very difficult to make any sort of accurate assessment.

[G-RFRY 1,712]

May be better to hold off on the BIOS update there are reports of random reboots for some that have updated, INTEL and motherboard manufactures are trying correct the issue.

[vortex681 806]

On 21/01/2018 at 7:12 PM, rjfry said:

May be better to hold off on the BIOS update there are reports of random reboots for some that have updated, INTEL and motherboard manufactures are trying correct the issue.

There's an informative article here about problems with the microcode: https://www.techarp.com/articles/intel-cpu-reboot-root-cause/, in particular the "What Should YOU Do? Updated!" section which advises users "Therefore, we recommend that you DO NOT apply any microcode update for your Intel system, if you are using any Intel processor manufactured since 2011."  - this was last updated on 22 Jan 18. Their link to a list of the processors potentially affected by the microcode updates (https://www.techarp.com/articles/intel-cpu-reboot-meltdown-spectre/3/) now includes many Skylake and Kaby Lake CPUs, including the i7-6700k and i7-7700k amongst many others.

I have yet to update the microcode on my system and will hold off until new, fixed microcode is produced.

[Uteman 51]

22 hours ago, vortex681 said:

I have yet to update the microcode on my system and will hold off until new, fixed microcode is produced.

Broadwell and Haswell seem to be the 2 problem lines. Intel from my reading have officially told everybody to stop deploying the microcode fix for the two series until they test out the perceived solution properly.

Detailed here: note the recommendations dated Jan 22 halfway down the page.

https://security-center.intel.com/advisory.aspx?intelid=INTEL-SA-00088&languageid=en-fr

Apologies if this has been posted elsewhere.

[vortex681 806]

6 hours ago, Uteman said:

Broadwell and Haswell seem to be the 2 problem lines.

They're certainly the two that have got the most publicity but many more CPUs have apparently been affected. My second link above doesn't seem to work any more - they've changed it to this one: https://www.techarp.com/articles/intel-cpu-reboot-root-cause/3/

Whilst Intel would appear to have found the problem for Haswell and Broadwell CPUs, they've said this:  “The progress we have made in identifying a root cause for Haswell and Broadwell will help us address issues on other platforms.” - the other platforms presumably being Skylake, Kaby Lake and so on. Some people have certainly updated systems other than Haswell and Broadwell with no problems but others have not been so lucky. I think I'll wait until they produce a verified safe fix.

[G-RFRY 1,712]

Still a hot topic Microsoft now pulling the plug on the patch and can be removed.

 http://www.tomshardware.co.uk/microsoft-disables-spectre-v2-patch,news-57805.html

[PaulGR 168]

Since I got new PC I updated every single driver two days ago and installed every possible update up to date.
I got a faster CPU now but I didn't notice anything at all.30% slow down would make my sim slower than I had before by a significant margin but at this point (8700K) my FPS went to the point where I'm flying with 70 traffic and I still land in heavy airports with 25-35 FPS (+/- 25fps - 50+ AI planes at location / up to 35fps when under 50 AI planes on location. NGX - use it for benchmarks as it's the heaviest on FPS for me). Visible improvement.

ORBX Global, Vector, NA, EU, WoAI pretty much every airline, AS16, Rex 4 Enhanced + Soft Clouds, FSFX Packages, Steve's Cloud Shadows and a bunch of freeware airports.

One thing I strongly recommend if anybody who uses WoAI is to fix the flight plans and make sure everything is as it is supposed to. Took me forever to do it but I believe I shaved a fps or two after I did it. Everything counts right? :D

Sorry for little OT.

[vortex681 806]

16 hours ago, PaulGR said:

30% slow down would make my sim slower than I had before by a significant margin

As has been pointed out a number of times, desktop systems with modern CPUs are unlikely to see hits of more than 5% (and often less than that).

[PaulGR 168]

1 hour ago, vortex681 said:

As has been pointed out a number of times, desktop systems with modern CPUs are unlikely to see hits of more than 5% (and often less than that).

Sure but according to specs my current OC shouldn't give me more than roughly 7% - 10% of boost (direct GHz to Ghz  #) yet I see significant performance upgrade.
With OC difference .3 is easy to see that if anything would affect me I wouldn't see much performance gains if any.
I need to point out that my previous setup didn't have the security patches installed.

 

[vortex681 806]

12 minutes ago, PaulGR said:

Sure but according to specs my current OC shouldn't give me more than roughly 7% - 10% of boost (direct GHz to Ghz  #) yet I see significant performance upgrade.
With OC difference .3 is easy to see that if anything would affect me I wouldn't see much performance gains if any.

I was just pointing out that the "30% performance reduction" was a worst-case scenario that only really affects servers, not desktops. Additionally, because of the variability of FPS when running flight sims, it's unlikely that you'd even notice any difference in performance after applying the patches.

[vortex681 806]

Latest update to the Spectre Microcode saga - it will be installed via Windows Update: https://www.guru3d.com/news-story/microsoft-will-distribute-spectre-microcode-via-windows-update.html

"Microsoft will be propagating Intel microcode updates to counteract the second variant of Spectre vulnerability to systems with the Windows 10 Fall Creators Update. Microsoft says it is now distributing those updates via the Microsoft Update Catalog.

Though Intel's release covered Kaby Lake, Coffee Lake, and Skylake platforms, Microsoft says its initial updates will only cover Skylake devices running the Windows 10 Fall Creators Update. The Skylake microcode update is listed as KB4090007. Microsoft says that more Intel microcode updates will be released as they're made available."

[CloudSurfer 16]

The sad part is, Intel doesn't appear to want to fix this problem in their hardware.

It's thought the next-gen of processors will have this flaw "designed in", for the simple reason that by ignoring the security isolation they can increase performance.

Who cares that hackers can steal your data, when gamers just want 10 FPS more?

Right now, Intel are even refusing to work with OS developers to add flags to later CPUs so that they can bypass the patches when they aren't required.

[vortex681 806]

41 minutes ago, CloudSurfer said:

The sad part is, Intel doesn't appear to want to fix this problem in their hardware

Intel can't fix it in the current hardware, they can only mitigate the effects.

 

[CloudSurfer 16]

No, Intel appear to not want to fix this in their hardware. I'm talking NEW processors released after today - they appear that they will keep this vulnerability as part of the design, because to properly fix it would hurt performance.

https://www.pcworld.com/article/3251171/components-processors/intels-plan-to-fix-meltdown-in-silicon-raises-more-questions-than-answers.html

https://www.bit-tech.net/news/tech/cpus/intel-details-spectre-meltdown-silicon-fixes/1/ (March 16th)

Interestingly, however, these parts will not be completely immune to exploitation: Krzanich has confirmed that Variant 1, the bounds check bypass aspect of the Spectre vulnerability, will continue to rely on software mitigation being present in the host operating system; hardware protection will only be available for Spectre Variant 2, branch target injection, and Meltdown, also known as Variant 3 rogue data cache loading.

In the same post, Krzanich confirmed that the re-released and now hopefully-less-crashy microcode mitigation patches have been released for all vulnerable Intel products from the last five years, but did not discuss when or if the company plans to release patches for older products.

.

Not cool Intel. Not cool.

Edited March 22, 2018 by CloudSurfer