After a Windows 10 Update Today Overclocking is lost. WTF Microsoft and Intel???

[MrGenius]

No. Because the only problem that anyone had in this thread is @xkm1948 and that was because he uninstalled the update, not installed it.

Yes. Because the update caused performance loss with my 3770K. I have a problem with that. And I think it's bullshit that I had to find a workaround, instead of being given the option of whether or not to accept that performance loss as a trade for added security(that I don't particularly want or feel any particular need for). They're forcing this shit on everyone. Whether you like it or not. Which...IMO...is fucked up. I have every other Meltdown/Spectre protection(exc. L1T fault, since I haven't figured out how yet) disabled ON PURPOSE! WhyTF would I want a microcode update for it?

 

[hat]

Yes. Because the update caused performance loss with my 3770K. I have a problem with that. And I think it's bullshit that I had to find a workaround, instead of being given the option of whether or not to accept that performance loss as a trade for added security(that I don't particularly want or feel any particular need for). They're forcing this shit on everyone. Whether you like it or not. Which...IMO...is fucked up. I have every other Meltdown/Spectre protection(exc. L1T fault, since I haven't figure out how yet) disabled ON PURPOSE! WhyTF would I want a microcode update for it?

Can't wait for the day these bugs are squashed in hardware and all this nonsense goes away...

 

[lexluthermiester]

parts of America i can tell you power outages are much more common

This is likely to be correct.

Have you tried using this script?

I find much more effective disabling the Windows Update service and it's dependent service, Background Intelligent Transfer Service(BITS) and then installing updates manually as needed.

 

[R-T-B]

WhyTF would I want a microcode update for it?

Just FYI if you've disabled the mitigations the microcode should behave as stock performance, providing what you want. Those keys exist for a reason you know...

 

[lexluthermiester]

I installed KB4100347 just now on my gtx 1070 laptop. and I have 0 changes or issues in any game.
lulz. the hype is real

This statement shows clearly that your understanding of how software and hardware work together needs improvement. There are so many fine-grained complexities involved that to make such a blanket statement is astonishing.

 

[MrGenius]

Just FYI if you've disabled the mitigations the microcode should behave as stock performance, providing what you want. Those keys exist for a reason you know...

I thought the same thing. So I made sure to test the theory by disabling the Spectre protection, that was suddenly enabled, after the update. Then running benchmarks to compare against pre-update records. There was still a significant amount of performance loss. Which I proved was due to the updated microcode by replacing it with microcode that wasn't, and rerunning those benchmarks. After which the numbers fell back in line with where they'd been.

I don't claim that I even fully understand the whats/wheres/whys. But I wouldn't have bothered finding a way to undo it if I could have proved to myself I didn't need to. Fact is, I did prove I needed to. If I wanted my CPU to perform as it did before the update. Which I've also admitted was not a huge difference in performance. But it was real. It was measureable. It was definite loss(not a gain). And I did not want it. Maybe there's more to why it turned out the way it did for me. But I don't think I need to know any more than I already do. Didn't like it. Don't want it(at the moment). Got rid of it. Plain and simple.

 

[R-T-B]

I thought the same thing. So I made sure to test the theory by disabling the Spectre protection, that was suddenly enabled, after the update. Then running benchmarks to compare against pre-update records. There was still a significant amount of performance loss. Which I proved was due to the updated microcode by replacing it with microcode that wasn't, and rerunning those benchmarks. After which the numbers fell back in line with where they'd been.

I don't claim that I even fully understand the whats/wheres/whys. But I wouldn't have bothered finding a way to undo it if I could have proved to myself I didn't need to. Fact is, I did prove I needed to. If I wanted my CPU to perform as it did before the update. Which I've also admitted was not a huge difference in performance. But it was real. It was measureable. It was definite loss(not a gain). And I did not want it. Maybe there's more to why it turned out the way it did for me. But I don't think I need to know any more than I already do. Didn't like it. Don't want it(at the moment). Got rid of it. Plain and simple.

Fair enough. I just read the ingredients list, I didn't do a chemical confirmation... good on you doing field research.

 

[Fangio1951]

Have you tried using this script?

WUMT Wrapper Script 2.5.2
https://www.majorgeeks.com/files/details/wumt_wrapper_script.html

There is that nasty KB4100347. Thanks but no thanks.

Bye Bye KB4100347.

hi m8,

Appreciate your help.

But, I have a few questions.

Have downloaded the package and unzipped the contents.

Do you create a short cut to the script file - say on the desktop and then ad it as an item in the windows start up ??

Also, should I deactivate the deferred windows update in the advanced sections ??

regards

 

[unclewebb]

Do you create a short cut to the script file - say on the desktop and then ad it as an item in the windows start up ??

I just run the script when I want to check for updates. The important updates are usually only released by Microsoft once a month on "patch Tuesday" so manually checking for updates once a week or once a month is reasonable. Adding this script to your Windows startup and checking for updates every time Windows boots up sounds overkill to me.

Some of these patches break more things than they fix. I have never had a virus that left my computer in a BSOD loop. Forcing untested garbage like this onto everyone's computer is unethical.

 

[DeathtoGnomes]

***looks down at the empty popcorn bucket***

that's it You should only update on a full moon and Clear skys

you forgot to lay down salt in a circle around the PC and do the chicken dance until you pass out.

Win10 was updated? really? #tinywall

 

[xkm1948]

Another victim:

https://rog.asus.com/forum/showthre...ers-watch-out-for-Windows-10-KB4100347-update

And more
https://www.overclock.net/forum/5-intel-cpus/1601679-broadwell-e-thread-191.html

 

[cadaveca]

Another victim:

https://rog.asus.com/forum/showthre...ers-watch-out-for-Windows-10-KB4100347-update

of you under another UID.... LOL. Gonna bug ya on this one, yep. Don't take it too personal.


It's fine that you guys want to avoid spectre fixes, but you'll find you'll constantly have problems like this. Properly patching this insane vulnerability is a multi-faceted thing that requires OS, driver, BIOS and nearly everything else to be patched. It fundamentally changes how the CPU and OS interact with each other. The performance loss is real, but its also really minor, and personally, I'll take slightly less performance on an overclocked system in order to have security. It's quite possible that those with issues have these issues because they are missing other spectre fix vectors, while those that are making these patches do not, so they don't see these issues when making these patches.

I've done extensive testing in the Windows environment about performance loss, as you'll note that most people that originated the performance loss aspect were NOT in Windows.. they were in Linux. It's quite different, and with all these Windows updates in place, the impact is minimal, and in the order of 3-5%.

@xkm1948 I pulled out my X99 delixe, and did not find any problems, so I wonder where your system really lost its OC-ability. It doesn't seem to be this update... as much of a problem it might have been for you.

 

[Aquinus]

I've done extensive testing in the Windows environment about performance loss, as you'll note that most people that originated the performance loss aspect were NOT in Windows.. they were in Linux.

...and not with every application either I might add. There are some very specific cases where performance is meh but, it's actually not terrible most of the time. I/O constrained situations seem to be hit the worst by these mitigations it seems. If you're mostly testing things that hit the CPU hard alone, it's unlikely to surface as a major performance hit. Tests targetting databases like PostgreSQL or Redis are likely going to show more of a hit than testing games or compression speeds; things that aren't I/O constrained.

 

[xkm1948]

@cadaveca

I don't know I have so many alias with different systems. Must be nice to love X99 so much by buying multiple system of them



Nothing personal. I am just annoyed it hits Broadwell-E particularly somehow. And it hits me when I had least amount of energy to deal with crap like this. You know my history on X99 and another round of problem I will be flipping out.

...and not with every application either I might add. There are some very specific cases where performance is meh but, it's actually not terrible most of the time. I/O constrained situations seem to be hit the worst by these mitigations it seems. If you're mostly testing things that hit the CPU hard alone, it's unlikely to surface as a major performance hit. Tests targetting databases like PostgreSQL or Redis are likely going to show more of a hit than testing games or compression speeds; things that aren't I/O constrained.

Noticed some slow-downs when I do SQL queries for RNASeq. Not terribly slow but noticeable.

 

[Nokiron]

5960X here.

No issues, no performance loss either.

 

[yotano211]

I'm happy to say that with a i7 8850h processor
In my laptop, no issues so far.

 

[cadaveca]

Nothing personal. I am just annoyed it hits Broadwell-E particularly somehow. And it hits me when I had least amount of energy to deal with crap like this. You know my history on X99 and another round of problem I will be flipping out.

I actually think you are dealing with multiple issues and not just one. I understand that its frustrating, for sure. You haven't had a lot of luck with that platform at all. I wish it was easy to find the commonality that causing these issues, and I'm afraid this Windows update isn't exactly the true cause.

At least you got your system up, but getting that OC back would be ideal, of course. How do we do that?

 

[xkm1948]

No it just the issue of this update. Not multiple issue.

Uninstalled the update. reflashed with 3902 BIOS used @MrGenius file to get in system. I used this tool to block that update
https://www.digitalcitizen.life/how-block-unwanted-windows-driver-updates-installing-windows-10

Now system OC is back online. So control all variables, that update was the only cause for all this shit storm.

Without the update my Microcode is now back at B00002A.

This MS update has the version number here

https://support.microsoft.com/en-us...or-windows-10-version-1803-and-windows-server


The two conflicting microcode is what has been causing havoc on the Broadwell-E systems. Not sure about Haswell-E at this point as it seems to be unaffected.


Also BIOS showing current microcode version. I don't think this was changed even during the update

 

[cadaveca]

So we need to push the microcode into BIOS, and then the update wouldn't be an issue...

 

[xkm1948]

And to add to that. Even with microcode B00002A everything seems to be patched according to InSpectre. Granted it probably does not guard against some of the latest discovered vulnerabilities but for my usage it should be secure enough.

Now the question is: what was Microsoft update doing? Security patch over already patched system? Or patching the newest vulnerabilities?


https://www.intel.com/content/dam/www/public/us/en/documents/sa00115-microcode-update-guidance.pdf
Intel listed their most recent Microcode for Broadwell-E will be B00002E. This MS update was on 2C. I am on 2A. So it seems more than likely the upcoming Windows 10 Fall 2018 update will have all these incorporated

, meaning I will probably have to delete that freaking file after the update.

So we need to push the microcode into BIOS, and then the update wouldn't be an issue...

I have already called ASUS customer serivce. Big help. Their tech over the phone replied "We'll look into that"

Their ROG forum engineer PMed me the chance of ASUS supporting X99 with even more microcode updated BIOS is "very slim"

So yeah, I am gonna need to learn on how inject updated Microcode myself somehow now.

 

[R-T-B]

So yeah, I am gonna need to learn on how inject updated Microcode myself somehow now.

My ASUS knowledge isn't as up to speed as it could be, but I do believe I remember it's easier if you have that "USB flashback" feature as it doesn't check the bios for integrity the same way the software flasher does. Do you have that per chance?

 

[xkm1948]

My ASUS knowledge isn't as up to speed as it could be, but I do believe I remember it's easier if you have that "USB flashback" feature as it doesn't check the bios for integrity the same way the software flasher does. Do you have that per chance?

Yes it has. At the same time i have 0 knowledge modding a system BIOS. So yeah, fat chance of me messing something up again.

 

[R-T-B]

Yes it has. At the same time i have 0 knowledge modding a system BIOS. So yeah, fat chance of me messing something up again.

I can cover the other base if you need help. Just tell me what image you want modded.

I'm more worried that it's not a "conflict" though because such a thing shouldn't be possible (only one microcode runs at a time I think). It sounds more like Intel just screwed something up in newer revisions. But we can always try it if your game. PM me.

 

[MrGenius]

I still say the best answer is just to replace the mcupdate_GenuineIntel.dll after the update(with the file dated 4/11/18). That way the system still thinks the update(the September 13 KB4100347, not the May 17 KB4100347, there's 2 with the same name) is installed. And it won't try to reinstall it(I've tested this theory and it's true). Then you don't need to uninstall the update or do whatever to prevent it from (re)installing. Or do any BIOS modding. You're done with it. And you keep a copy of both the pre and post update mcupdate_GenuineIntel.dll files so you can manually control the microcode that the OS uses(if you decide you aren't done with it). I realize it's not the best answer in every circumstance. But it's pretty close if not.

 

[xkm1948]

I can cover the other base if you need help. Just tell me what image you want modded.

I'm more worried that it's not a "conflict" though because such a thing shouldn't be possible (only one microcode runs at a time I think). It sounds more like Intel just screwed something up in newer revisions. But we can always try it if your game. PM me.

Awesome! Thanks Froggie! I will dig around Win-RAID forum and see if I am conformable with modding.


@MrGenius I am definitely saving that file for the upcoming Fall Windows 10 update.