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

[DeathtoGnomes]

If users don't like being on the bleeding edge .,,. would suggest not being the 1st one on your block to install new updates. Ask Woody has probably the biggest collection of "warnings" ... . When discussing preview rollups, I always respond "What advantage is there to installing this today ? Whether the chance is 10, 25, 50, 100% of having a problem why take the chance ? Wait 30 days and see if you are affected, by what and if it's been fixed when the next month rolls around."

There a bunch of ways to delay updates in windows, the sad part is m$ doesnt always respect the users choices and updates anyways. Using 3rd party apps are the preferred method of preventing windows from phoning home to check for said update.

 

[CloudXPS]

Just wanted to say thanks to all the fellow nerds in here. Just deleting the microcode dll in System32 will get things smooth again. Regained control of my FSB and overclocking from the BIOS. X99 with a 6800k. Very cool, I had noticed it being weird after every update.

 

[1stFalloutboy]

Hi there, realize that this is a month old but needed to ask has anyone attempted to patch the microcode on the CPU on the Rampage V Edition 10, my flatty has done it for the Rampage V and his older intel CPU 5960X I think he's running, I have the 6900K and noticed that 1903 bjorked overclocking so deleted the microcode update from microsoft which fixed the overclock issue but not the security flaw. I would be interested to know if applying the microcode fix direct from intel would fix the Spectre problem and still allow overclocking.

 

[lexluthermiester]

Hi there, realize that this is a month old but needed to ask has anyone attempted to patch the microcode on the CPU on the Rampage V Edition 10, my flatty has done it for the Rampage V and his older intel CPU 5960X I think he's running, I have the 6900K and noticed that 1903 bjorked overclocking so deleted the microcode update from microsoft which fixed the overclock issue but not the security flaw. I would be interested to know if applying the microcode fix direct from intel would fix the Spectre problem and still allow overclocking.

No, but if you are gaming as a primary usage, you can disable hyperthreading to nullify the vulnerability. However, you shouldn't really be worried. These vulnerabilities are very difficult to take advantage of in a real-world attack.

 

[1stFalloutboy]

I agree with the fact that they are very difficult to take advantage of in a real world attack but I also feel it is the duty of the manufacturers in this case ASUS and Intel to fix things correctly without negatively impacting the user experience which was something that was quoted directly by Microsoft themselves as being "virus activity". When it comes down to it my board is still under warranty and I expect ASUS to patch the BIOS which they are responsible for doing.

 

[R-T-B]

Never liked the idea of UEFI and how Microsoft\Intel could just change shit. just to much control over some thing you payed good money for.

external MC loading has been around since BIOS (since the Pentium Pro, to be precise). UEFI has nothing to do with it.

 

[lexluthermiester]

I agree with the fact that they are very difficult to take advantage of in a real world attack but I also feel it is the duty of the manufacturers in this case ASUS and Intel to fix things correctly without negatively impacting the user experience which was something that was quoted directly by Microsoft themselves as being "virus activity". When it comes down to it my board is still under warranty and I expect ASUS to patch the BIOS which they are responsible for doing.

While that is true, none of these vulnerabilities were intentional and we the public are expected to protect ourselves. If you have a DSL/Cable/Fiber modem that has a router built into it, enable the firewall and NAP. Those two settings alone will keep 99.997% of remote attacks out, including what would be needed to take advantage of the vulnerabilities in question. If you don't have a modem with a builtin router and you don't have one already, get one and enable the settings described above. Next, get into your Services manager and disable anything to do with "Remote Desktop". The next is simple, DON'T download anything that you don't know something about.

 

[1stFalloutboy]

While that is true, none of these vulnerabilities were intentional and we the public are expected to protect ourselves. If you have a DSL/Cable/Fiber modem that has a router built into it, enable the firewall and NAP. Those two settings alone will keep 99.997% of remote attacks out, including what would be needed to take advantage of the vulnerabilities in question. If you don't have a modem with a builtin router and you don't have one already, get one and enable the settings described above. Next, get into your Services manager and disable anything to do with "Remote Desktop". The next is simple, DON'T download anything that you don't know something about.

Been into computers since I was six years old and am TCP/IP network engineer, been in the industry 43 years and know about security. The point I am trying to make though is that if you buy a car and it has a serious fault that say could let someone take control of the steering and control it remotely and note I said COULD the problem would be fixed the same thing should apply to people expensive computers as a matter of principle, my Car wasn't cheap but my computer cost more. There is a moral obligation on behalf of the manufacturers to fix vulnerabilities in their products it shouldn't be up to end users no matter how talented they are and manufacturers should not be allowed to get away with half assed fixes which break usability, especially when that usability was part of the reason for purchasing the products in the first place.

 

[R-T-B]

The thing you are missing is that short of a recall, there really can't be a "better fix." These vulnerabilities exist in the silicon, and can only be worked around at expense of perfomance, not truly "patched."

 

[1stFalloutboy]

The thing you are missing is that short of a recall, there really can't be a "better fix." These vulnerabilities exist in the silicon, and can only be worked around at expense of perfomance, not truly "patched."

If that is indeed the case how is it my flatmate managed to patch his CPU microcode with a file supplied by Intel for Linux patches on his 5960X and have it both still overclock and Spectre protected? The way I see it my CPU is just a later refresh gen of his so the board supplier should be able to do this too and are just choosing not to and letting microsoft and intel provide the bjorked fix that kills the overclock.. That's just lazy, irresponsible, unreasonable and not beneficial to maintaining your customer base.

 

[lexluthermiester]

Been into computers since I was six years old and am TCP/IP network engineer, been in the industry 43 years and know about security.

Then you shouldn't need help from us. I offered you confirmed working solutions. Take it or leave it, I'm out.

 

[R-T-B]

If that is indeed the case how is it my flatmate managed to patch his CPU microcode with a file supplied by Intel for Linux patches on his 5960X and have it both still overclock and Spectre protected?

Because I did not say anything about overclocking, only general performance loss. The microcode windows uses is the issue regarding overclocking. It's an older, earlier fix. Updating your bios mcode may help but is hardly simple.

 

[1stFalloutboy]

Because I did not say anything about overclocking, only general performance loss. The microcode windows uses is the issue regarding overclocking. It's an older, earlier fix. Updating your bios mcode may help but is hardly simple.

Fully aware of that and fully aware of your ability to bjork your board completely which is why the manufacturer should be fixing the problem not the end user.

Then you shouldn't need help from us. I offered you confirmed working solutions. Take it or leave it, I'm out.

No need to be rude, simply stating the facts and I know despite what microsoft says about security on it's systems that the only way to have a system fully secure is to never connect it to anything, USB, internet or network but these days that sort of likely not to happen. Confirmed working solutions however is not would should be coming down the pipeline - when the board is still under warranty the manufacturer has an obligation under law to fix the problem. In this case they are simply ignoring it and that is unacceptable.

 

[rkt]

Finally went to Ryzen

 

[FragManSaul]

instead of changing any Windows 10 setting I downloaded a new bios for the Rampage V edition and this sorted it, overclocking back to normal. EZ update is poo

｜｜ASUS USA

www.asus.com

 

[basco]

i know its an old thread but the win10 microcode update KB4497165 breaks my memory overclock on x99 with 5960x.
took me a month to find why its not working. thx ms.

ucode 0x40 was last good for me\newer one is ucode 0x43 easy to see in throttlestop under the FIVR tab

 

[ThrashZone]

Hi,
Which bios are you using ?
I'm still on old 2101 released 1-6-2016 lol for x99 sabertooth and didn't notice anything at 3200c14 being killed
This system was torn down couple weeks ago for a z490 build though so no way to test anymore until I get another gpu.

 

[basco]

i am on 3801 since 2017- last bios before spectre fix

 

[ThrashZone]

i am on 3801 since 2017- last bios before spectre fix

Hi,
I'd try 2101 and see what happens
MS installs micro codes so no advantage of newer bios on haswell-e if anything broadwell-e bios exaggerates vccio to 1.25v+ which is way more than it needs which is 1.05v it's default.

 

[basco]

no auto volts for me - never.

 

[ThrashZone]

Hi,
Good policy
Soon as memory frequency is raised from 2133 default broadwell-e bios shoots vccio through the roof
Haswell-e bios never did that always hair above 1.05v even at 3200c14 xmp.
Think most say safe limit on vccio on haswell-e chips is 1.15v
Lots toasted as well as some broardwell-e chips on oc.net reported all were on auto vccio and xmp.

 

[Mussels]

Bump for relevancy!

Was fixing a family members PC (full of dust), which was my old Z77 system. 3770k @ 4.5GHz, 16GB DDR3 R2400 C11, RX580 (My god that hardware aged so well)
Except... it was running 3.5Ghz, 1333. Every BIOS setting, even fan speeds, was totally ignored - like it was reading from another, second set of settings. CMOS clear, reset to defaults - no change.


Remembered this thread and checked the similar posts elsewhere, and found one suggestion "Intel management engine corrupted by windows update, reflash the BIOS"

Found one mystery BETA update from 2019, flash it... and bam. All settings working again.
These spectre and meltdown "Fixes" are still breaking things in 2021

 

[Mussels]

Ran into this yet again, this time on intel 4th gen

flashed the bios, installed 11, everything was peachy - then the new owner got the PC and it failed to boot
Had to short out pins on the bios chip (1 and 6) to get the backup bios to boot, which then had the exact same thing happen

Reflashed the BIOS *again* and now it's all good - i worry that another OS install in the future will trigger the exact same bullshit again, it's like the updates to the IME just corrupt the BIOS on these older boards

 

[AsRock]

Ran into this yet again, this time on intel 4th gen

flashed the bios, installed 11, everything was peachy - then the new owner got the PC and it failed to boot
Had to short out pins on the bios chip (1 and 6) to get the backup bios to boot, which then had the exact same thing happen

Reflashed the BIOS *again* and now it's all good - i worry that another OS install in the future will trigger the exact same bullshit again, it's like the updates to the IME just corrupt the BIOS on these older boards

Was only until a few months ago i moved from a 25k and a 3770k and was not having these issue's, how ever i never put windows 11 on either of them.

 

[mkdr]

https://www.reddit.com/r/intel/comments/10dcjd9