hmmm - solution is to disable real time Defender...
No one said to disable Windows Defender Real-time protection. It is the Windows Defender Real-Time Protection notification feature that is causing the high CPU usage and reduced performance. The Local Group Policy Editor (gpedit.exe) shows this explanation.
This policy setting turns off real-time protection prompts for known malware detection.
Turning off the prompts and turning off real-time protection are two different things.
download an app by author
I wrote a free app called Counter Control so users can test for this problem as well as fix it. No use in releasing an app that only reports a problem if it cannot also fix it. Counter Control includes a TechPowerUp logo so anyone that sees this free app in a forum will know where to go to safely download it. It is a "proof of concept" type app to help users understand this issue.
My 10850K shows a repeatable drop of 1000 points in Cinebench R23 when Windows Defender sets the fixed counters to mode 2 and starts running this portion of its algorithm. No one is trying to scare anyone. Those are just the facts.
It does not seem to be all of Windows Defender that is causing this. Other users with completely different processors have reported similar results. A 6% drop in Cinebench performance seems to be typical. This decrease in performance has been reported in a wide variety of Core i CPUs from the 7th Gen through the 11th Gen. Both desktop and mobile CPUs are affected by this issue.
I must not be very good at self promotion. I forgot to include both my name and email address within the program itself. Adding an About box is on the things to do list but realistically, that will likely never happen. Too busy for that.
BTW, I do not need any self promotion. I am already quite well known in the user community. My popular free utilities have been downloaded more than 16 million times from TechPowerUp alone. These programs are used by millions around the world every day for the last 15 years. Counter Control was written to inform and make user's Intel based computers faster if that is what they choose to do. Counter Control was not written to make a quick buck.
It may not affect all users
That is correct. User testing has shown that Intel's 12th Gen CPUs do not have this problem. Intel's 12th Gen CPUs have the exact same fixed function and programmable counters as all of the previous Intel Core i CPUs have. For some unknown reason, the part of the Windows Defender algorithm that is being run on the 7th to 11th Gen CPUs does not get started on the 12th Gen CPUs. None of the counters are being started so this part of the Windows Defender algorithm must not be starting.
Can't reproduce the 1000+ point difference noted by unclewebb.
A person will only see this sort of difference if Counter Control reports that the Windows counters are initially set to 0x222. You have a 12th Gen CPU so this problem does not affect your computer. Your counters are reporting 0x000 which means they are not being used. If you do not have a problem that needs to be solved, running Counter Control is not going to change anything. Cinebench performance will be exactly the same.
Cant say i've noticed anything
That is not surprising. When you have six or more cores like most people do along with hyper threading, an inefficient antivirus program is not going to be noticed by most users during normal use. This issue has been around for years because very few people ever noticed anything out of the ordinary. Just recently, I finally found a way to reproduce this issue and I came up with a simple way to fix it. That is why I wrote Counter Control. Most users appreciate some extra performance, especially when it does not cost anything.
Now that this problem is a little more out in the open, maybe Microsoft will notice this. They can decide if there is any room for improvement in the efficiency of their Windows Defender algorithm. Most minor issues like this are likely to be quickly forgotten before they are ever fixed.
