but why

MLID is a bullshit mill that spits out a bunch of self-contradictory guesses based on absolutely no evidence, in the hopes one of them will be close to correct.

A platform running it can remove them. Same way platforms running GPT-based software can set their own limitations currently.

No sane platform will remove them entirely, unless their goal is demonstrating what a complete circus of misinformation and schizophrenic ranting results from doing so.

Remember early public chatGPT’s totally insane behavior? No company wants to use something that will potentially say literally anything to a customer.

Those things have literally nothing to do with each other. An open source system or model still needs configurable restrictions to stop it going off the rails in basically any real-world use.

You need process lasso for some programs that windows is baffled by. Most things will work without it, but a good chunk of programs will use the wrong cores, or stuff will be parked for no good reason.

Everyone cried about how this cant be a windows problem and AMD needs to fix it 👀

You would need a 120Hz+ display to get any vague benefit out of it to begin with. It is not very impressive compared to FSR/DLSS framegen.

I think this is the worst example i have seen so far, most people with messy configs only manage 100w.

Do patchnotes count as statements? Because this has been addressed repeatedly and fixed for most users.

The problem is that it is a complex issue that is mostly due to badly configured monitors causing unnecessarily aggressive refresh cycles and high memory clock, and is being fixed for each individual monitor, one by one.

There is a reason it was basically ignored for years (it goes way back and impacts all GPUs, RDNA3 is nothing new aside from having higher memory power draw). Nobody wanted to open this can of worms.

Also worth looking into how much power your actual monitors use if you are worried about idle power use, since this mostly impacts people with 3+ mismatched monitors.

It should not impact it meaningfully. Sounds like this may finally be a years late fix (everyone else did already/never had) for fTPM stutter.