Good progress.

I'm still going through XCOM 2, a bit over 82 hours in. I'm at what looks to be the last mission and I think I could have finished it yesterday... if the game hadn't crashed. And I don't know how long had it been since my last save.

I hope I'll finish it today so I can move on.

And here I thought it was gonna be a slow boring Monday.

A couple weeks ago I wrote that MLiD reported that Intel were to ditch hyperthreading, and got met with much [x] Doubt due to the source.

Well, now hardware times also comes with this, and explains how rentable units are apparently actually better than hyperthreading to fill up the pipeline of a given CPU:

https://www.hardwaretimes.com/intel-15th-gen-cpus-to-get-rentable-units-why-hyper-threading-is-going-away/

If this is true, then no wonder Intel were to ditch hyperthreading in favor of rentable units (also: worst name ever, as it sounds very much like unlockable cores against paying Intel tons of cash when it's something different entirely). The drawback of RUs are a need for big and close caches, so raising the sizes of L2 caches seems very much like working into that direction, as is the explosion of core numbers with Intel's e-core inflation.

I mean, if you follow the source thread, it all goes back to MLID.

Well, it also links to an Intel patent about the subject.

Another thing is if something like this will happen sooner rather than later.

Sure, but Intel files 6,000+ patents every year. I'm certain you can find one to apply to every narrative.

It might even be something they are considering, but between that and coming up in an actual product is the remaining 99%. If it succeeds is another story entirely, as Intel's recent string of failed products attests to.

Meh, I still think it's a big doubt in the consumer space. Intel, AMD and Nvidia all patent stuff all the time that goes unused. Nvidia has had tons of research papers when it comes to chiplet designs for their GPUs going all the way back to pre-Turing era. Yet we are in the Lovelace generation with still no chiplet in sight.

Maybe it could happen in the distant future but who knows. The reason why everyone dislikes MLID is because of how many things he consistently gets wrong. He throws shit against the wall until something sticks which is why most if not everyone will continue to doubt him. Remember, he said N33 would = 6900XT performance when in reality, N33 is no where near that. He also deletes posts and etc when he gets stuff wrong.

"Very high confidence" lmao

I'm not saying this patent will come to fruition, almost all patents end being nothing (I remember Nintendo filling a patent for holographic storage many years ago), but at least this debate doesn't come from a leak with no source.

Could you give us some summary of it? I'm not really keen to watch an over 40 minutes "rant" video, if that's part of what this is.