> he gets no shekels from either intel or amd > yet he

>>59077726
Cucky Mcfucky

>>59077726
To explain the fiasco a bit further, GPUs are NOT CPUs and vice versa. That means, CPUs will NEVER become technology that can get away with shit single-thread performance. This is because while GPUs work by definition in parallel - and therefore the NVIDIA/AMD cartel can just make bigger/more expensive/easier chips and get away with it - CPUs work by definition very lightly in parallel, and largely in a serial manner because simply that's how software logic works, it's a fact of mathematics and life in general, so you will always need serial performance.
The only way to make a CPU get away without serial performance, is if you are 100% offline, 100% noninteractive, like for instance offline rendering of something, or offline encoding, are just pure video recording or something else simple like that.
When you have interactive applications like a game, even if you make the task manager to show "100%" on many cores, you will still need serial performance, because interactive applications still need a global loop and that global loop can not be hogged by 150,000 mutex locks every single procedure.

>>59077789
>Answering to a thread saying to not be a cuck for either of them
>he calls that cucking
I guess that's what cucks say to themselves. "I'd be a cuck if I'm not a cuck".
Come to think of it, it's exactly how real cucks work in real life: They are scared their wife cucks them, they become jealous as fuck, they become appalling as a result, their wife hates, and therefore cucks them.
It comes in parallel with what I said in the OP. If you shill for AMD when they pull their dirty tricks you effective make Intel better. You have to expose the hypocrity. Intel isn't better because they are "evil", they are better because they are better capitalists and they managed to have their own Foundries. AMD still has no Foundries anymore and the access it has to them is expensive or lower quality.
It's why the NVIDIA/AMD cartel still works on GPUs. They hog the patents of GPUs but neither NVIDIA have their own foundries. If Samsung or Intel or another foundry had access to making full blown GPUs, both NVIDIA and AMD would be dead withing 1 to 2 years.

>>59077919
Cucky Mcfucky

>>59077833
>CPUs work by definition very lightly in parallel, and largely in a serial manner because simply that's how software logic works, it's a fact of mathematics and life in general, so you will always need serial performance.
Nah
Vast majority of serial code is only serial because it can.
If for some reason current CPU technology suddenly disappeared and took serial software with it, all those "mandatory" serial segments would shrink into lightweight fences.

Though programming would instantly start requiring 10 IQ more to comprehend and most pajeets would lose their jobs.

>>59077987
Keep bumping my thread with your ignorance. Suits me, and AMD's progress. PS. sage doesn't work fully anymore either so keep bumping.

>>59078017
You obviously have never tried to code an interactive application. They are not x265, encoding, you must have a global loop for at least a large part of it, and it must have mutex locks.
Even when people sperg about their "100%" task-manager cpu usage, there is nothing telling them how efficient that "100%" is being done.
I'm pretty sure the majority of games nowdays only harness about max 50% of what they would do on a purely non-interactive application like encoding.
Single-thread WILL NEVER be not needed on interactive software, it's simply NOT possible on conventional silicon computing at least.

>>59078068
>you must have a global loop for at least a large part of it
Unless you learn how to use events and asynchronicity. Then you don't.