Does it fit 32K? Does it have some weird aliasing issue because you caused cache extinction with too many power of two sizes? And if you don't know the answer to these just check L1d hitrate with perf.
The original UNIX literally swapped processes, as in write all their memory to disk and read another program's state from disk to memory, it could only run as many processes as many times the swap was bigger than core, this is a wholly unacceptable design nowadays.
But the point is, why would you trust it for anything at all, when it can't do an incredibly simple thing reliably at all? (Yes, I understand the tokenizer makes this hard, but still, it's a quick demonstration that it's just bad technology.)
But fiber optics are different, they stay good for a very long while and you just need to add better transceivers at the ends to upgrade them, it's questionable whether the current GPUs are even worth powering up once 2nm ones are out, and we also don't know how long they last, electromigration might start killing them in 5 years.
The AVX2 version was still 64x faster than the C one, so AVX-512 is just 50% improvement over that. Hand vectorized assembly is very much the key to the gains.
That was very different, here the problem is the game's threads get scheduled on the weak E-cores and it doesn't like that for some reasons, with the PS3 that would have been impossible, SPEs had a different ISA, and didn't even have direct access to memory, the problem was developers had to write very different code for the SPEs to unlock most of the performance of the Cell.
But history showed exactly the opposite, if you don't have an already existing software ecosystem you are dead, the transistors for implementing x86 peculiarities are very much worth it if people in the market want x86.