You're still reading news articles from the 90's. You need to scroll ahead in your feed and catch up. It might make you feel better to think this way in the short term, but one underestimates and denigrates their competition at their own peril.
Maybe...
It remains a fact that every CPU I know of out of China is very disappointing. They're all foreign ISA (eg x86 or ARM or MIPS or RISC-V) and the performance is what, in the US, you'd consider 3rd tier. I also don't think I've ever seen any interesting micro-architectural ideas in the literature.
And certainly the shenanigans with ARM China don't help in dispelling the claim that they've stolen whatever they wanted...
I wouldn't want to make any claims about whether this is structural, or bad incentives, or poor college education (and everyone who is any good at CPU design goes to the US and then works at a US or Taiwanese company).
I'm simply pointing out that this is what we have seen so far.
===================
OK, I've had a chance to skim Twitter (the ONLY useful source of truth when it comes to anything technical) and this is basically a big nothingburger. The CPU and GPU are both ARM.
Huawei IP is the ISP and NPU (which it already was in prior chips).
In the case of Apple, their version of this step was the A4, with clear plans (though we didn't know them at the time) for the A5, A6, and then A7. And there was sophisticated Apple IP on the chip (eg, the NoC and memory controller, and the strategy for abstracting out IO) though again we did not know it at the time.
But in this case I think even that is assuming too much. This is just this year's Huawei SoC, like last year's Huawei SoC and the year before that.
Journalists bothering to publish Huawei's "Rivals Apple" hype are being played! Pure and simple. There is nothing interesting and significant here.
Noteworthy elements include
- the area devoted to the NPU is, eyeballing it, about 1.5x the area devoted to "two P cores". For A18 Pro these areas are about the same. This COULD represent the NPU being less efficient. Or it could represent the sensible decision (one Apple will likely follow) that NPU is important enough, and the field is fluid enough, that might as well give it some extra area just for future-roofing
- ISP appears massive, like as much area as NPU. Again unclear if this represents low efficiency or Huawei knowing their market, and knowing that their users really really care about computational photography.
Of course these comments assume the chip annotation is correct, which may be optimistic...
I'm not sure the extent to which Geekerwan and friends are doing this by genuine knowledge as opposed to vibes...
