I said “compares favorably”
I’m gonna give you the benefit of the doubt since I’m bad at communicating, and explain further:
“Compare favorably” meaning that the M1’s performance, for a chip meant for passive cooling and long battery life, is trading blows with x86 cpus which consume more power.
Now, do I expect the M1 to outperform the best of the best x86 desktop chips in every metric? No, I’m not delusional. However if you don’t define a cpu which compares in performance to it’s competition (often above its weight), while drawing much less power, as “unimpressive”, then I’m not sure what you’re expecting.
Now, with that cleared up, the chess benchmark, shows the M1 performing over 10x worse than other cpus which it compares in performance in other metrics. That alone should tell you that something is off.
It’s been stated, over and over, to you specifically, that the M1’s performance is known, and it’s definitely not 10x slower than any x86 cpu. If it were, nothing would run on it, and certainly not to the degree it does.
This is because of specific avx optimizations (as stated by the intelligent people itt), that exists on the x86 code. The numa optimizations, which would be the apple silicon equivalent, are not in use.
The “generic x86 code” (quote, you), in fact has more optimizations than the “specialized arm code”, (quote, you). And as such is not a true comparison.
If you want to argue that that is fair game, then by all means. But, then you must admit that the benchmark code is favorable to x86.
Now, to put all of this in context, the point on benchmarks in the first place id to give potential buyers an idea of the performance of the system they are buying.
Of the M1 users across the web, I have never once seen these machines described as “slow”. Which if your benchmarks were accurate, people would notice these machines being painfully slow.
