WriterDudeAZ

melgross said:

But a couple of years ago Intel made a proposal that instead of using process node, which these others aren’t totally honest about, they should use transistor density. That’s a much more useful way of measurement, as it actually accounts for all the manufacturing points. But the others said no. Why, because it takes the marketing advantage of using smaller numbers, which they can do, and using bigger numbers instead, which they can’t.

I agree completely. Revolutionary chip design doesn't lend itself to linear comparisons. So for example, Apple has been very quiet about processor speeds in the M1. They've described the unified memory architecture, the video cores, the processor cores (performance and power-saving), but unless I've missed it (entirely too many things I miss in this world), they're not talking about these processors in terms of typical clock or bus speeds. I wonder if the reason for that is the idiot tech media and marketing twerps out there who are only bright enough to say "this megahertzthingy has more megahertz than that other megahertzthingy." They would pooh-pooh the processor because it's "only" clocked at 1.2GHz, or because it uses "slow" memory, or what have you. 

I think the unified memory architecture alone has improved performance to the point where they can run everything slower, save power and heat, and still rock the benchmarks. Not to mention that these chips give Apple an architecture they can use to super-charge their cloud computing data centers. 

I think the M-series chips are very impressive. I'll never buy an Intel CPU again.