Discussion - New MediaTek chip could solve the silicon crisis

If you're going to spend between 600 and 2800 dollars for a new smartphone, would you pay an $84 higher price to be sure you got the most powerful SoC possible? Would saving $84 by going with an inferior substitute truly be a bargain?

First, where do you get your $84 price delta? Second, what SOC is inferior in what way? Do you have access to comparative real world performance numbers or are you just making assumptions? This insane instance on higher and higher SOC performance numbers without any study/analysis of how much performance is enough to provide a devices required performance and what becomes pure overkill resulting in unessacery power consumption and heat generation is just dumb. These SOC manufacturers (especially Qualcomm) have convinced everyone that "there is no such thing as too much horsepower " which is total and utter nonsense.

So to answer your question in a more appropriate way, yes I would by a flagship with a non Qualcomm SOC as long as said SOC provides a smooth and satisfying user experience.

Where from? From the byline, "Qualcomm's next flagship SoC is rumored to start at $300, while MediaTek's alternative costs 28% less."

And yeah heat dissipation, or lack thereof, is the current bottleneck for flagship SoCs. In terms of which is better than which though, their Geekbench and Wildlife Extreme tests showcase that pretty well. Dimensity doesn't make SoCs that are as competent as Qualcomm.

You are making the samse mistakes/assuntuons most laypeople do.

If you pay $84 more for a component, when you price the product you have to apply your sales margins to the product. Example common 50% sales margin ( very common in the mobile industry) on a $84 component results in an additionsl sales price of $168.

Benchmarks do not measure real world performance but only maximum isotropic performance. On a mobile SOC, operations like texting, streaming AV, browsing, voice communications and most AI functions almost always require significantly less functional power that the max performance figures published in benchmarks. You can buy a car with an engine capable of a top speed of 200 mph but if you only drive it in downtown traffic you will never come close to using the cars speed capability and worse will waste gas in the process. Qualcomm counts on you buying into the marketing coolaid created buy these benchmarks without even questioning their meaningful application. My local internet provider keeps running ads about upgrading to their 2 Gbs fiber by pushing the obvious improved bandwidth. Many people buy into this hype not understanding that a standard 4k stream only requires an average of 15 tk 25 maps and given how the internet functions even the most demanding households would never even use more than 300 mbs of bandwidth. In the electronics Industry we call this specmanship. Great sales tool, often meaningless in real world applications.

Dimensity's best SoCs have only one advantage versus Snapdragon SoCs: superior ray tracing performance. That's the only measure by which they are superior.

And yeah 28% of $300 doesn't include the markup but it does represent the difference in price for the SoC to its end-users, the phone manufacturers.

I look at the lowest FPS observed in a 20 loop Wildlife Extreme test because that tells you what is going to happen to performance after the phone heats up and starts throttling.

I appreciate that you have a different opinion than me, and you are entitled to think what you want.

I think ypu hit the nail on the head. I look at things through a specific set of technical lenses that causes me to arrive at different conclusions but that does not invalidate your viewpoint. I value only imperical testing using actual real world applications running in an user mode and value benchmarks very little. From my viewpoint there is a point of diminishing returns for frame rates and computational metrics at which if you blind tested a user (gave them identical looking hardware but with very different performance level internals) they would not be able to pinpoint which was the high performance SOC vs one of lesser specs/performance. These are not the tests that are run, but need to be run in my opinion, to make sure we are not trying to crack an ege with a sledgehammer so to speak. I do however very much respect your opinion to dissagree.