Expensive phones can’t get much faster but that’s no reason to stop buying

Phone speeds have hit a brick wall

But that’s okay

Nobody is advertising chipset speeds anymore. Just a few years back, when a manufacturer announced a new flagship phone, the superlatives “50% faster processor” or “twice faster graphics subsystem” were pouring from the stage.

The iPhone 12 still promised outrageous performance gains | Image credit – Apple

All the geeky hardware specs were that much more exciting then, because the generational processor manufacturing node jumps brought significant power gains. Even Apple, which on paper only focuses on what you can do with your phone, rather than what’s under the hood, was showing colorful charts to compare its newest A-series chipset capabilities. Well, usually with something a few generations back, so the gains look more impressive, but still…

Performance gains promises were more modest for the iPhone 14 | Image credit – Apple

Nowadays, the talk of the town are auxiliary chips, artificial intelligence calculations, connectivity gains, or new graphics subsystem features like ray tracing. The talk about clock frequencies has all but vanished, along with benchmark bragging rights, as the scores have hardly budged in three chipset generations.

Why? Well, phones are already fast enough as they are. In fact, if it wasn’t for the AI craze, there is barely anything software developers can throw at them to make them screech to a halt, unless it is a badly written code. 

We are already in the realm of 3nm chipsets and the clock frequencies will be hitting 4GHz peak counts, so don’t expect much more speed from your flagship phone going forward.

Processing power is overrated

Quality before quantity

Such an increase in processing speed is pretty unremarkable, but has been going on ever since we got down from the 7nm high horse. Even that potential 15% gain in benchmark scores will never materialize, though, as manufacturers are usually balancing performance with power draw gains, and settle for something in between.

The 4nm/5nm nodes were around for a while, and mobile chipsets on the 2nm/3nm process will stay for even longer. That’s simply because performance gains will be nowhere to be found, barring some radical discovery that resurrects Moore’s Law from the deathbed it will be lying on this decade. After all, that which is left to work with are some measly 1.4nm, and we’ll literally need a quantum leap to make our phones run much faster than they already do.

The situation is similar to the cellular connectivity modem craze that initially brought ever-growing speeds, but the 5G chip in our phones has now been stuck on 10Gbps download maximums for a good while. Qualcomm is adding more bands, filters, or aggregation goodness instead, since the carrier networks can’t even take advantage of these modems as they are.

The same goes for the whole mobile chipset industry, as we can expect more AI, graphics, connectivity, or image processing subsystem features added, rather than drastic leaps in clock speeds to simply make phones faster for the sake of it.