2017 Kaby Lake i7

>2017 Kaby Lake i7
>only 25% faster than 2011 Sandy Bridge i7

Did we hit a wall on single core IPC performance?

Is it going to be like this forever?

Or, will it get even worse (less than 25% speed-up in the next 6 years)?

Who knows.
They have been focusing on reducing power consumption and heat before charging forward with more speed.

In the past they could bump the clock speed up....

Now they actually have to make improvements to the arch

Mobile computing is the future

yes, and to make you buy new hardware they make non-free OSes and software that magically starts to work slower after some time forcing you to buy new hardware
the next part will be replacing all software with javascript shit like atom, maybe we'll even see a whole OS written in javascript

They are going to bake new arch. We've been using core arch for like 10 years now.

>mobile SoCs are consistently increasing by 30% or more every year when Intel can't even make that kind of jump in 7
How long until laptops and Windows tablets start using ARM chips?

mobile SoCs also didn't start with a huge amount of performance like x86 is at right now.

Sure at high levels like desktop parts it'll be forever till ARM can compete but at low power consumption stuff like the core m and core u chips ARM is really getting close.

Kaby Lake is over 50% faster than Sandy shit

lol no

Honestly that hasn't gotten much better either.
Especially considering it's 14nm vs 32nm what is the perf/watt difference between sandybridge and skylake? it's gotta be somehting around 30%

Probably not long, microsoft has been showing off how it can emulate windows 10 on those snapdragon 820s because qualcomm finally got off their ass and stopped using those atrocious cortex cores.

...

...

i dont know anything about ARM

what makes them so special?

what are they doing thats different?

There's nothing really special about them, They're just well built for low power draws.
Intel's cores are good at ~10w-130w but any lower or higher and perf/watt gets to the level that it's more efficient to build a new architecture, and that's why ARM exists.
Also I don't think intel's core Ms consume as little power as intel claims, intel's rated TDPs seem to be more of a random guess than actual measurements on a good day.

Haha you fucking whish

Well compared to a 2600k the ~25% IPC gains of skylake + the stock OCs of kaby lake could reasonably get 50% performance improvements in highly multithreaded workloads like video rendering.
It's not a totally ridiculous claim, it's just you could always OC the sandy bridge.

>4.5
>4.2
>7% difference in clock up
>net result shows 7% difference and 8% difference
kek
Literally 1% IPC on multicore and 0% IPC gain at single thread

bump