ARM Aims To Deliver Core i5 Like Performance At Less Than 5 Watts

[RussianNeuroMancer]

There isn't a lot that ARM can do in this regard even if they wanted to.

They can provide upstream open source Mali driver.

[cj.wijtmans]

CPUs are so good nowadays for the majority of people can run even run most of their games on 5 year old CPUs just fine. With an exception of AI heavy games.

[starshipeleven]

They can provide upstream open source Mali driver.

You would still have no hardware acceleration for media, an d no support for embedded peripherals (like wifi or bluetooth on SD interface).

That's cool but not game-changing imho.

[c117152]

I'd like to point out that until they make decently powerful processors, none will even remotely care about making "games, CAD and graphic design software" that run on it.

If it was an issue of performance we would all be using linux instead of windows by now. Fact of the matter is, the mid-high end is an Intel backwards comparability entrenched market. Even now, it's easier to use a linux dom0 and gpu pass-through windows vm to run games, CAD and graphics design software than to actually port software to linux. So, thinking ARM can just produce high-performance cores and have developers rushing over porting their existing software is purely wishful thinking. Best ARM can do is slowly chip away at the market lower-end while making sure their emulation is rock solid so by the time they get good enough clocks, they'll be able to emulate the x86 one way or another.

I don't need a toy processor for internet browsing, regardless of how "free" it might be.

My current setup is a cheap, silent and cool i3 client and an 32cores AMD compile server. Sure, not an option for low-latency applications. But that's exactly why I repeated "games, CAD and graphics design software" in my posts....

[Wilfred]

It's a 2017 chip. By 2020 it will be 3 years old and that's the current gap ARM already has with Intel's mobile chips. This is just ARM observing the trend line. The reason it's more relevant now it because Intel isn't expected to come out with better chips until Q3-Q4/19 since they're reengineering the whole x86 family to counter the speculative execution vulnerabilities. But that is left unsaid.

arm64 also has the speculative execution vulnerabilities, so ARM has to do that too.

[Apokalypz]

Marketing BS at its best. They only released this "news" to show they are dedicated to staying relevant in the near future. It helps with things like investors and entices SoC makers to buy their latest IP (or to plan to buy their future IP). If this "news" was really important, it would be about a product IP that's already out and have benchmarks to show their claims.

Any news on their roadmap (even for them) is all speculation. And we all know how speculation is these days...(pun intended)

[coder]

and ARM can't see beyond Windows laptops.

Ever heard of chromebooks?

Also, ARM is very interested in server chips.

[coder]

Are we really supposed to believe that two thirds of the power budget of the i5-7300U is "things intel hasn't figured out" and "x86 overhead"?

Intel's TDP includes their GPU. AVX2 also consumes quite a bit of power, and I don't know how that factors into Intel's TDP estimates.

[coder]

Not to mention they compare their "future 5nm" node with 10nm or 14nm processors. Because that makes sense, not.

Intel's 14 nm CPUs are a known quantity, and therefore a useful point of reference. We don't know exactly what Intel will deliver on 10 nm or beyond, so it makes less sense to speculate about that.

Yes, you're supposed to use your brain and account for the fact that Intel will certainly have something faster on the market, by then.

[coder]

I'd rather see them compete with the desktop CPUs rather then the poor performing U variants

The laptop market is stronger than desktops, and one in which ARM at least has a foothold. Also, it plays on their power-efficiency advantage.

Desktops will surely come later, as they become more established in the laptop and server markets.