So now that x86 is approaching ARM's energy efficiency is there any reason to hang onto ARM anymore...

So now that x86 is approaching ARM's energy efficiency is there any reason to hang onto ARM anymore? I'd really really like to buy a phone that could do more than play angry birds and do social media. Can't exactly fit a steam deck in your pocket.

Expertbook = 3.99 watts/hour for web browsing (66Wh/16.5 hours)
M1 Air = 3.2 watts/hour for web browsing (49.9Wh/15.5 hours)
X1 Carbon = 5.12 watts/hour for web browsing (51Wh/10 hours)

Attached: Untitled23.png (759x285, 19.8K)

any "X per Y for web browsing" metric is pure bullshit
x86 will never reach performance per watt in idle and low to medium load scenarios, by design

Explain why the steam deck can achieve 4+ hours of battery life (ie less than 10W total system power consumption) playing full blown PC games (ie not angry birds).

Attached: VpUjF3nqNZfWYm5MEkJ73f-970-80.jpg (970x600, 45.36K)

Battery life and power draw in general is an irrelevant metric. Raw instruction throughput, and especially integer and floating point arithmetic performance, is what matters.
ARM will never be relevant and amd64 is just okay. The majority of computers are in data centers with effectively unlimited input power and fast I/O anyway, the ones not in server racks belong in landfills.

>4+ hours of battery life
>playing full blown PC games
>most games listed don't even manage 4 hours even when run at low graphics settings
>the ones that go over 4 hours are some indy game and some 2D platformer (that has also been ported to ARM) that both have extremely low system requirements

Attached: my smug is augmented.png (1080x717, 1.6M)

The point is ARM operates under 10W in "low to medium load scenarios". x86 can now do that as well.

Web browsing involves a lot of mostly idle while you read pages, and still x86 can beat ARM at it. Nobody just leaves their laptop running doing nothing, they close it and put it to sleep.

I take one 1500W CPU please and make sure not to cheap out on the cooler. Don't worry about running 2V through the core just give me 8Ghz and enlarge the pipeline to push 4096bit vectors please.

>The point is ARM operates under 10W
Sure, if you just set an arbitrary limit to stop measuring at at 10W, and don't measure any battery life gains beyond that point. The Steam Deck runs at 6.4W when playing that 2D platformer I mentioned, which has also been ported to the Nintendo Switch. For comparison, the Nintendo Switch runs at between 2.45W and 6.38W depending on the game for the original model that used an older processor made on a 20nm process (3 process node shrinks behind the 7nm processor the Steam Deck used), or at between 1.77W and 3.54W for the newer model with an updated processor made on a 16nm process (which is still 2 process node shrinks behind the one used in the Steam Deck).

Not arbitrary in the slightest, anything under 10W = passive cooling, which is where ARM has been curb stomping 90% of x86 (bar dogshit celery/atom). The fact that an x86 CPU that isn't a weak pile of dogshit can perform " low to medium load scenarios" at under 10W is a huge monumental leap and a threat to ARM.

Where are my 10Ghz CPUs anyway? It seems like most CPUs are topping out around 5Ghz without overclocking. Single-thread performance helps every compute task in the long run.

>30fps
unplayable

actually, the human eye can only perceive up to 6fps

By god i hope you're baiting.

intel had a presentation about this a while back, noting that if the jiggahurtz race went on, eventually CPUs would exceed the temperature of the sun
do you have enough power and cooling to handle a sun in your house?

>eventually CPUs would exceed the temperature of the sun
And then, despite giving up on the niggahurtz wars, they did anyway. Funny how that happens. Don't even get me started on GPU TDPs either.

Attached: 1617489193830.jpg (299x450, 18.07K)

>noting that if the jiggahurtz race went on, eventually CPUs would exceed the temperature of the sun
Bullshit.
Just make the die bigger, and your heat dissipation can scale with package size. CPUs shouldn't be the same size they were in 2003, if it means that physical/chemical limitations limit capacity.

Because it barely manages to achieve 2 hours, playing any half-decent PC game of the last 10 years (aka no PS2-tier JRPGs or platformer shit). Also let's not talk about how it sounds like a jet engine and you literally cannot touch the fan outlet without burning your fingers off
>t. steam deck owner

>32% bugger battery
>only 6% more battery life
Truly incredible. ARM shills on suicide watch.

>just make the die bigger
>how to reveal you don't understand anything about circuits and photolithography
Make the die bigger how, retard?
The CPU does exactly what it's designed to do, and that design has a footprint.
You can't make it bigger without adding features, and you can't make it smaller without removing features

You're essentially asking them to stay on the same process node, which will result in zero performance gains

Increase the spacing between the transistors dumbfuck. Same design, same components, spaced further apart so that upping the clock rate doesn't light the casing on fire.

Are you fucking retarded?
The spacing between the transistors is EXACTLY what makes die shrinks faster
Have you ever heard of "the speed of light"?

Good performing x86 laptops have been gnawing on 10+ watts even when doing light web browsing which meant the jet engines came on even when browsing a spetsnaz emotional support forum. Going down to 4 watts while maintaining good performance (ie not celery/atom trash) for x86 is a pretty huge accomplishment. We used to rely on ARM for that kind of efficiency but now we don't have to which is a pretty huge deal since ARM is absolute dogshit for running x86 software and muh gaymes (ie make up 90% of all useful software on the planet) and there's very little hope much of that is going to get ported to ARM because x86 emulators exist.