Thread 106622744

x86_64 sisters, not like this...

>>106622744 (OP)
>a system so confined you can't do anything useful with it, powered by an architecture so limited you'd need an extra build per CPU manufacturer, claims it's faster than AMD's based and flexible as fuck x86 cpu which is powering the entire world full of meaningful(unlike iOS) systems.

oh ... wow....

am I supposed to clap now because some retard pleb here thinks it's more significant than my morning routine at the john ?

Zoomers were a mistake.

*throttles*

>>106622744 (OP)
The fastest hardware is useless when there is no software support.

>>106622744 (OP)
>geekbench
sure

>>106622744 (OP)
>Overheats and dies

>jeetbench
worthless

>>106622744 (OP)
lol who is this article's audience. nobody is this dumb

>>106623755
No coping, OP is just a pajeet.

>>106622744 (OP)
And yet the most intensive thing you can do with it is run "the" browser to watch porn in.

The fuck do apple people even do with these retarded devices?

>>106622799
>>106622892
not with the innovative new vapor chamber+aluminum unibody design

>>106624014
you can use it as 3D scanner and process it on the phone and it can take minutes on my 13 pro which should be much faster on the 17 pro

I make web applications that require a lot of CPU power.

I've noticed my web applications run faster on the M1 Macbook Air I use for testing than on the AMD Ryzen 7 5800HS gaming laptop I use for development.

Was always curious how it was even possible because AFAIK single thread performance should be worse on the M1.
Is there something about Apple chips that make them better for Javascript?

We could be exploring the galaxy by now if the purpose was computational advancement wasn't to brute force worse and worse code.

>>106622744 (OP)
what's single thread performance even good for nowadays? web browsers? video games?

>>106622792
>a system so confined you can't do anything useful with it,
What does this even mean? Any C programmer for instance would prefer MacOS a hundred times over Windows.

>>106622792
The Snapdragon 8 Elite G5 is even faster. And the Snapdragon X2 Elite for PCs will have the same architecture but with 2.25x as many cores as its SD8 sibling, and a significantly higher TDP. It's going to be faster than the 9950X

>>106622744 (OP)
I remember when apple fags kept saying this about the M1/M2 and it was some hyper-specific video nonsense that only beat amd if you turned off hardware acceleration because like any other high end processor they typically offload that work to the gpu.

>>106624320
I have noticed the same, the answer is the extra memory bandwidth. A lot of those js/Java frameworks go though a lot of memory so it makes sense that whatever device gives higher bandwidth, the better performance
>But how do you know if it is bandwidth or something else?
I have a 12700k I use for myself, also an M2 pro for work (provided by my employer) and a HP Z6 G4 workstation for my personal projects. (Xeon gold 6184)
For fun, I decided to build Java projects, and the order from fastest to slowest is: the Xeon, then the M2, then the 12700k. Given that the Xeon has the worse IPC, you might think it would finish last, but Alas, it finished first.

>>106624119
>>106624705
The OP is about the A series, so iOS.
That shit is locked up tighter than the average itoddler's brain functions.

>>106624804
That makes a lot of sense to me too because apart from using a lot of CPU power my JS apps also use a lot of data, even up to a gigabyte or more.

>>106622744 (OP)
even 14th intel beats 9950x3d in single core

>>106622792
>he still runs builds for x86_64 triplets
lol
lmao
all of my infrastructure is aarch64 or riscv64, has been for 2 years. if you are still using x86_64 you are BEHIND

>>106622744 (OP)
If Apple chips are such amazing performance at such low power usage, how come some datacenter company hasn't already paid Apple billions to manufacture server chips? The energy savings alone would make it worth the effort

>>106624866
Post systems

>>106624937
>Datacenter
It's coming soon (2 more weeks)
https://www.datacenterdynamics.com/en/news/apple-working-with-tsmc-to-develop-ai-data-center-chips/
Start buying $AAPL while it is still cheap.

>>106624829
It's fine for whatever a mobile is meant for.

>>106624730
What does it matter? A19 will be used for a $499 macbook, while shitdragon for a $1999.

>>106622792
If you wanna be a fucktard why don't you come here and say it to my face so i can answer you with a swift fist to the nose.

For same price, I can buy x10 x86_64 CPUs.

>>106624320
>I make web applications that require a lot of CPU power.
You are a cancer upon the world. Depart! DEPART!

>>106624320
>I make web applications that require a lot of CPU power.
So bad webpages, got it.

>>106624705
>Any C programmer for instance would prefer MacOS a hundred times over Windows.
not me

>>106622744 (OP)
> this year's chip has more single-core perf than last year's chip
Rly unusual. Should have tested against 9950x3d tho.
> geekbench
Useless.

>>106623908
cope

saar geekbest best benchmark saar

>>106624866
>I'm a contratian because NEET

>>106625861
I highly doubt you would survive

Apple SoCs are great, but their software is absolute dog shit. If someone put a gun to my head and forced me to operate a MacOS-based system I'd hesitate for a moment.

>>106622744 (OP)
I checked the numbers to be sure and it's on par with a 5.5GHz 9950X3D CCD in 1T in SPECint too. That seems good.
But considering the power difference it's amazing. The x64 retards really really need to do something different. They keep saying it isn't an architectural problem then prove it retards

>>106628292
x86_64 can't be fast by definition
you have to emulate CISC and also run a minix-based OS for security (tm) reasons and also strong memory model means every core needs to know what are other cores doing etc etc etc
it's just garbage ISA to run legacy software

>>106628267
>software is absolute dog shit.
such as?

>>106624320
>I make web applications that require a lot of CPU power
You should kill yourself.

>>106624674
Benchmark.

>>106624674
multiprocessing performance is never better than the combined single-core performance. thinking it doesn't matter is delusional.

I wonder what an experimental build for the iphone would be like if its you only used like apps written in low level langs and had minimal UIs. Could you boost battery life meaningfully?

you have already been btfo one week ago rajesh
https://desuarchive.org/g/thread/106573026/#q106574813

>>106622744 (OP)
Just imagine a world where you can install a useful operating system on apple's chips...

don't care until I have vidya and common productivity programs running on it

>>106622792
Kek. Have an upvote because the seething replies are hilarious.

>>106622744 (OP)
3nm vs 4nm

AMD will add one gAI AMX unit in the IOd at somepoint then GB score will go up bigly. I run M4 w/o SME and w/ SME and the Memebench 1T score goes up 9%.

>>106622744 (OP)
Isn't that made on like a 2nm node, and its only 11-12% faster, running arm code?

>>106628354
> you have to emulate CISC
Both x86 and ARM decode instructions to micro-ops. Neither executes directly. Also, CISC/RISC does not mean anything.

>>106630809
maybe qualcomm does it but not apple

>>106622744 (OP)
Who gives a shit? It's a phone.

>>106622744 (OP)
Nice, let's see how it runs windows+a random AAA from this year.

>>106622744 (OP)
>geekbench
Posting this should be bannable offense.

>>106632042
https://www.youtube.com/watch?v=nFsy4mb3T1w
now let's jeetdroid

>>106630524
>AMD will add one gAI AMX unit in the IOd at somepoint then GB score will go up bigly.
Na, Geekbench will then switch to a new version that reduces the scores for AMD again.