Is the future ARM, or are these Benchmarks bullshit?

Is the future ARM, or are these Benchmarks bullshit?

Other urls found in this thread:

geekbench.com/doc/geekbench4-cpu-workloads.pdf
twitter.com/NSFWRedditImage

ofc its the future

Intel dragged their feet for too long and apart from Ryzen nothing exciting is happening on x86

>Geekbench

Geekbench is absolutely retarded and does not reflect performance, whether overall mix or single task, in any way shape or form.

My 1600x shits over it anyway 16000+

single thread is the same who cares its outdated asf

An iPhone X has half the muli and 2/3's the single of an i7 7700k according to Geekbench..

Is there a way to accurately benchmark ARM vs. x86?

Is your 1600x 1/4th the size of your thumbnail, passively cooled, and driven by a phone battery?

Geekbench is trash. Truth is arm can do what 2010 CPUs could do.

Probably only SPECint or SPECfp.

Not very good, but better than the Geekbench trash.

a chip that has an ISA that only works on IOS and geekbench has be optimised for IOS
brings out the best result vs fragmented OS

ofc its not the future this is bending reality you simply cant compare them

>ARM @ 24 MHz

Seems legit.

can your applel run actualy games or applications?

ios is dogshit

A11 beats Ryzen 1200, impressive

applel has no games tho

And the A11 is only running at 24MHz

why is there so much retardation in these threads

RM is the future because Apple has more money than anyone for R&D.

This chip was designed 3 years ago.

Arm is garbage and far from the future.

>People keep explaining why Geekbench is fucking trash and not directly comparable
>Keep posting it anyways
Almost as if someone was paying for these posts!

2010 had some great cpu's like the i7 920.

its 10nm too, imagine the 5nm ones in 2020

its from q4 08 actually

A11 has 4.1b transistors
I7-7700 has 1.75b

Is that due to the GPU in the A11?

Now post the cinebench r15 results.

It's a system on a chip.

2 + 4 ARM cores
3 GPU cores
motion coprocessor
neural network processor

Here is the A10 (125 mm2) of which each big core takes only 4mm2.

comparison of the A10 big core to other mobile cores.

Can't wait to see diagrams of the A11.

>neural network
apple fags just eat this shit up, don't they?

>Is the future ARM
No.
>are these Benchmarks bullshit?
Yes.

No, it literally has small ML chip, think of Movidius ones.

RISC is gonna change everything

You mean RISC-V?
Then no, it's so autistically RISC it's unsuitable for any high-performance designs.

So wait, there actually ARE alternative CPU designs to x86?

What the hell has taken them so long?

How the fuck will everyone on x86 adapt if nothing x86 will work on arm or risc?

No one will bother to port anything to ARM though.
That's why AMD shelved K12 you know.

I hope you dont screw like you type

Bullshit. Iirc, geekbench overvalues sha and aplel cpus have hardware sha. They arw much slower than x86 in real eorld performance.

You can't compare different architectures with GeekBench.

Shouldn't the question be why Intlel doesn't have hardware support for the most used algorithm in the entire world.

It's because the ARM is a SoC.
Even the last Atoms have more transistors than an Intel I7-7700 just because it's a SoC and not just a CPU.

>KikeBench

It's all thanks to Acorn. Long live the ARM - the Acorn RISC Machine

Except you can. All Geekbench versions have the same workload.

>no one will bother
except linaro who specializes in ARM compilers
except arch
except debian
except google for android
except google for chromebooks
except arduino
except raspberry pi
except windows for pi / shitty rt implementations

Except TI, Samsung, Atmel, Apple, Freescale, Rockchip, and pretty much every other major hardware manufacturer.

Intel runs windows gaymin though, gotta love that proprietary philosophy locking software into platforms to keep said platforms alive. Otherwise ps3, xbox, and pc would have died years ago in favor of centralized linux gayming on all devices. This is why you cant have nice things.

No one cares until big enterprise shit like SAP and Oracle DB gets ported. From then on it's smooth sailing.

>iTard believes that a smartphone cpu is better than a desktop/laptop cpu because of geekbench
That's why you buy iShit and I don't

A11 calls A11 bionic neural engine, until moment maybe begins GPU making 600GFlops in 16bits.

All workload almost only exist in cache memory

geekbench.com/doc/geekbench4-cpu-workloads.pdf

They're big as fuck they weren't joking for dekstop classes performance, give it a time we will see it soon on macbook maybe.

The workloads don't represent real world performance.

The absolute state of Sup Forums

>inb4 iphone x > i7 7700k

>are these Benchmarks bullshit?
If you compare ARM to ARM. No they're not
If you compare ARM to x86. Yes

Doing the geekbench my 1700x gets a LZMA score of :

28602
44.7 MB/sec
(geekbench compresses and decompresses a 450KB html ebook)


in 7zip I get
31MB/s compress
500MB/s decompress

Either the implementation is crap, or it's heavily weighted in favor of compressing (Which makes no sense as you will be decompressing much more often) in a arbitrary way (not based on the time each step takes)

Yes Ryzen excels at decompression. This is a fundamental trait of the architecture.

They're bullshit. geekbench is a horrible benchmark even for a synthetic one especially across different CPU ISAa.

is this bait

I know. that should reflect in the the geekbench as well.