Holy shit, Qualcomn is finished, Samsung is bankrupt.

What does s7 edge using exynos get?

iPhone already destroys samsung in real world performance. A10 is going to btfo anything else.

Even the OP3 that is half the price destroys samshit.

File: aww.png (214KB, 1760x1586px) Image search: [Google]

214KB, 1760x1586px

>>56457872
not sure if this is legit though.

Remember Apples CPU is a dual core while Samsung's is octa core.

>>56457999
moar cores
moar pixels
moar lag

File: Screenshot_20160906-141834.png (199KB, 1440x2560px) Image search: [Google]

199KB, 1440x2560px

>>56457872
Beats it in multi-core
>>56457999
Old version of Geekbench

>>56457735
I see they increased the SHA performance.

>>56457999
Remember that the octacore CPUs don't use all eight cores, under load they use the four big ones and when relatively idle they use the four small ones.

>>56458084
>scaling results to be more favorable to apple
I thought this was supposed to be a benchmark.

Reminder that most task are single threaded, especially internet browsing.

Thats why Apple invest its time and money in strong single core performance, since thats what normies need.

>>56458598
>use perfmon
>load up web browser and a website
>all my (four) cores ramp up to top speed and high degrees of utilization

>>56457735
>iphone9

are you from the future?

>>56458894
El psy congroo

It's John Titor

>>56458894
yes

>>56458209
Why don't they just use a powerful quad that can down clock during low loads / idle and boost clock during peak loads

>>56460552
Because big.LITTLE works better.
Each cores have entirely different voltage ranges. Big cores cannot operate at a low enough power level to match a little core.

>>56457735
>Samsung is bankrupt.
Samsung manufactures CPUs for Apple though.

>>56457735
> tfw this is a 4W cpu
intel btfo

>>56460682
Only the transistors (which is like 10% of the cpu)

also, this new chip is apparently 100% Samsung free.

>>56459704
Go back to /a/ with your cancer

>>56460772
Nope, the Xeon D 1541 is still the king of energy efficiency right now. Also the A10 is not a 5W chip, it's a 10W one.

>>56457735
>caring about smartphone performance
they're literally facebook machines. Serious people use UMPCs for their /g/ needs

>>56460920
There are rumours intel may use Xeon D cores in phone SoCs soon. We may start seeing legit UMPCs soon.

>>56460805
>CPU
>only 10% transistors

Get out of here and never come back

>>56460971
There is no such thing as a Xeon D core.
The Xeon D line uses Core i arch exactly like all of their mainstream chips, Core M, the i7E line, and the rest of their Xeons.

Intel canned all plans for smart phone Atom chips because they couldn't gain any market share, and they were bleeding billions of dollars into it. They aren't going to try again with an even larger more expensive approach.

File: 1362960884713.png (28KB, 499x322px) Image search: [Google]

28KB, 499x322px

>using your phone for anything besides phone calls and light internet browsing

You idiots probably fell for that retarded phone VR meme, too.

>>56460971
it's a faggot phone who cares

>>56461057
Yeah god forbid people use their pocket computers for any computing tasks

>>56461080
Phones are supposed to replace desktops and laptops for the majority, grandpa.

>>56461160
>majority
aka peasants

>>56461022
>There is no such thing as a Xeon D core.
Yeah, meant to say the Broadwell-DE core that those SoCs use.

>The Xeon D line uses Core i arch exactly like all of their mainstream chips, Core M, the i7E line, and the rest of their Xeons.
But this specific core architecture is low power as fuck.

>Intel canned all plans for smart phone Atom chips because they couldn't gain any market share, and they were bleeding billions of dollars into it.
True

>They aren't going to try again with an even larger more expensive approach.
It could work if they made dual-booting phones that ran Android and x86 windows 10. They could make UMPCs great again.

>>56461190
>we all need 24-core Xeon processors to watch cat videos and do work
k

File: 1471759307621.jpg (259KB, 2048x1152px) Image search: [Google]

259KB, 2048x1152px

>using the whole cuckbench 3 score instead of just the FP performance
every time

>>56460878
>it's a 10W one.
[citation required]

>>56460852
>>>/reddit/

>qualcomm is finished
who will build iphones then

>>56462276
foxconn, you fucking retard

>>56462164
>Its a 5w one
[citation needed]

Holy shit some vague numbers! <X> is finished and bankrupt!

>>56460878
>it's a 10W one.
I'd really like to see them try dissipating 10W in a cell phone.

File: krakenpowersm.png (20KB, 650x495px) Image search: [Google]

20KB, 650x495px

>>56462164
He doesn't have one, because its wrong. Apple doesn't release TDP figures because they aren't selling chips to anyone, they don't need to publicize that figure.
Anandtech tore down an iPad Air and put some probes on it to measure power consumption though. In a power virus they got the A7 to pull 9w peak. Under typical heavy load the chip pulls 3.5w~.
The A8X showed 16% improved power draw, evidenced by longer battery life with a smaller battery. The successive SoCs used in their mobile devices all followed the same trends, reducing nominal power draw.

If you rate a chip by how much it pulls in a power virus workload then your average i5 or i7 is nearly double its TDP. No one considers these fringe cases to be the norm.

http://www.anandtech.com/show/7460/apple-ipad-air-review/3

Pic related is total power for the whole device, including screen. SoC draw is seen by the delta from idle to load. They note that an intensive GPU bench brought platform power to 6w, in game power draw was 5w, so the chip itself was pulling 2-3w.

>>56458193
Geekbench 4 doesn't test SHA1 anymore.
All those Flagship Android owners are butthurt as fuck since Samshit and Mediatek employed SHA1 cheating methods.
The Redmi Note 3 Pro with the 600 series Snapdragon scores very closely to the 820 in Geekbench 4.

File: ayylmao_by_google.jpg (63KB, 680x472px) Image search: [Google]

63KB, 680x472px

>>56462446
>probes
ayy lmao

File: tmp_8271-Screenshot_2016-09-06-23-48-02-2025844343.png (156KB, 1200x1920px) Image search: [Google]

156KB, 1200x1920px

>>56457735
>Nvidia K1
>destroy everything mobile in gaming
>scores are shit
What gives?
Is this what happens when Nvidia can't use GimpWorks?

File: lion_kek.png (216KB, 393x391px) Image search: [Google]

216KB, 393x391px

>>56462797
> Nvidia

>>56462797
Maybe your one is throttling. The K1 gets an average of 1000 single core and 2700 multi core.

>>56462797
Tegra K1 is either two of Nvidia's custom ARMv8 Denver cores, or four Cortex A15 cores.
Cortex A15 is old, ARMv7 arch, and not anywhere near what one would consider high performance. They wouldn't have used these at all, but their Denver cores pull an ungodly amount of power.

All Nvidia had going for them was their IGP arch being more fleshed out and having better drivers than the competition.

>>56462837
>Maybe your one is throttling.

Jesus christ

>>56457735
literally: make better cores(apple) vs. make moar cores (mediakek, qualcomm, samsung)

File: amd.jpg (10KB, 288x175px) Image search: [Google]

10KB, 288x175px

>>56463647

>>56458598
>internet browsing.
>single threaded
b8
Any respectable web browser has to do HTML/CSS parsing, JS execution, rendering and compositing. Do you still think browsers only do one thing at a time?

File: tmp_12046-Screenshot_2016-09-07-01-22-32-2025844343.png (188KB, 1200x1920px) Image search: [Google]

188KB, 1200x1920px

>>56463123

>>56462512
>SHA1 cheating methods.
since when offloading cryptography and hashing functions to a dedicated crypto hardware is cheating?

>>56463703
How useful really are fixed-function SHA1 units in a mobile SoC?

Makes sense for the server though.

>>56458598
You are retarded.

>>56463739
Wireless encryption, IpSec VPN throughput, maybe even TLS/SSL though sha1 is depreciated

Having an aes module in your cpu for IpSec makes a massive difference, for example my netflix piracy VPN box I have can accelerate Ipsec to fast enough throughput to wirelessly stream HD

>>56463703
Because SHA1 is deprecated and is being phased out by everyone. Google will penalise site search rankings for sites that continue using SHA1 encryption past Jan 2017.
It is a useless benchmark.

>>56458026
Or: Java.

>>56461160
Obviously you cab sell more units of a product that costs $100 and barely lasts a year compared to a $1000 that has greater utility and lasts longer.

>>56463703
Because nobody gives a fuck about hashing 1 GB per second vs 14 GB per second on a phone where the SSL handshake alone results in a high latency roundtrip.

>>56460609
not that anon but ty for that

File: Безымянный.png (442KB, 3200x1080px) Image search: [Google]

442KB, 3200x1080px

>>56457735
Geekbench is fake test for apple fags.

File: Безымянный2.png (1MB, 3200x1080px) Image search: [Google]

1MB, 3200x1080px

>>56465887
Can u imagine that phone soc can beat desktop cpu?

>>56460609
>because big.LITTLE works better

Lul. Apple have demonstrated with their supreme architecture that focusing exclusively on large, powerful cores and race-to-idle is more effective than dedicated power-saving cores.

>>56460805
only 10%? Just what is in the CPU, Doritos?

>>56458209
Not true anymore, some snapdragons can use all now

>>56465904
Just cuz mobile browsers strip out a lot of shit

>>56466040
> browsers
I thought we talkin about syntethic test.

File: Geekbench 4_20160907_151421.png (183KB, 1200x1920px) Image search: [Google]

183KB, 1200x1920px

>>56462797
>>56463695
Just upgrade to Marshmallow and dont use fake tests.

>>56465931
So that's why all the iPhone owners I know are racing to charge their phones by lunch time?

>>56465904
But it can't. Real world performance tests have shown that the iPad Pro doesn't match an i5.

Geekbench is useless.

>>56466359
> Geekbench is useless.
/thread.

>>56466338
>>56465931
kek. both of these are actually really good points though.