Apple A14X Bionic Rumored To Match Intel Core i9-9880H

[Vya Domus]

Given the rate of improvement, even if A14X is even half as powerful as 9880H they will catch up within a few generations.

In a few generations Intel (or AMD) will have new processors as well, the 9880H is still basically an ancient Skylake CPU. Also the rate of improvement has hard limits, there are so many execution units and in-flight instructions you can add before it makes no difference in the real world.

 

[Fourstaff]

In a few generations Intel (or AMD) will have new processors as well, the 9880H is still basically an ancient Skylake CPU. Also the rate of improvement has hard limits, there are so many execution units and in-flight instructions you can add before it makes no difference in the real world.

That's true, but if they are at Skylake level of performance it will be "good enough" for most people. According to Steam hardware survey, most people are still at 6 cores or less: https://store.steampowered.com/hwsurvey/cpus/. Hardly any with cutting edge 8C or better processors.

 

[Frick]

Might not be the best descriptor, although with the increasing dependance on cloud... hmmm

By that logic any internet connected machine that has client software is a terminal.

 

[Vayra86]

By that logic any internet connected machine that has client software is a terminal.

That would be correct. Consider a Chromebook...

With the push for cloud, we are fast going for global mainframes. Anyway. Grossly offtopic I guess

 

[Punkenjoy]

The first 80% performance are easily obtainable when making a CPU, it's the last 20% that get harder. And the closer you get to 100% the more work you have to put.

It's also always funny that everybody future CPU is beating 1-2 years old CPU. But in the end, they will fight different architecture.

And Apple have a lot of silicon dedicated to many accelerator and since they live in a closed environement where they control everything, they can easily make use of them. That is actually a good strategy but it come with downside.

The truth is it will be hard to really get a real idea of performance between Apple CPU and the rest of the market. It might end up in a fight between a closed and controlled platform where everything can be set the way apple want and an open environment where everyone is free to do what they want.

 

[king of swag187]

No, it wont. God I hate geekbench, literately the only benchmark that you see or hear about whenever there is something Apple related.

Finally, someone who realizes how flawed it is. Also, really? @author of the article, using some random "tech" youtuber who has no idea what he's talking about for a news piece? Wow this site has gone down in quality recently

Yes, Apple, as long as you have your slow as molasses IOS to pair with your great CPUs, it all looks very smooth and fast. Meanwhile, latency is pretty high on all your devices. Its a nice hiding trick but in raw performance, its not all that special as many think it is.

I'll take x86 and some real responsiveness, ty

ARM Is a joke

device	latency (ms)	year
ipad pro 10.5" pencil	30	2017
ipad pro 10.5"	70	2017
iphone 4s	70	2011
iphone 6s	70	2015
iphone 3gs	70	2009
iphone x	80	2017
iphone 8	80	2017
iphone 7	80	2016
iphone 6	80	2014
gameboy color	80	1998
iphone 5	90	2012
blackberry q10	100	2013
huawei honor 8	110	2016
google pixel 2 xl	110	2017
galaxy s7	120	2016
galaxy note 3	120	2016
moto x	120	2013
nexus 5x	120	2015
oneplus 3t	130	2016
blackberry key one	130	2017
moto e (2g)	140	2015
moto g4 play	140	2017
moto g4 plus	140	2016
google pixel	140	2016
samsung galaxy avant	150	2014
asus zenfone3 max	150	2016
sony xperia z5 compact	150	2015
htc one m4	160	2013
galaxy s4 mini	170	2013
lg k4	180	2016

realistically what the hell is this even supposed to mean

 

[dragontamer5788]

Lets actually talk Geekbench for a sec. I know Geekbench3 was highly flawed, but why does everyone think that Geekbench4 is bad?

Here's Geekbench4's workload: https://www.geekbench.com/doc/geekbench4-cpu-workloads.pdf

Now I recognize that a lot of Geekbench4's benchmarks fit inside of L1 cache, but that's more of a testament to how big L1 caches have gotten. (128kB on the iPhone). Lets be frank: if 128kB L1 cache is what's needed for the modern consumer, then we should be blaming AMD / Intel for failing to grow their L1 to 128kB (AMD / Intel still have 32kB L1 data caches).

Lets really look at Geekbench4's benchmarks. Unlike Geekbench3, AES is downgraded to be just another test instead of its own category. (And mind you, AMD Zen2 and Intel Xeons have doubled their AES pipelines recently: AES remains an important workload). There's JPEG compression (emulating a camera), HTML5 parse, LUA scripting, SQLite database, and PDF rendering. Lots of good workloads here. Very similar to a wide variety of workloads of the modern, average consumer. Even an LLVM compile (3900 lines of code).

There's a bunch of "synthetics" too: 450kB LZMA compression, Djikestra, Canny (Computer-vision), a 300x300 Raytracer, etc. etc. A bunch of tiny synthetics.

--------------

Geekbench4 is what it is: a small test for testing L1 cache and Turbos of modern processors. Its probably closer to the average phone-user or even desktop-user's workflow than SPEC, LINPACK, or HCPG.

But yes, the iPhone crushes Geekbench. Because the iPhone has 128kB L1 cache. But is that a legitimate reason to call the test inaccurate? We can't just hate a test because we disagree with the results. You should instead attack the fundamental setup of the test, and tell us why its inaccurate.

Its pretty insane that the iPhone has a 128kB L1 cache per core. Yeah, that's its secret to crushing Geekbench4 and its pretty obvious. But Intel Skylake's L2 cache is only 256kB and AMD Zen2's L2 is 512kB. Having such a large L1 cache is a testament to the A12 design (larger caches are usually slower. Having such a large cache as L1 must have been difficult to make).

 

[Searing]

View attachment 165917

If you run all the CPU & GPU cores at Max frequency the chip can consume up to 20W, but of course apple doesn't allow that to happen and limits the maximum power available.

You say 20W yet you post a picture that clearly says 4.61W for SPEC integer and 5.04W for SPEC floating point. So no, it isn't 20W. And the chart also shows the 2.6ghz A13 beating the i9-9900k in integer performance, so imagine the A14.

Sometimes I feel like I'm a developer and have a basic grasp of graph reading, and I'm arguing with people who "know things" and will even post proof opposite what they are saying.

Total system power is 3W to 6W during gaming.

Imagine believing that a single digit W SoC will outperform a 45W Intel chip with 4.8 Ghz single core turbo.

Apple fanboys are something else.

Educate yourself. Limited TDP is where Intel does very badly right now, that's why AMD is ahead with Ryzen 4000. Run your 9900k with 2 cores and 5W and watch it squirm. We are not fanboys, we've been watching Intel do absolutely nothing for years, no improvements in manufacturing node, and no changes to core or GPU design. Hopefully Tiger Lake keeps Intel in the race.

 

[M2B]

You say 20W yet you post a picture that clearly says 4.61W for SPEC integer and 5.04W for SPEC floating point. So no, it isn't 20W.

That 5W figure is for A SINGLE LIGHTNING CORE at max frequency, the phone itself won't surpass the power/thermal limits of course and if you run a proper multicore workload it will throttle down. Seems like You are not that good at graph reading Mr. Developer.

 

[TheoneandonlyMrK]

Imagine believing that a single digit W SoC will outperform a 45W Intel chip with 4.8 Ghz single core turbo.

Apple fanboys are something else.

Dreamy lot aren't they , it's like apples cinebench , that geek bench ,except worse.

At least cinebench can be and is loop run so 30second turbo mode's don't cheat the figures, total ball's comparison software.

 

[Searing]

That 5W figure is for A SINGLE LIGHTNING CORE at max frequency, the phone itself won't surpass the power/thermal limits of course and if you run a proper multicore workload it will throttle down. Seems like You are not that good at graph reading Mr. Developer.

That's not how it works. Again, developer here. You don't go to double the power consumption using both cores, there's no complete power gating here. That's why I showed you the 6W max for everything in the GPU test.

The iPhone doesn't turn off one core and run at half power when you do the test. In singe core performance it takes less than 6W and beats the 9900k in integer performance already. In multicore it throttles down slightly and runs two cores at about the same power. And anyways we are talking about the 5nm EUV A14, that one will beat Intel easily. Give it higher clock speeds ala laptop or desktop form factor and it will beat Intel in FP also most likely.

Don't quote Anandtech and then ignore where they say Apple is faster in integer than the 9900k already and that was a year ago. I have a Epyc 24 core server, a 10900 development machine and a Macbook Pro and Ryzen 4000 latop in the house, and iPad Pro and iPhone (work pays for stuff). I have no problems with performance. The question is why do you believe it isn't fast? Anandtech, every benchmark, every program, and tons of youtube videos are out there, go have fun.

 

[TheoneandonlyMrK]

That's not how it works. Again, developer here. You don't go to double the power consumption using both cores. That's why I showed you the 6W max for everything in the GPU test.

The iPhone doesn't turn off one core and run at half power when you do the test. In singe core performance it takes less than 6W and beats the 9900k in integer performance already. In multicore it throttles down slightly and runs two cores at about the same power. And anyways we are talking about the 5nm EUV A14, that one will beat Intel easily. Give it higher clock speeds ala laptop or desktop form factor and it will beat Intel in FP also most likely.

Simply impossible, the node they use(5Nm) limits any possibility of higher clocks, you think they are not already pushing it via burst algorithms, and apples chips would get wrecked on a sustained workload against that 9900K, never mind a more modern x86 core.

if its a philips head use a philips screwy
flat head use a old shool screwy


just surfin or lightweight tasks use arm .
do actual work or run simulations 24/7 etc use x86.

simple.

 

[Searing]

Simply impossible, the node they use(5Nm) limits any possibility of higher clocks, you think they are not already pushing it via burst algorithms, and apples chips would get wrecked on a sustained workload against that 9900K, never mind a more modern x86 core.

We are talking about the chips, not the form factor. Stick them in an iPad the sustained performance is higher. Sustained means nothing, you just stick the A14X in a laptop form factor and it would be sustained. They are only at 6W and 2.6ghz and you think they can't get higher. Ok.... *backs away slowly*

ARM is an ISA it has nothing to do with how fast the CPU can be.

ARM-based Japanese supercomputer is now the fastest in the world

Fugaku is being used in COVID-19 research

www.theverge.com

 

[Vya Domus]

Educate yourself.

No, you should be the one educating yourself on what's feasible and what isn't.

Sustained means nothing

Come on. Sustained means nothing, right, the one thing that you know Apple's chips are horrible at in terms of scalability means nothing. Got it.

Geekbench4 is what it is: a small test for testing L1 cache and Turbos of modern processors.
...
Its pretty insane that the iPhone has a 128kB L1 cache per core. Yeah, that's its secret to crushing Geekbench4 and its pretty obvious.

It's also pretty obvious why it's a horrible benchmark precisely because of that. We both know those patterns have little to do with the real world, Samsung tried to do optimize their cores for Geekbench as well and indeed they are second to Apple except their chips perform worse in real world tasks than vanilla ARM designs that get half the score Samsung's cores do. Is that still not enough to prove something is terribly wrong with this benchmark ?

Having such a large L1 cache is a testament to the A12 design (larger caches are usually slower. Having such a large cache as L1 must have been difficult to make).

Anyone can put large caches, there is nothing amazing about that. In fact, it's a pretty poor strategy especially in a mobile chip, caches don't just get slower when they become larger they also use a lot of power as well. Probably one of the reasons why their chips always had horrendous multi-threaded scalability, having one core turbo up with such a wide design is fine, when you have 2 or 4 or more you inevitably need to drop the frequencies into the ground. That's fine for a phone, it fits the typical usage pattern but on a desktop not so much.

Mr. Developer.

Cut him some slack, he said he's a web developer, this stuff is not exactly within his area of expertise.

 

[Searing]

No, you should be the one educating yourself on what's feasible and what isn't.






It's also pretty obvious why it's a horrible benchmark preciously because of that. We both know those patterns have little to do with the real world, Samsung tried to do optimize their cores for Geekbench as well and indeed they are second to Apple except their chips perform worse in real world tasks than vanilla ARM designs that get half the score Samsung's cores do. Is that still not enough to prove something is terribly wrong with this benchmark ?



Anyone can put large caches, there is nothing amazing about that. In fact, it's a pretty poor strategy especially in a mobile chip, caches don't just get slower when they become larger they also use a lot of power as well. Probably one of the reasons why their chips always had horrendous multi-threaded scalability, having one core turbo up with such a wide design is fine, when you have 2 or 4 or more you inevitably need to drop the frequencies into the ground. That's fine for a phone, it fits the typical usage pattern but on a desktop not so much.



Cut him some slack, he said he's a web developer, this stuff is not exactly within his area of expertise.

Every forum is full of ignorant people like you just ignoring every benchmark (it is Geekbench 5 now, and there are many other benchmarks you can use), ignoring every expert Anandtech included, ignoring actual real world results. World's fastest computer is ARM based? Ignore it. Amazon offering ARM server instances? Ignore it. This is why the world passes some people by. They just can't accept that something has changed. There is an interesting question about psychology here, why does ARM being fast bother you? Why do you not accept basic reality? ARM is just an ISA, 68000 was fast, PowerPC was fast, x86 was fast, ARM was fast, it is just an ISA.

"Come on. Sustained means nothing, right, the one thing that you know Apple's chips are horrible at in terms of scalability means nothing. Got it." Any chip can run with sustained performance with a bit more cooling and power, yes it means nothing. We are comparing the CPUs, not the form factor.

 

[Vya Domus]

Every forum is full of ignorant people like you just ignoring every benchmark (it is Geekbench 5 now, and there are many other benchmarks you can use), ignoring every expert Anandtech included, ignoring actual real world results. World's fastest computer is ARM based? Ignore it. Amazon offering ARM server instances? Ignore it. This is why the world passes some people by. They just can't accept that something has changed.

You mean Anadtech the site that exposed several times how little worth Geekbench has as an accurate benchmark through their tests : https://www.anandtech.com/show/12520/the-galaxy-s9-review/4

It’s when we try to compare the Exynos 9810 versus the Snapdragon 845 where we start to see issues when trying to reconcile the fact that the Galaxy S9 is powered by both SoCs. With its new microarchitecture and significant silicon budget, the Exynos 9810 only manages a 22% and 17% lead over the Snapdragon 845, a stark contrast to the much larger discrepancy that we had previously analysed in GeekBench 4 measured coming in at 37% and 68% for integer and floating point workloads.

You know what they say, you can lead an Apple fanboy to water ...

We are comparing the CPUs, not the form factor.

Because it's the form factor that gives you performance not the CPU itself and it's underlying architecture ? What are you smoking ?

The chip has to the be designed to be scalable under an increased power envelope. The fact that you believe you can just put any chip out their under better cooling and more power and it will just magically run faster shows how primitive your logic and understanding is on the matter.

There is an interesting question about psychology here, why does ARM being fast bother you? Why do you not accept basic reality? ARM is just an ISA, 68000 was fast, PowerPC was fast, x86 was fast, ARM was fast, it is just an ISA.

ARM is an ISA as you said, it can't be fast, careful there you preach what you don't believe yourself it seems. Only thing related to ARM that I mentioned are it's vanilla designs which are fast, in a mobile device.

 

[M2B]

That's not how it works. Again, developer here. You don't go to double the power consumption using both cores, there's no complete power gating here. That's why I showed you the 6W max for everything in the GPU test.

The iPhone doesn't turn off one core and run at half power when you do the test. In singe core performance it takes less than 6W and beats the 9900k in integer performance already. In multicore it throttles down slightly and runs two cores at about the same power. And anyways we are talking about the 5nm EUV A14, that one will beat Intel easily. Give it higher clock speeds ala laptop or desktop form factor and it will beat Intel in FP also most likely.

Don't quote Anandtech and then ignore where they say Apple is faster in integer than the 9900k already and that was a year ago. I have a Epyc 24 core server, a 10900 development machine and a Macbook Pro and Ryzen 4000 latop in the house, and iPad Pro and iPhone (work pays for stuff). I have no problems with performance. The question is why do you believe it isn't fast? Anandtech, every benchmark, every program, and tons of youtube videos are out there, go have fun.

That's not how AnandTech measures power.

Take a look at this:

As you can see, the measured power draw of a small thunder core inside iPhone 11 is around 0.3W, this literally proves my point, that 0.3W figure can't be for the whole SoC, RIGHT? It's just a single thunder core. The same story is true for their big core graph.

 

[TheoneandonlyMrK]

We are talking about the chips, not the form factor. Stick them in an iPad the sustained performance is higher. Sustained means nothing, you just stick the A14X in a laptop form factor and it would be sustained. They are only at 6W and 2.6ghz and you think they can't get higher. Ok.... *backs away slowly*

ARM is an ISA it has nothing to do with how fast the CPU can be.

ARM-based Japanese supercomputer is now the fastest in the world

Fugaku is being used in COVID-19 research

www.theverge.com

You quoted the part wherein i mentioned chip technology yet you commented on the throwaway form factor comment ,go back, have a go at a smart ass way of beating the laws of nodes again, they decide speed that and the design whats bionic designed for again, is it speed? purely speed?, speed costs transistor budget and energy simple ,readup on chip design, you cant make a fork a into a spoon.

Stick them in an Ipad and you got a good web browser yes but do any gaming ,3d modeling , engineering or simulation work on it and it will lag way behind that 9900K, which couldn't possibly sit in that form factor tbf.

sustained means nothing to your perception, That 5Nm chip is made for the platform its in, stick it in a laptop and it will clock about the same ,it's silicon limit is what it is.

 

[Searing]

hahahaha like I said, something about Apple brings out the ignorant haters. Have fun spewing nonsense. There's so much in the last 3 comments, no point. You didn't read or understand the earlier comments anyways.

Suddenly Apple has a 2.6ghz silicon limit, you can't do anything except light work, ARM is an ISA that can't be fast blah blah (despite the world's fastest computer being based on ARM), still trying to suggest Apple uses more power than they do, when you can literally measure it at any time, pretending Anandtech didn't say the A13 was shockingly fast and performant.

No benchmark represents all performance, it creates a statistic that represents complicated information with one number (I guess you'd hate my masters mathematics and statistics education since you hate my developer experience also, btw I hate macs, but I have to make all the web code work with apple devices, iOS in particular). Geekbench 5 is one. Spec2006 is one. How long it takes you to export a video on iPad (faster than my PC since it uses dedicated hardware) also one. Go find 100 benchmarks that all show Apple CPUs at the top in performance efficiency and come back here. That was the A13, wait for the A14X.

Apple is leaving Intel behind for a reason. My 10900 is fast, but nothing special. Same cores from 4 years ago.

 

[TheoneandonlyMrK]

hahahaha like I said, something about Apple brings out the ignorant haters. Have fun spewing nonsense. There's so much in the last 3 comments, no point. You didn't read or understand the earlier comments anyways.

Suddenly Apple has a 2.6ghz silicon limit, you can't do anything except light work, ARM is an ISA that can't be fast blah blah (despite the world's fastest computer being based on ARM), still trying to suggest Apple uses more power than they do, when you can literally measure it at any time, pretending Anandtech didn't say the A13 was shockingly fast and performant.

No benchmark represents all performance, it creates a statistic that represents complicated information with one number (I guess you'd hate my masters mathematics and statistics education since you hate my developer experience also, btw I hate macs, but I have to make all the web code work with apple devices, iOS in particular). Geekbench 5 is one. Spec2006 is one. How long it takes you to export a video on iPad (faster than my PC since it uses dedicated hardware) also one. Go find 100 benchmarks that all show Apple CPUs at the top in performance efficiency and come back here. That was the A13, wait for the A14X.

Apple is leaving Intel behind for a reason. My 10900 is fast, but nothing special. Same cores from 4 years ago.

fine example you raised " How long it takes you to export a video on iPad (faster than my PC since it uses dedicated hardware) also one. Go find 100 benchmarks that all show Apple CPUs at the top in performance efficiency and come back here. That was the A13, wait for the A14X"

your example uses a coprocessor, a special accelerator, something others also do, and certainly a hardware feature that helps with efficiency and performance while doing daily tasks ,like browsing , but they are not the cpu are they, apple wins on efficiency is a given, as i said ,put to task, as i think every desktop processor should be ,its entire life, then the 9900k gets way more work done, as for ryzen 4800X/5800X, who knows.

I use everything but Apple devices personally but i have used them , and there's no hate just understanding, I see their wares, i like the Os but no still ,Apple simply cannot do all that i want from one device ,quickly.

funny perspectives aren't they, honestly it doesn't matter to me, I think they will sell well and perform well for their target market , mostly.
and few will complain ,because the perspective that gets one of those devices into your hands in the first place means you knew what you wanted and were going to both get from the device and do with it , apple is built on consistency,ease and reliabilty, with panash, f knows how you spell that ,even checkers baffled.

you still havent told us how you have both cutting edge nodes and high clocks yet either.

 

[Vya Domus]

ARM is an ISA that can't be fast

Unironically believing that an ISA is what determines if something is fast or not is by far the dumbest and nonsensical idea from all of the posts here. You outclassed all of us.

 

[Searing]

Unironically believing that an ISA is what determines if something is fast or not is by far the dumbest and nonsensical idea from all of the posts here.

I know. He said that, not me, I was quoting him.

 

[Vya Domus]

I know. He said that, not me, I was quoting him.

I can't find anywhere that quote.

ARM is an ISA that can't be fast blah blah (despite the world's fastest computer being based on ARM)

Implying there is a correlation between ISA and speed.

You did the same a couple of posts above :

ARM is just an ISA, 68000 was fast, PowerPC was fast, x86 was fast, ARM was fast, it is just an ISA.

 

[Searing]

I can't find anywhere that quote.



Implying there is a correlation between ISA and speed.

You did the same a couple of posts above :

Now you've really crossed in to the deep end. I was stating things in opposition to him saying ARM can only be slow and you are actually quoting me pretending I said the opposite. I SAID the ISA doesn't matter. I get it, you want to win every argument. I post quotes from reviewers or experts or show slides, no matter. You just keep going with your own thoughts. Good day, bye.

 

[Vya Domus]

I was stating things in opposition to him saying ARM can only be slow.

Yeah, by saying that it's fast instead . Nice 200 IQ backpedaling bro.

I post quotes from reviewers or experts or show slides, no matter.

So did I, except you ignored them because it went against your fantasy world.

Now you've really crossed in to the deep end.

I'm glad it took that long, you did so from your very first words with your avid fanboysm. Nice try though.