Possible Benchmark for A12 Processor From 2018 iPhone Surfaces Featuring 10% Faster Speeds, 4GB RAM

[Val-kyrie]

https://stratechery.com/2018/intel-and-the-danger-of-integration/

Seems that Intel's problem was that it had pigeon-holed itself into designing chips based on an inefficient and outdated design, not so much that processor speeds were reaching a peak. This is a problem of Intel's own making, and if the competition such as Apple can leapfrog it, Intel deserves every bit of what they had coming to them.

Thank you for the link to a fascinating article. I found it very interesting that Intel refused to introduce Ice Lake at 14nm. I think that decision was as much about their overconfidence in their perceived lead over AMD as it was about their overconfidence in their ability to develop and enhance their manufacturing process.

Still, the article's warning about the pitfalls of integration also apply to Apple:

"... what makes disruption so devastating is the fact that, absent a crisis, it is almost impossible to avoid. Managers are paid to leverage their advantages, not destroy them; to increase margins, not obliterate them."

Apple may not be manufacturing CPUs which it designs, but, if Apple shifts to ARM as many expect, it is betting on its ability to integrate its hardware design with its software.

 

[geromi912]

Why would Apple use the efficiency gains to improve battery life when they can make a thinner phone by using a smaller battery instead? I'm being half way sarcastic, but serious at the same time because this is typically what Apple does if they see an opportunity for something thinner that will maintain roughly the same run time as the previous gen.

That’s so true, if they didn’t made the MBP thinner it would’ve retained the 99Wh battery instead of the 75Wh one and would’ve got 12 hour battery life and more thermal headroom for the RX560.

 

[ROGmaster]

Nope.

Nope what? Geekbench barely gets an X86 CPU lukewarm. It's a really bad benchmark to test the limits of such CPUs. That's why most of the time the leaks for newer CPUs that are going to be released by AMD and Intel show geekbench scores, because of how unreliable this benchmark is.

They’re quite relevant. For example, a quad-core Kaby Lake Refresh U from Intel can briefly reach similar performance as a quad-code Kaby Lake H, even though the latter needs thrice the wattage, but only for brief periods of time.

It's not relevant when I was talking about the architectural capabilities of X86 cpus vs ARM cpus.

What does this even mean?

Simple
X86 can run way more complex and resource heavy applications without breaking a sweat(sometimes more than one such application at a time)
ARM cpus perform great in simpler workloads where the data is fed to them in a very meticulous, organized fashion.

 

[Feenician]

That's why most of the time the leaks for newer CPUs that are going to be released by AMD and Intel show geekbench scores, because of how unreliable this benchmark is.

AMD and Intel leak geekbench scores because they’re unreliable. That’s your contention here? Can you explain the logic behind these leaks?

 

[ROGmaster]

AMD and Intel leak geekbench scores because they’re unreliable. That’s your contention here? Can you explain the logic behind these leaks?

Yes as a member of the enthusiast PC community that's exactly why we often see quite often leaked geekbench scores of upcoming AMD or intel CPUs. And it happens because these companies want to get the hype train going.
Simple

Geekbenck is no Cinebench that's for sure. It doesn't scale well with threads, clocks, IPC improvements, Ram speeds etc. like Cinebench does so most of the time if nothing else is known Geekbenck only creates speculative talks based on the scores it shows.

 

[chucker23n1]

Nope what? Geekbench barely gets an X86 CPU lukewarm. It's a really bad benchmark to test the limits of such CPUs.

The purpose of a benchmark isn't to "test a limit". It's to compare.

If your assertion is that Geekbench unfairly favors ARM CPUs over x86 somehow, you're going to have to substantiate that.

If you want to "test limits", open one Terminal window for each core and run `yes>/dev/null`.

That's why most of the time the leaks for newer CPUs that are going to be released by AMD and Intel show geekbench scores, because of how unreliable this benchmark is.

Er, what? Leaks for new CPUs often show Geekbench scores for the same reason Geekbench scores generally show up a lot: because Geekbench is a popular benchmark.

It's not relevant when I was talking about the architectural capabilities of X86 cpus vs ARM cpus.

If you want to enumerator "architectural capabilities", that's fine, but that's not the purpose of Geekbench. Most people are more interested, at the end of the day, in how fast a CPU runs.

Simple
X86 can run way more complex and resource heavy applications without breaking a sweat(sometimes more than one such application at a time)
ARM cpus perform great in simpler workloads where the data is fed to them in a very meticulous, organized fashion.

Sounds questionable to me, especially as such a sweeping generalization.

Does "x86" include Atom? Or just Pentium M-derived CPUs? Does Xeon behave differently? What about AMD's or VIA's CPUs?

Does this apply only to ARM-designed CPUs, or include Apple designs? Qualcomm designs?
[doublepost=1530880417][/doublepost]

Geekbenck is no Cinebench that's for sure.

It isn't intended to be.

 

[macduke]

Why would Apple use the efficiency gains to improve battery life when they can make a thinner phone by using a smaller battery instead? I'm being half way sarcastic, but serious at the same time because this is typically what Apple does if they see an opportunity for something thinner that will maintain roughly the same run time as the previous gen.

The new iPhone A12 CPU is supposed to be a lot more efficient on the 7nm process. The new iPhone benchmarks are not that much faster. The new iPhone enclosure is rumored to be the same thickness but bigger for each model aside from the iPhone X. Therefore it is reasonable to assume that the battery life will be higher. Although additional RAM does eat into battery life somewhat, I doubt it will be enough to overcome these other gains in efficiency and internal space.

 

[ROGmaster]

--

LoL it's painful how clueless you are.

 

[chucker23n1]

LoL it's painful how clueless you are.

Great rebuttal