[TechReport]ARM unveils Cortex-A72 CPU, Mali-T880 graphics, and more

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
Toggle sidebar Toggle sidebar
I

imported_ats

Senior member
Mar 21, 2008
422
64
86
Fjodor2001 said:
84% CPU performance increase! :eek: That's actually massive. How long can they continue improving at this rate, while we're seeing ~8% performance increase on Intel desktop CPUs?
Click to expand...

That can probably sustain it for -10+ years. Yes that's a minus, as in its never been done and never will be done and people should probably take pure marketing slides on performance with no details on an uArch that has never been tested on a process that currently doesn't exist with a bit of a grain of salt. Or several mountains full.

Here's a first clue, anyone think that 20nm A57 is 1.9x better performance than 28nm A15. We actually have some benchmarks for these two, fyi...
 
Last edited:
N

Nothingness

Diamond Member
Jul 3, 2013
3,315
2,385
136
imported_ats said:
people should probably take pure marketing slides on performance with no details on an uArch that has never been tested on a process that currently doesn't exist with a bit of a grain of salt. Or several mountains full.
Click to expand...
I definitely agree and would like all Intel bulls to treat Intel marketing slides in the same way.

Here's a first clue, anyone think that 20nm A57 is 1.9x better performance than 28nm A15. We actually have some benchmarks for these two, fyi...
Click to expand...
Yep, and the important disclaimer is "sustained performance". Basically we don't know how much faster this chip is IPC wise.
 
ShintaiDK

ShintaiDK

Lifer
Apr 22, 2012
20,378
146
106
Another slide.

CoreLink.CCI500.Diagram_575px.png
 
A

Andrei.

Senior member
Jan 26, 2015
316
386
136
Sweepr said:
ARM claims 2.5GHz sustained frequency for mobile devices. That could easily account for a large portion of that 84% performance increase (depending on the sustained frequency of Cortex A57 @ 20nm). A57 @ 14nm should provide better comparisons.
Click to expand...
Sustained perf on the 5433 is like 1.2-1.4GHz, again depending on what load is used.

As mentioned in our pipeline post, careful with the marketing numbers:
http://www.anandtech.com/show/8957/arm-announces-cortex-a72
 
SOFTengCOMPelec

SOFTengCOMPelec

Platinum Member
May 9, 2013
2,417
75
91
Source:

the Cortex-A72 processor will deliver CPU performance that is 50X greater than the leading smartphones from just five years ago.
Click to expand...

I'd love to see what a 48-core (per compute node) Cortex-A72 based (micro/Blade) server can turnout, in terms of performance, power consumption savings and overall cost.
 
S

SlimFan

Member
Jul 5, 2013
92
14
71
Idontcare said:
ARM's #'s are measured. Only Intel's silvermont data are based on Intel's projections.
Click to expand...

I think that ARM's presentation used ARM's projections of Intel's Silvermont power and perf, and not Intel's projections.

Interestingly enough, I don't think ARM has published a slide like that since Silvermont products have been available in the market. The last thing I recall them talking about with regards to Atom power was expressing concern about the overhead due to emulating ARM instructions with Houdini for apps that are only available as ARM NDK.

In other words, ARM stopped talking about Atom CPU power.
 
NTMBK

NTMBK

Lifer
Nov 14, 2011
10,458
5,844
136
witeken said:
desk.png


With a 16nm process it would be able to sustain that boost frequency like Silvermont can.
Click to expand...

Does Silvermont really sustain >2GHz clocks in a smartphone? I kind of doubt it, though I haven't seen good reviews for an Intel phone annoyingly.
 
Arachnotronic

Arachnotronic

Lifer
Mar 10, 2006
11,715
2,012
126
NTMBK said:
Does Silvermont really sustain >2GHz clocks in a smartphone? I kind of doubt it, though I haven't seen good reviews for an Intel phone annoyingly.
Click to expand...

I keep waiting for one to become available so that I can buy it and test it...
 
III-V

III-V

Senior member
Oct 12, 2014
678
1
41
imported_ats said:
Here's a first clue, anyone think that 20nm A57 is 1.9x better performance than 28nm A15. We actually have some benchmarks for these two, fyi...
Click to expand...
Do you not remember how bad initial revisions of the A15 were? I wouldn't be surprised at all if their 1.9x claim is true.
Arachnotronic said:
I keep waiting for one to become available so that I can buy it and test it...
Click to expand...
Isn't the ZenFone 2 due next month?
 
Last edited:
H

Haserath

Senior member
Sep 12, 2010
793
1
81
Exophase said:
They won't come anywhere close unless they want to reach similar levels of power consumption. You don't design a uarch that can scale to those frequency levels unless that market segment (desktops, high end laptops, servers) is a critical target. Because to do so means sacrificing efficiency at the lower end. This doesn't make sense for an ARM CPU whose most popular markets will remain phones and tablets by a wide margin.

For comparison see Atom which also isn't projected to exceed 3GHz any time soon, let alone 4GHz.



The memory controller/interface used doesn't have anything to do with the CPU core, and if ARM's estimations are remotely honest they'll be of the core(s) only (and maybe the interconnect)
Click to expand...

A few more nodes down the line and 3-4Ghz is within reach in a phone/tablet(turbo) with a lower performance core than Intel's Big cores. Similar levels of power would be only if they made it as wide, and Intel is also bringing 3-4Ghz down to that envelope(meaning each node is faster for the same power in mobile).

Arm's performance line has not been about absolute efficiency. It's probably why they have big.LITTLE. Looks like the new CCI-500 helps improve upon that and it also has a 30% memory improvement. So they might be testing with similar memory speed but with this chip and its other improvements.

So 30-40% from the process, 15-20%? from architecture, a little from the 'uncore', and 20 bajillion from marketing that one corner case where it's fantastic.:sneaky:
 
oobydoobydoo

oobydoobydoo

Senior member
Nov 14, 2014
261
0
0
Fjodor2001 said:
The level of performance increase being displayed by ARM is still so substantial that even if you remove any such marketing fluff it will be truly impressive. So I think the question in my previous post is still valid. How long can they keep up this improvement rate, when Intel delivers 8% desktop CPU performance increase? Won't ARM hit a wall soon?
Click to expand...

I have to agree, I think ARM has become the company that is truly moving CPU and GPU design forward in 2015. Intel still maintains a healthy lead in peak performance and I highly doubt ARM will be beating them in that specific stat anytime soon, but I don't think that is even their objective. Since the whole windows software stack is stuck with intel and their performance cadence it really makes no difference whether or not ARM can beat them, they may not even want to. If ARM can offer a competing core with even half the peak performance of intel you will see Apple, Google, and others tripping over themselves to get out from under intel's suffocating margins.

I would like to note that it is interesting to see the same posters point to this and claim ARM's claims are just marketing hyperbole, or "hype", are the exact same posters who paint these boards with endless lectures on each iteration of intel's roadmap, the whole while claiming it is set in stone, and not marketing at all. To them I say: Why do you not suspend disbelief now?
 
III-V

III-V

Senior member
Oct 12, 2014
678
1
41
Arachnotronic said:
Yup. Kinda a long wait for a 22nm SoC...
Click to expand...
Just a bit.
oobydoobydoo said:
I would like to note that it is interesting to see the same posters point to this and claim ARM's claims are just marketing hyperbole, or "hype", are the exact same posters who paint these boards with endless lectures on each iteration of intel's roadmap, the whole while claiming it is set in stone, and not marketing at all. To them I say: Why do you not suspend disbelief now?
Click to expand...
I see your point, but I feel like performance claims and roadmaps are on entirely different levels. It's a bigger deal to say "we're gonna beat our competitors by x amount" than it is to say "we'll have a new product in this timeframe." Really though, I don't think roadmaps -- unaccompanied by other statements -- are hype at all.

Intel did do the same thing, back when they disclosed Silvermont's architecture. It was probably, at least in part, a response to ARM's claim that Intel was about to get beaten with Silvermont, which was not even close to being the case. Given this, I'm not really surprised that people are wary of ARM's claims, but I personally find them reasonable.
 
Last edited:
witeken

witeken

Diamond Member
Dec 25, 2013
3,899
193
106
NTMBK said:
It's a good job their competition hasn't been shipping phones with 20nm SoCs for 4 months already D:
Click to expand...

I wonder how 20nm vs 32nm would go. 32nm probably doesn't have a hard time competing against 20nm.
 
You must log in or register to reply here.

TRENDING THREADS

COMPANY
RESOURCES
FOLLOW
Top Bottom