The ARM vs x86 Wars Have Begun: In-Depth Power Analysis of Atom, Krait & Cortex A15
by Anand Lal Shimpi on January 4, 2013 7:32 AM EST- Posted in
- Tablets
- Intel
- Samsung
- Arm
- Cortex A15
- Smartphones
- Mobile
- SoCs
Modifying a Krait Platform: More Complicated
Modifying the Dell XPS 10 is a little more difficult than Acer's W510 and Surface RT. In both of those products there was only a single inductor in the path from the battery to the CPU block of the SoC. The XPS 10 uses a dual-core Qualcomm solution however. Ever since Qualcomm started doing multi-core designs it has opted to use independent frequency and voltage planes for each core. While all of the A9s in Tegra 3 and both of the Atom cores used in the Z2760 run at the same frequency/voltage, each Krait core in the APQ8060A can run at its own voltage and frequency. As a result, there are two power delivery circuits that are needed to feed the CPU cores. I've highlighted the two inductors Intel lifted in orange:
Each inductor was lifted and wired with a 20 mΩ resistor in series. The voltage drop across the 20 mΩ resistor was measured and used to calculate CPU core power consumption in real time. Unless otherwise stated, the graphs here represent the total power drawn by both CPU cores.
Unfortunately, that's not all that's necessary to accurately measure Qualcomm CPU power. If you remember back to our original Krait architecture article you'll know that Qualcomm puts its L2 cache on a separate voltage and frequency plane. While the CPU cores in this case can run at up to 1.5GHz, the L2 cache tops out at 1.3GHz. I remembered this little fact late in the testing process, and we haven't yet found the power delivery circuit responsible for Krait's L2 cache. As a result, the CPU specific numbers for Qualcomm exclude any power consumed by the L2 cache. The total platform power numbers do include it however as they are measured at the battery.
The larger inductor in yellow feeds the GPU and it's instrumented using another 20 mΩ resistor.
Visualizing Krait's Multiple Power/Frequency Domains
Qualcomm remains adament about its asynchronous clocking with multiple voltage planes. The graph below shows power draw broken down by each core while running SunSpider:
SunSpider is a great benchmark to showcase exactly why Qualcomm has each core running on its own power/frequency plane. For a mixed workload like this, the second core isn't totally idle/power gated but it isn't exactly super active either. If both cores were tied to the same voltage/frequency, the second core would have higher leakage current than in this case. The counter argument would be that if you ran the second core at its max frequency as well it would be able to complete its task quicker and go to sleep, drawing little to no power. The second approach would require a very fast microcontroller to switch between v/f modes and it's unclear which of the two would offer better power savings. It's just nice to be able to visualize exactly why Qualcomm does what it does here.
On the other end of the spectrum however is a benchmark like Kraken, where both cores are fairly active and the workload is balanced across both cores:
Here there's no real benefit to having two independent voltage/frequency planes, both cores would be served fine by running at the same voltage and frequency. Qualcomm would argue that the Kraken case is rare (single threaded performance still dominates most user experience), and the power savings in situations like SunSpider are what make asynchronous clocking worth it. This is a much bigger philosophical debate that would require far more than a couple of graphs to support and it's not one that I want to get into here. I suspect that given its current power management architecture, Qualcomm likely picked the best solution possible for delivering the best possible power consumption. It's more effort to manage multiple power/frequency domains, effort that I doubt Qualcomm would put in without seeing some benefit over the alternative. That being said, what works best for a Qualcomm SoC isn't necessarily what's best for a different architecture.
Facebook
Twitter
RSS
140 Comments
View All Comments
some_guy - Saturday, January 5, 2013 - link
I was reading comments about how Intel can't compete on price because it will cannibalize profits from higher margin chips.My sense is that Intel will aggressively attempt to squash ARM just like cored AMD. I think they are aware of the imminent threat.
Cannibalizing their own profits from higher margin chips is inevitable and may be a better choice than being eaten alive by ARM.
(BTW I didn't read all 11 pages of comments, so sorry if this was already stated.)
nofumble62 - Saturday, January 5, 2013 - link
Intel current margin is >60% but its stock price didn't rise an inch.The concern is about architecture and future.
As long as they can get into mobile and get market share back, their stock price will spike.
So YES, they can and will drop the price to compete. Their cheap smartphones are already in store in India, China, and Europe.
Don't forget that there are plenty of companies doing well with much smaller margin.
mugiebahar - Saturday, January 5, 2013 - link
While theory will agree, it's not so easy. Simple illustration (only in smart phone tablets)Company 1 (ARM) - employes 1 person and works from home.
Company 2 (Intel) - employes 100 people and rents a warehouse.
Cost to run business 1 - ?
Cost to run business 2 - ???
ARM owns the market now (just for now is all I'm saying) - revenue per expenses = great
Intel is negligible - revenue to expenses = disaster (not that won't change, but the
money spent already over the years)
Company 1 has loyalty and legacy in this market.
Company 2 is HOPING it has loyalty and the it legacy support will matter in this market!
The article makes it seem that it's a one deal Intel will be in the marke and we just have to wait. But that's wrong. Right now we are having phone that can do everything we need at the moment. Yes it will get better but we don't need the leaps n bounds anymore @ a cost we really don't need to spend.
Also if Intel sells a core CPU SoC and it goes for 20-30 like ARM SoC. uneducated people will look at it as an insult that the desk top costs so much. And while its completely different levels of power average joe won't know, and humans are the worst critics correct? I stated before its not about intel not being able to make a chip. It's their company s structured so wrong for this market. You have to be small and agie. AMD would have a better shot at it, if they had the smarts.
some_guy - Sunday, January 6, 2013 - link
It could get ugly.Intel will not give up the high end ARM market without a fight because next stop is desktop market. And Intel may not succeed.
Apple is already majorly moving away from Intel which gives the lots of leverage.
This may be Intel's long twilight, and AMD's shorter twilight.
CuriousSoul - Saturday, January 5, 2013 - link
Without knowing the Frames Per Second, Games settings for 3D Game 1 how is is possible to compare all of the graphs in any sense other than maximum mW ?It's like saying a VW Golf is more frugal around a race track than a Ferrari !
One resistor could draw 1mW and another 1000mW but which is better at the actual 3D Game 1? (in this case they offer an identical experience).
Cold Fussion - Saturday, January 5, 2013 - link
Yea this was mentioned multiple times before, It doesn't do very much to alleviate the claims of Anand's Intel bias does it?jameskatt - Sunday, January 6, 2013 - link
The first problem for Intel is that every major ARM chip company develops their own ARM chips. Apple does it. Samsung does it, etc. This means that these companies will NOT buy third party CPU/GPUs for their ARM devices. By developing their own ARM chips, they cut out Intel as the middle man and Intel's profits. For low priced devices, it is crucial to do this since margins can be very low or in Apple's case, the need to make profit is very high. These companies will not be Intel customers for chips. Intel is locked out. This is why Texas Instruments got out of the ARM business.The second problem for Intel is that the ARM-chips have a huge ecosystem supporting it. Apple has their iOS software kingdom with over a million apps. Android has it hundreds of thousands of apps. These ecosystems have a heavy momentum of their own. Their number dwarfs the number of apps available to the Intel platform. Thus Intel has a difficult job convincing anyone to incorporate its CPUs in to devices which will automatically be incompatible for the existing ecosystems.
What Intel can do is to simply keep aggressively developing its own chips so that they can compete strongly with the ARM chips so that Intel can keep the ecosystem it currently has - laptops, desktops, mainframe and supercomputers.
Intel won't be able to compete on price. it has to make its own profits. It won't be able to compete at the bottom of the device market. But it certainly can keep and expand its own arena. Microsoft surface tablets are an example of an attempt to keep its own ecosystem.
omind - Monday, January 7, 2013 - link
James,Intel is not trying to sell IP to existing ARM chip companies (Samsung, Nvidia etc) but instead are creating complete SOCs, including GPU, Video en- and decoders and so on. They are positioning themselves as competitiors to SOC manufacturers not as suppliers.
Secondly, the Atoms already in the market (Medfield) can leverage the Android app ecosystem. This is accomplished by a version of Android's Dalvik adapted to the x86 ISA. I personally own a Medfield-powered Motorola Razr-i and I can say that every single app I downloaded from Google's play store does run without any problem whatsoever on it. Contrary to the popular belief, there is no ecosystem problem for the end user.
Best regards
mrdude - Wednesday, January 9, 2013 - link
I think his point is that today the big device makers are also have their own SoCs, thus locking Intel out from the beginning. The push towards vertical integration threatens Intel drastically. And unlike a Qualcomm, Intel currently has very little to offer that stands out from the rest of the competition in the tablet and smartphone space (I say tablet because the sales of x86 tablets has been really poor. Consumers don't value x86 compatibility on their tablet)Intel just can't compete on price. I know this gets brought up endlessly but it's something that's absolutely true. Intel operates on sky high margins and that's something that's impossible to do if you're only making SoCs for tablets and smartphones. If the high end is cannibalized at an accelerating rate going forward (and the figures I've read point to this) then Intel is going to be in big big trouble. There's just no way the company can maintain its sky high margins and even keep its own fabs (and R&D) by selling ~$20 SoCs. It's just not gonna' happen. Either Intel provides something that only they can offer and that's wanted by every device maker and consumer and warrants a higher price tag or they begin their slow march downhill.
wsw1982 - Friday, January 11, 2013 - link
Intel sell much more processors today than they were in 2010/2009, if they can maintain their own fabs then, I don't see the reason why they cannot now. The $20-$30 SOC is new market for Intel, not a replacement (at least in near feature). The additional market should be welcomed, right? It reminded me that back to 2008 when Intel first introduced the ATOM, a lot of genius started to cry that Intel opened the netbook market and sale chip at $30 and it would cannibalize the notebook market and hurt Intel's profit. But the years after are one of intel's fast growing age.As for the price? Why cannot Intel compete? doesn't it have enough revenue (PC&Server revenue is much more than 2010 and 2009) to even up the R&D? does it need to pay a bloody foundry fee to TSMC (who alone enjoy 40% margin)? does it have to pay a license fee to ARM? Doesn't it already spend a forturn on PR? Doesn't it already have the manufacture volume ? (50% in idle, anyway will be wasted, why don't produce something at least earn some money)
And I think Intel is never as competitive as now. It gaining it position in server market for the last few year, it's pc business has never such high market share, it's mobile chip start to crush ARM in any respect.
http://www.phonearena.com/news/Intel-Atom-powered-...
check this out, the smart phone lenovo K900 score more than 25000 in Antutu on a 1080P display, which not only crush snapdragon pro (4 krait) in Optimus G but also the newly release Samsung Exynos 5440 (2 A15) in Nexus 10