Since the release of the first Galaxy S devices, Samsung has relied heavily on its in-house System-on-a-Chip (SoC), the Exynos. The CPUs in the Exynos chips are based on the Cortex series, licensed directly from ARM. Samsung frequently includes Mali GPUs in Exynos chips, which are also licensed directly from ARM. In contrast, Samsung’s primary SoC competition comes from Qualcomm and Apple – both of whom merely license the ARM instruction sets for compatibility and then design their own CPU architecture. While Samsung’s relatively boilerplate designs have fared well – Exynos chips are consistently among top industry performers – there are increasing signs that, with the Exynos 5 Octa, the chip may not be all it’s cracked up to be.
Samsung faces two major obstacles in the SoC market. First, it’s simply unable to compete with Qualcomm’s LTE baseband products, which severely limits Samsung’s ability to deliver SoCs with integrated mobile connectivity. Second, Samsung is largely at the mercy of others for real improvements to its chips – either by ARM in design or by industry limits on fabrication. Arguably, GPU development is another issue for Samsung, but the inclusion of the beefy PowerVR SGX544MP3 with the Exynos 5 Octa (5410) proves that the company is willing and able to simply buy its way around that issue.
Fortunately for Samsung, its LTE connectivity issues can be resolved in the same manner. To bring LTE connectivity to the Exynos 4 Quad (4412) in the Note 2 or to the Exynos 5 Octa in the Korean Galaxy S4, Samsung paired the Exynos chip with a Qualcomm baseband chip. Apple does the same thing with the iPhone 5. The ongoing problem for Samsung in this space is one of timing. Qualcomm tends to release its latest baseband chip embedded in the latest Snapdragon SoCs and only releases the baseband as a discrete (stand alone) unit months later. This works for Apple because it releases new iPhones in the fall, after Qualcomm’s discrete units are available. By releasing the flagship Galaxy S phones in the late spring, Samsung is left to choose between using the latest baseband technology – and therefore the rest of the Snapdragon SoC – or pairing last year’s Qualcomm baseband with this year’s Exynos. For both the S3 and the S4, Samsung chose the Snapdragon in major LTE markets around the world, chips that have been cannibalizing Exynos sales along the way.
Samsung Galaxy S2
The recent history of the CPUs in Samsung’s Exynos processors might suggest that LTE connectivity is not the only compelling reason for the company to prefer not to use its own SoCs. Looking back, it’s possible that the Exynos 4 Dual (4210) in the Galaxy S2 may have been the high water mark for the SoC line. That chip included a dual-core ARM Cortex-A9 CPU built on 45nm. A year later, Qualcomm was shipping the Snapdragon S4, which had dual-core Krait CPU built on 28nm. Krait is Qualcomm’s self-designed architecture built to compete against ARM’s Cortex-A15. The expectation was for Samsung to deliver something similarly advanced in the Galaxy S3. Instead, it released the Exynos 4 Quad (4412). With a quad-core ARM Cortex-A9 CPU built on 32nm, it amounted to only a core increase and size decrease over the previous generation. With double the number of cores and a severely overclocked GPU, the Exynos 4 Quad was strong enough to regularly outperform the Snapdragon S4. It was reasonable at this point to attribute the decision to release the Galaxy S3 in the U.S. with the Snapdragon SoC entirely to fix Samsung’s LTE problem, but the lack of innovation in the Exynos 4 Quad portended larger problems on the horizon.
After it was unable to produce the chip in time for the Galaxy S3 launch, later in 2012, the company finally released the Exynos 5 Dual (5250). It includes a dual-core ARM Cortex-A15 CPU built on 32nm. The chip delivers decent performance, but Samsung’s inability to control the power consumption of the Cortex-A15 design relegated the chip to use only in the Samsung Chromebook and the Samsung-built Google Nexus 10. This is the chip Samsung was supposed to deliver in the Galaxy S3 and instead it came extremely late and was not, in truth, particularly good.
All of this leads up to the Exynos 5 Octa and the Galaxy S4. The Octa is used in the GT-I9500 variant of the Galaxy S4 and is supposed to solve the problem of Cortex-A15 power consumption by using ARM’s big.LITTLE architecture. big.LITTLE allows the use of two core clusters, one for high-performance tasks and one for low-performance tasks. In the Octa, this is a quad-core Cortex-A15 cluster and a quad-core Cortex-A7 cluster, all built on 28nm. In big.LITTLE, there are supposed to be three modes for managing threads across all of the cores in both clusters. Evidence thus far suggests that the Octa really only supports one of these modes – the least efficient. Even worse, it appears that this limitation is due to crippled hardware in the SoC and not something that can be fixed in software.
The first unsupported mode in the Octa is called core-migration. In this mode, each of the four Cortex-A15 cores is ‘paired’ with a Cortex-A7 core. At idle, one A7 core in the first pair would be running at minimum speeds and the others are all deactivated. As load increases, either another pair would come online with its A7 core, or the first pair would ramp up to the A15 core. Each pair is able to independently toggle between the A7 and A15 cores as needed depending on load and threading. This adds efficiency and greatly reduced power consumption. The second unsupported mode in the Octa is called heterogeneous multi-processing (HMP), which allows tasks to be scheduled across all 8 cores independently. This has obvious benefits in maximum power output, but might not always yield improvements in power consumption.
Like this post? Share it!
I still prefer the Exynos Octa core over Qualcomm. Baseband be damned
Well, the Exynos is still beating the Krait 300 (the CPU in the Snapdragon 600) in per-core performance. But it’s losing on per-core performance per watt, which is a much bigger deal for mobile phones (less for tablets).
Samsung also cannot do much with the Cortex-A15 core in terms of continuing advancement. They are limited by ARM’s designs. In contrast, the Krait 300 is already the 2nd production revision of Krait and the Krait 400 in the Snapdragon 800 coming later this year will be the 3rd. Still more revisions will come after that.
The problem with the Exynos 5 Octa is that it doesn’t absolutely crush the Snapdragon 600. The Octa is Samsung’s ‘big’ chip release this year. The 600 is NOT Qualcomm’s ‘big’ release, the 800 is. With Octa as only a minor performance winner against the 600, it means that Samsung has little room to improve to compete with the 800 at all.
The Snapdragon 800 at the same clock speed will still be inferior to the Exynos 5 Octa because the Krait cores aren’t based on the A-15 architecture which is allot powerful whereas the Exynos 5 Octa is. So Samsung still has time and room for improvements and maybe they’ll unleash its full power on the Note 3.
In short the Krait cores lie between the A9 and A15.
Well sure is a good thing the 800 will be much higher clocked then the exynos
Yes but at the same clock speed the most Snapdragon 800 can do is match the Exynos that’s it. Or they’ll have to use the A15 architecture to beat it.
It doesn’t matter, because they won’t be at the same clock speed the 800 will beat the exynos easily
Clockspeed is directly proportional to power consumption and heat production so its a bad way to match the Exynos 5.
Well as its not the same architecture Qualcomm chips have lower heat and power at same clock speed
Lower power, I agree on that. The Exynos 5 is power hungry.
You seem to be missing the point. At the same clock speed, the Krait 400 in the 800 beats the A15 in the Octa by about 10%. And it does so while consuming less power. And after achieving greater performance for less power at the same 1.9Ghz clock speed, the Krait 400′s reduced power envelope allows it to clock another 30% higher still.
Qualcomm doesn’t need to do anything to outperform A15. They already can.
The benchmarks showed a different image though. Well let’s see.
Ah, the benchmarks listed in this page are for the S4 with the Snapdragon 600. The benchmarks for the Snapdragon 800 are elsewhere, and also still pre-production.
The real problem here is that Qualcomm can iteratively improve the Krait architecture. Samsung really can’t change the A15 design, only the way it’s implemented in the overall SoC. Their fabrication is top notch, so that gives them some room to work. But the chips that were known to be in their pipeline have vanished, there aren’t even rumors about the next Exynos, and the company is continuing to fabricate more for other vendors (including Qualcomm).
It really looks like Exynos is dying.
Maybe but when ARM releases the A-57 architecture which will be allot advanced then the A-15, Samsung might have another chance.
Yep, that is the next big thing. And A57/A53 big.LITTLE configurations should be extremely powerful and efficient. The only problem is that they aren’t likely to get produced until 2015 – and that’s a lot of time for Qualcomm to reign unopposed.
2015 ? That’s just too far!
Yeah, exactly my point.
The benchmarks I have seen from the 800 were run locking it at the same 1.9 Ghz as the Octa it was tested against. The 800 was consistently about 10% better. Pushed to the full 2.5Ghz max, that translates to a rather massive performance gap.
I would agree that Samsung’s A15 cores offer greater performance than Qualcomm’s Krait 300 cores, though only marginally. But Qualcomm also says that the Krait 400 will be substantially faster. Samsung has no ability to improve the design of the A15.
-the Note 3 is rumored to have an 800.
-of course Krait isn’t based on A15. It’s based on Krait. They are different architectures entirely. But yes, the one is built to compete against the other.
AnTuTu benchmarks showed that Snapdragon 800 at 2.3 GHZ reported a score near 30000. Where as the Exynos Octa 5 in GS4 when overclocked to 1.8 GHZ had the same result and at times surpassed the 30k barrier which shows A15′s are still advanced than Krait 400 cores.
As these snapdragon 800 benchmarks are only rumoured they mean nothing
The Snapdragon 800 will NEVER be faster clock to clock to an A15. Snapdragon is pretty much a stripped down A15 and some areas changed and customized such as the IPC etc. The architecture is similar or between A9 and A15. They remove the power hungry components from A15 and add there own to make it more power efficient.
Qualcomm always says that it’ll be more power efficient than previous Snapdragon series, but the 600 and S4 Pro heats up like crap when playing games. At 2.3Ghz, I can guarantee it’ll overheat and throttling will appear.
What i’m more interested is in the Tegra 4i, not the full blown A15 Tegra 4.
Tegra 4i uses Cortex A9R4 clocked at 2.3Ghz. To simply put, the A9R4 should be similar (clock to clock) or faster than Apple Swift A9′s.
iPhone 5 at Dual-Core 1.3Ghz can score a Geekbench of 1600. While the Tegra 4i is clocked at 2.3Ghz and 4 Cores, it should achieve around 4000+ on Geekbench in ‘theory’.
I’d prefer the Tegra 4i, uses less power, smaller to conserve energy. Only problem is the GPU is slower than the Adreno 3xx series on the Tegra 4i. Not to mention it’s half the size of Snapdragon 600/800. 80mm2 (Tegra 4i and Tegra 4) vs 147mm2 (Snapdragon 600)
And the Tegra 4i – you know, the chip that was promised way back at CES 2011 – will be entering mass production some time in early 2014. And in devices some time in Q2 of 2014. That is, if NVIDIA can find anyone willing to buy one. They aren’t having much like with the regular Tegra 4.
Meanwhile, Qualcomm will be releasing whatever comes after the 800 at that point. So, concerns about the 4i vs. the 800 are about as relevant as concerns about the Snapdragon S4 vs the Exynos 4 Dual.
But since NVIDIA acknowledges that the 4i will not even be as powerful as the 4, and early reports suggest that the 800 will outperform the 4, I still don’t think the 800 will have trouble standing up against the 4i when it hits market 9 months later.
I don’t know where you get all the bullshit from, but Tegra 4i is going to be released in Q3 2013 not 2014. So your point is invalid. Actually a few company are using it, ZTE, Xiaomi, Asus, Toshiba etc. More will come as the device get released, it’s a cheap SoC with LTE. Plus Tegra 4 devices are coming now.
What reports? You seem to make loads of shit up to make yourself seem right. On Antutu the Tegra 4 at 1.9Ghz scores 35k, while leaked benchmark of the Snapdragon 800 is around 30k. The Snapdragon 800 GPU is faster, thats what I know as on GLBenchmark the LG Optimus G2 (with S800) scores 12xx frame while Tegra 4 11xx frame. So it’s a bit faster.
My ‘bullshit’ here comes from NVIDIA’s public statements. The 4 is entering mass production right now. And has, thus far, one single announced product not from NVIDIA. The 4i is not scheduled to enter mass production until Q1 2014. NVIDIA hasn’t been able to get Icera modem production up and running. This is why they announced all the details of what came to be the 4i (at the time just calling it project grey) way back at CES 2012 and then failed to deliver the chip that year. They will fail again to put it to market this year. Just like they failed to hit their Q4 2012 release timeframe for the 4.
Very well said
At 1.6 GHZ the Exynos 5 Octa is 1.2 GHZ lower (overall) than the 1.9 GHZ Snapdragon 600 and still has a higher score and speed. If Samsung can optimize it and update it then I’m sure it will break records.
Unfortunately, Samsung cannot update or optimize hardware that does not function.
Well they do have a chance in the Note 3 lets see.
Thanks for arguing with author Daniel Charlton I didnt have the energy I knew it would be a brawl. Good job :-)
whatever Soc they use it will LAG ! thanks samsung touchwiz ! So there no point in arguing -_-
I really, really don’t think so. Forget in depth detail of each core instruction list and architecture, the Snapdragon 600 as a whole SoC matches and sometime beats the Exynos Octa in both benchmarks and real world usage. This has been shown everywhere by now. Although, that could be the result of the 600 being clocked at .3Ghz more than the Octa. Another factor that probably matters the most is graphics performance. You can see the Adreno 320 in the 600 has outperformed the 554MP3 of both the iPhone 5 and Galaxy S4 already. The 800 is “rumoured” to be even more powerful, so you know it will outperform the Octa. Remember how well the S4 Pro on the Optimus G wiped the floor with the Exynos Quad 4412? This will be the case with the 800 against the Octa.
That said, let me tell you, being the owner of a I9500 (S4 Octa Variant) the phone is a speed demon. Coming from an Xperia Z, the increase in overall performance is very apparent. The only problem is that some games already optimized for the 600 are not yet available for the Octa.
Unless you’re looking for that microscopic speed difference, and honestly, with a unit like the Galaxy S4, I dont think there will be a noticeable difference between the two in terms of performance.
If I was to throw my two cents in the discussion, I’d say what matters isnt Exynos or Snapdragon. Its Tegra 4. That thing has already released official benchmarks and has shown to outperform the Octa and the rumoured benchmarks of the 800 by a pretty big margin. And you know what matters the most nowadays is graphics processing, like it or not. Which is why I couldnt care less about Cortex (Until the next gen comes out with the rumoured Mali T678 GPU) or Snapdragon at the moment. All eyes on Nvidia.
In Qualcomm (and sometimes nvidia) we trust
That’s why the next Galaxy Tab 3 will use Intel processor…
Give me the snapdragon 800 on the note 3 and im a happy camper!
I hope huwei and other October core makers don’t screw it up like Samsung
Is the s4 pro the same krait version as the snap 600, are both krait 300?
The S4 and the S4 pro are both on the first generation Krait. Krait 300 is the new name for Krait v2. This is one of the reasons why the 600 so handily outperforms the S4 Pro (that and poor implementation in the S4 Pro code by a lot of OEMs).
Its funny on my nexus 4( over clocked to 1.7 GHz) I score 22 000 on antutu and 2700 on geekbench not far off the gs4 or HTC one
That’s probably because you are running a custom kernel. That will usually solve the implementation issues I mentioned.
Are really thats awesome wondered why I was getting such better performance then my mates( un rooted) just for having a .2ghz faster processor
tell me the battery life and the kernel for that perfomance plox my N4 really makes me cry on those departments
Franco kernel, over clocked to 1.7 GHz, slightly under volted by 25 ma all around
yes they are both kait 300 core architecture but with the 600 there are a bunch of tweaks that allowed for a more stable oc of the entire soc.
That’s not what Daniel said…
The S4 and S4 Pro are not Krait 300. Krait 300 is a newer revision of the architecture. This is why the 600 so handily outperforms the S4 Pro even though they have the exact same GPU.
Folks nothing to see here. Just another HTC fan boy spewing bs. Samsung clearly has the best socs go read the reviews.
Given that HTC isn’t in the SoC business, it’s a bit hard to spin this article as HTC-fanboyism. You could certainly argue that it’s pro-Qualcomm, but even still, I am focusing on objective problems in the Octa and places where Samsung can flat-out not compete with Qualcomm (seriously, Samsung’s LTE basebands are a joke – they don’t even use them anymore).
Keep your one.
I don’t have an HTC One. But keep on making false assumptions.
Im sure you don’t. Here’s the thing, Samsung will own hard. Why? Cause theyre a great company that gives users what they want. And in returne they make mi.. Billions. Why you hating on Exynos? Whats your problem?
In roughly 70% of Galaxy S4 sales, ‘giving users what they want’ means a Qualcomm SoC.
That’s not me hating on Exynos. That’s Samsung hating on Exynos.
All his points were objective and factual. Stop being such a Samsung fanboy.
This just means there’s plenty of room for improvement :)
Loved the Linus’s comments on ARM. They never really liked each other, never..
Also nice observation on the ‘either or’ configuration issue. Its true no amount of patches can make up for hardware issue.
I work in the lower-power chip industry. From what I know, Samsung is having big problems with their silicon manufacturing below 32nm. They don’t know how to bring the leakage power down. Power consumption of the chips are only acceptable when the manufacturing conditions are very specialized. This is why they are unable to make a lot of these chips (only 30% of S4 uses octa).
Qualcomm, on the other hand, had TSMC make the chips for them.
make it 10%
The Octa is down to 28nm, so that would explain the yield issues compared to the 32nm Exynos 4 Quad and Exynos 5 Dual.
Everyone has been struggling with the drop to 22nm. It’s was supposed to come in early 2014, but it’s looking like TSMC will skip it and go straight to 16nm in late 2014. That will make ARM better able to compete with the Intels that will be at 14nm at that point (and looking to 10nm soon thereafter).
If Samsung could solve the issues with the 20/22nm process and move to it before Qualcomm can drop to 16nm, then they could really get back in the game. But I don’t see it happening.
Tegra 4 Vs. Snapdragon 800 Vs. Exynos 5 Octo?
Need to wait to get our hands on a couple of those first…
Octa is DEAD
Honestly I like the concept in the Exynos Octa, I think the design from ARM is in fact flawed and Samsung is either locked into the ARM Reference spec by contract or they simply don’t want to invest the $$$ needed to really improve things…
If Samsung were to take a full year and some real R&D $$$ they could use the ARM instruction set to make a ground up chip that really blows the competition out of the water, with all the integration of Qualcomm along with a better power consumption/performance ratio but that would take a real commitment on Sammy’s part…
Personally I think they take their huge market share on Chip Fabrication a bit for granted, they focus more on producing chips than designing them…
Just my 2 Cents though…
that said I would like to see an S800 in the Note 3 at this point… Power consumption reduction is worth the minor performance loss.
Samsung is locked into ARM’s reference specs, but only in the same way as other ARM licensees. NVIDIA has the same license and they innovate outside the core architecture design. Samsung just chooses to make boilerplate CPUs. It would be a lot easier for the company to invest in that sort of re-engineering than to push their own architecture. Apple only just made that switch with the A6, but they bought the design company like 5 years ago. It would take Samsung a lot more than a year to come up with something like that.
On the other hand, their fabrication is quite good. Their edge over TSMC in this space is what allows their boilerplate designs to perform as well as they do. If they scrapped Exynos altogether and tried to muscle out TSMC, the company would probably be just as well off financially. It’s not like they sell the Exynos chips to many other customers anyway.
Excellent article. For the last year I’ve been saying that the S3 and S4 are over-rated. Don’t get me wrong they are competitive top end smartphones but neither were the revelation that the S2 was.
One minor criticism, saying the mali 400 @440MHz is ‘severely’ overclocked sounded negative. Once again don’t get me wrong it was underpowered even @440MHz for the S3 but I have the mali in my S2 clocked at 400 with a small voltage bump and am pretty sure it could go to 440 too. Now the S2 is a 45nm die yet the S3 is 32nm die and users report overclocks in excess of 700Mhz. ‘Severely’ overclocked it is not!
You are correct that my depiction was not sufficiently nuanced. The clock speed was well beyond the spec for the Mali-400, but it was achieved without much power loss due to the drop in process size. And that definitely gave it the oomph to outperform the Adreno 220 in the Snapdragon S4 of the time.
Didn’t realise the Mali 400 outperformed the equivalent Adreno gpu. Maybe it wasn’t underpowered. Having said that if you look at pixel fill rate it’s good, in comparison triangle rendering on the Mali is an order of magnitude slower than other devices. Some rare but specific usage scenarios can bring the mali to its knees which I’ve yet to observe in other gpus.
The Adreno 220 was largely a stopgap. The 3xx series wasn’t ready until the S4 Pro, so Qualcomm updated the older architecture. It was certainly capable, but the overclocked Mali 400 in the Exynos 4 Quad could beat it pretty handily. That’s what accounts for most of the Quad’s performance edge over the S4 – not the 2 extra cores.
In the Exynos 5 Dual, Samsung used the next gen Mali (I think T604). That GPU was simply not able to cut it – particularly with the pixel-dense Nexus 10.
For the Octa, Samsung has switched to IC’s PowerVR (the same thing Apple uses). These GPUs are top-of-the-line in mobile right now. The GPU in the Tegra 4 could top them, but right now, the PowerVR is the chip to beat for GPUs.
It’s great to see power VR gpu’s make their way to android smartphones. It’s probably the only thing iPhone users have that makes me jealous.
Had an interesting discussion with @Gilles LeBlanc about lack of smartphone bandwidth. Do you think with increasing Mpixels gpu’s will or already partially are bandwidth starved?
apple’s new gpu will own anything qualcomm can put out. And Nvidia always talk big but that was what happened with the tegra 3 which was garbage.
Apple doesn’t currently make a gpu nor was I aware they had plans to develop one. I assume you meant the new powerVR implementation they use.
yes, that is what i meant, though I wouldn’t be surprise if they did start making their own gpu. At any case I’m sure apple will implement some version of powervr sgx554 or even rougue.
Assuming rogue is ready and Apple follows it’s previous trend the rogue would launch in iPad while the SGX554MP4 would trickle down from the current iPad to the new iPhone.
The true battle would be between Rogue and the ULP GeForce in Tegra 4.
The Adreno 320 has overtaken the 554MP3 in benchmarks but fell short to the 554MP4. Tegra 4 has already handily beaten all of them, so it holds the crown (assuming the 800 wont beat it after all), until it faces off against Rogue.
Another factor is the platform. Dont forget, its iOS vs Android, the latter of which is a far heavier an OS along with normal level real time multitasking, which will bog down overall performance. So iOS has a lot to do with the impressive graphical benchmark scores from Apple devices, not to mention superior system level optimization by Apple as well.
This just in http://www.androidpolice.com/2013/06/03/arm-announces-cortex-a12-cpu-and-mali-t622-gpu-architectures-mali-v500-video-processing-solution-coming-in-2014/ New midrange Mali gpu launching 2014, T622, 50% faster than the T604.
so do i want the exynos octa or snapdragon800 in the note 3????
Then I did a bad mistake getting the octa gs4 from uae since its unavailable to europe. (Upgraded from S2) My next phone is going to have the A50 cores. Maybe galaxy s6?