There's just no interest in MIPS anymore. It's all ARM or x86 in the "popular" SoC/CPU categories, and as large as the Indian and Chinese markets are, the ARM cores largely dominate in both cases.
Actually, the top Mediatek processor shipping today in a phone is MT6592, an octal-core Cortex-A7 with Mali-450 MP4. And MT6595, a octal-core with for A17 + four A7 and a 2 clusters PowerVR 6200 seems just around the corner.
Was about to comment on this; mediatek is not quite as diminutive as the initial remark suggests. Their current flagship SoCs already match the snapdragon 600 on several benchmarks and the cortex A17 one on the way could seriously compete with Western premium SoCs with a little luck.
Indeed, MediaTek is pushing huge volumes in Chinese market. They still have a lot to work before catching Qualcomm, as a SoC is a lot more than CPU and GPU alone, but they are unbeatable on price.
Chinese manufacturers are flooding the Chinese market with cheap smartphones using MediaTek SoCs; they recently started to establish a foothold in European markets as well.
This article is focused on mobile SoCs, where AMD is has a negligible presence. And if you look only at smartphones SoCs, AMD doesn't even cater for this market.
Sorry -- somehow I missed the hexa and octa MediaTek options. They're still looking like Exynos levels of performance at best, which isn't the fastest chip around, and octa-core Cortex-A7 just seems... silly. There, I said it: I can't think of many use cases for eight of the same core in a smartphone or tablet. In fact, I think having two really good cores is still a better approach, though there are edge cases where quad-core can still prove useful.
Anyway, MediaTek is huge in terms of volume, I agree. That's not even a question. But the MediaTek devices still tend to be cheap, budget offerings. The MT6595, MT6752, and MT6795 will improve performance in some areas, and for $200 devices they should be decent overall, but in most respects the SoCs are still a generation behind.
8 cores is silly but Qualcomm is doing it too with A53. It's silly and it's just marketing but it is what it is. And a generation behind what? Behind the Qualcomm A7 or behind teh fastest SoCs, how many generations are other behind? Intel hasn't even integrated 3D. And the market is cheap and why would anyone sane in China pay 830$ for an iphone when they can get a decent 100$ phone or last year's flagship at 230$? And why would enthusiasts be that wasteful and inefficient?. Are we afraid to get a little bit dirty now? It feels like Anandtech got stuck in a past that is notthing like the real world anymore. How much coverage does Intel gets in mobile and what have they done so far? The US mobile market is just dead because of the carrier monopoly but there is a big world out there with lots of brands and lots of devices. And the next giants are out there too and the next few years will be very bloody with prices tanking and more of the old guard being killed off. But you guys are just missing it all.
The upcoming MediaTek chips look quite similar to the existing big.LITTLE Samsung Exynos chips, that's what I'm getting at. Okay, so A53 is coming, but the MediaTek has traditionally gone light on the GPU side of the equation, and that doesn't really look to be changing. 2-cluster Rogue is already half of the A7 performance (and that's if they hit equivalent clocks), and 4-cluster Mali-T760 isn't exactly going to set the world on fire either. This is why I say they look like last generation products more than current generation.
Hi Jarred, MediaTek is not going to held the performance crown anytime soon, that's for sure. I'm pretty sure high end devices will continue to be based on Qualcomm chips, with the exception of Apple's and, to some extent, Samsung's.
However, the big question is whether their chips will be good enough for medium and even medium-high devices in terms of both CPU and GPU performance. As far as I remember, low and medium smartphone market is growing more rapidly than high-end, at least in terms of volume.
In my opinion, 2 major MediaTek's weak points are: - poor software & developers support - I believe they don't publicly release source code, kernels etc.; this is the reasons you don't see custom ROMs for phones using MediaTek SOCs, as opposed to Qualcomm and Samsung (Exynos) powered models - poor GPS performance - it really sucks compared to Qualcomm's
An Octa-core with Big.Little is technically a quadcore chip unless it is all A7 in the eight cores which is actually silly. In China, people are practical and do buy low-end devices at low prices just to be able to use them. iPhones are status symbols for the rich and that is about it. Apple cannot make much head-way due to high pricing and huge competition in the mid and low ranges of phones. I am surprised that manufacturers/designers are not making TriCore chip which are purposed such that the third is a Little core dedicated to I/O onlywhile the two large cores can switch up and down in speed or off if not used. This lessens the context switch needed for Big.Little design to a large extend thus saving power while performing close to a quad practically. Nvidia's 5th power saving core is an example of such implementation but there was no dedication to I/O processing as such.
you try playing any high profile high bit rate 1080p or better with a dual or lower quad core and anything that uses FFMPEG for the audio etc and you soon see the need for an octacore that ffmpeg/vlc and x264 and downstream apps can use to good effect without stuttering
its even hard for lan streaming your DVB-t2 content to your browser too, a good quad or octacore is the way to go for a purely tech POV...
How many minutes would the battery last with the cores going at full blast? A octacore going full on would likely thermal throttle in most designs rather quickly anyway.
These tasks need to be offloaded to a dedicated DSP.
If that is your benchmark for getting mentioned then Intel has no business on this list either. Regardless of their overall size and talk Atom has gotten zero tracking in the mobile space. As the article itself likes to say its always "wait till X platform is release next year" with them. Meanwhile ARM becomes more dominant every year.
@Jumangi :"If that is your benchmark for getting mentioned then Intel has no business on this list either. Regardless of their overall size and talk Atom has gotten zero tracking in the mobile space."
Totally agree!. Intel is just using the WRONG tech to fight the Arm competition. IF Intel has any sense they ought to be jumping on the RISC-V initiative which actually claims HALF the power consumption of Arm for a given performance level. It is this kind if advantage that x86 can NEVER get to unless Intel has 7nm process node!. Of course Intel can get an Arm license and do their own implementation. But "Not Invented here" syndrome is rampant in Intel. That is for sure. So getting into a competing architecture is the ONLY thing worth doing to counter the Arm attack so far. So Intel: Are you "Blind or Stupid ?". I am pretty sure they are not stupid since they are still around ....
"already match the snapdragon 600 on several benchmarks" is a meaningless claim. Obviously SoCs that are fields of wimpy cores can do well on a certain class of throughput benchmarks, but that doesn't translate into any sort of value for a cellphone or tablet. (They could be valuable for a certain type of server, except there are enough other issues there, starting with 32 bit, that that's also a non-starter.)
As for the A17, well, my vaporware can beat up your vaporware. I assume the A57 based Snapdragons will ship around the same time the A17 Mediatek's ship.
It's a bit strange to me that the article skipped Microsemi, whose "SoC" division -- originally Actel -- created the first FPGA with a built-in ARM processor + micro controller system + 50MHz oscillator + DDR controller + CAN bus + I2C + RTC + NVM. The FPGA itself is built out of flash cells so it doesn't lose its configuration on power loss, and is architected for security, low power, and immunity to SEU.
First of all, nice to see that you're so loyal to your company, coming to post product spec sheets on a public forum! ;-)
Second of all, it must be noted that the manufacturers focused on here are ones which are catering to the smartphone/tablet market. Microsemi's offering seems more targeted towards industrial applications, and has features that aren't as useful in a smartphone (FPGA, CAN bus).
thats all well and godd remembering the old innovators but you forgot to mention and first people to create todays tech that being Ferranti International plc, the UK electrical engineering and equipment firm that operated for over a century from 1885 until it went bankrupt in 1993.
among the related tech they produced were the With the invention of the laser in the 1960s the company quickly established itself in the electro-optics arena. From the early 1970s it was delivering the Laser Rangefinder and Marked Target Seeker (LRMTS) for the Jaguar and Harrier fleets, and later for Tornado.
In collaboration with the University of Manchester they built a new version of the famous Mark 1, Some of the technology of the FP 6000 was later used in its Ferranti Argus range of industrial computers which were developed in its Wythenshawe factory. The first of these, simply Argus, was initially developed for military use.
the Semiconductors they produced where the forerunners of the FPGA In the early eighties, Ferranti produced some of the first large uncommitted logic arrays (ULAs), used in home computers such as the Sinclair ZX81, Sinclair ZX Spectrum, Acorn Electron and BBC Microcomputer.
now when Microsemi have a current ARM cortex SOC with octacore ,x4 SATA,x4 micro PCIe and at least x2 1gigabit ethernet in an ODROID form factor AND price point , then we will talk... what is it with all the ARM oems today limiting their SOC to only phone/tablet specs and noting Quad/Octacore in capable for the SME NAS/SAN range!
Most so called innovations are incremental improvements of existing technology or combining existing technology in a new way. There are very few paradigm shifting discoveries, such as the laser that you mention. These things take time. After all, Einstein gave the theoretical basis for the laser in 1917.
Power envelope is the limiting factor right now, and thus the settling for "good enough". Much of the power used is due to i/o. I expect the next big step to be "new" i/o methods. In particular, on-chip silicon photonics has the potential to drastically increase both the speed and efficiency of chip-to-chip i/o, perhaps even inter-chip i/o, freeing up more power for compute operations. At that point we will define a new level of "good enough", since human wants and desires are endless and we will always want more. ARM is limiting their SOC based on power envelope. Raising that power envelope would pit them against higher performance competition. Intel is trying to enter ARM's low power niche. At the same time, ARM is trying to enter Intel's high performance niche. Neither have yet reached their goals, mostly due to physical limitations that will eventually require replacing electronic interconnects with something else.
Where is the Snapdrogon 410 in shipping devices? Some showed up in benchmarks but there isn't a single device with it already announced as far as i can tell and i do pay attention to emerging markets since they define the market now more than Samsung or Apple do. For the 810, during their latest results call, they said it will launch late this year, if launch means shipping or devices in store no clue really but with Samsung shipping 20nm parts,they might be in a hurry. On Mediatek you guys really got to get more informed and it seems you need to reset your definition of first tier phone makers. Folks like HTC, Moto , Nokia, Blackberry and even Sony are fading away and Mediatek has wins with some of those too. They got almost 30% of the smartphone market this year, you can't keep ignoring them. And in revenue they are bigger than AMD and Nvidia and vast majority of semi companies ( now that they bought MStar). They are also just starting to sell discrete LTE with existing chips and about to launch their quad A17+quad A7 with integrated LTE, the MT6595. And can you really claim they cut corners or that they are less than others when you guys have reviewed 1 single device with Mediatek- as far as i can remember that single device was the Asus Memo pad HD 7. And you never report on them at all. If you ignore them and belittle then it should be on facts not popular beliefs- if we go by that , Mediatek is from China not Taiwan. If you ask me Mediatek is very likely to go for a custom core of their own soon too, it makes sense and they are big. They are about to reach 2B revenue in a quarter, they have gross margins just under 50% and R&D budget above 19% of revenue so they actually are on of the few that can make a custom core and makes sense to go there or even buy someone surprisingly big,You like it or not, they are the only real competition to Qualcomm right now and a serious competitor for that matter. The Allwinner A33 is new i don't think you can see it often in budget tablets as you are saying. I've seen a single tablet with it so far, ofc there could be a handful of them but the thing is new and products with it just starting to show up. Rockchip RK3288 is in a bunch of tablets like the ifive mini 4 , Pipo P1 , P1, P9, P8 , Teclast P90HD , Chuwi V88 and my list is a couple of weeks old.Not all listed hit retail yet. 2 names that you've missed although they are more significant than most names mentioned are Spreadtrum and Marvell , got the feeling Marvell is working on a custom core too and they are one of the very very few that has been shipping LTE for a while now. Broadcom is also not quite out yet. Anyway ,it might be time to look at SoCs in far more detail ,not just the CPU, GPU and connectivity and not even just the SoC. You've got a bunch of other compute units and likely more will be added , while companion chips matter too when it comes to price and the financial aspect decides who makes it or not.
Phones with Broadcom are still launching, just today i've seen one announced in India and that's why i said they aren't out quite yet nut yes you are right they are on their way out (unless something changes).
They are on the way out, so to speak. You can imagine the exit will be gradual, as they have prior obligations to fulfill. Supply chains deals are made with months or years in advance.
Modem business is definitively over, despite having a ready LTE solution. They plan (or have by now) to let staff go in Finland. Renesas/Nokia-LTE team is no more. Doesn't mean Broadcom still won't deliver other products to the mobile market though. They have WiFi, GPS, ARM-SoC's with custom GPU/video-processor for different markets. Even if mobile stuff will get no new app processors stuff like STB's should get new chips. It's 2500 people from the cellular modem divisions they are firing.
From what I see, Broadcom is leaving mobile (phones/tablets) for good and is focusing on other markets. There's no mention about mobile SOCs on their site, as opposed to Marvell.
custom Mediatek ARM core ? don't think about it, they don't have the right guys to do that. They are only hardware assemblers, putting IP blocks in their SoCs. Their support is nonexistent and their software stack/team is pathetic...
You're right, they have a lot to do before even thinking to get into the big league. They software support is really poor, just started to offer Android 4.4 for their SoCs and only for octal-core models, if I'm not mistaken. However, they are on the rise and who knows what future may hold...
A7 is weaker and much smaller that A15, search in previous AnandTech articles, you'll find more detailed info about that. And by looking on Lenovo site, there are only a handful of phones with KitKat: http://shopap.lenovo.com/in/en/smartphones
Qualcomm has a lot of expertise in designing custom cores, both discrete and integrated communication modems, memory interfaces and knows a thing or two on how integrate all the blocks into well rounded SoCs; is not by accident they are the leading smartphone SoCs vendor, Currently, even the behemoth which is Intel is struggling to compete with them; this can give you an idea about the engineering expertise Qualcomm owns and MediaTek is still far from touching them in this area.
In fact, no. I will quote Anand. "The dual-core A7 is now the fastest SoC we've tested under SunSpider, even outpacing Qualcomm's Snapdragon 800 and ARM's Cortex A15"
Apple A7 is a custom ARMv8 (64bit) dual-core design, in a class of its own, for those who don't know. In fact is the first shipping ARM 64bit SoC in a mobile device, which no small feat; it took everyone by surprise, as AnandTech mentioned in Iphone 5S review. That's why Qualcomm is using standard ARM cores (A57, A53) in their first 64bit products; they have to come up fast with something, but their custom 64bit cores (which I'm sure they are working on) are not ready yet for prime time.
@Xinthius: ROFLMAO, you are confusing Apple A7 with ARM Cortex A7, its very clear that @Liviutm is quoting ARM Cortex A7 as generally A7, A8, A11, A57 etc are referred to standard ARM nomenclature and the Apple's stuff is labelled as Apple A6 swift, Apple A7 cyclone etc.. The only reason Anand has mentioned as A7 in that sentence is because you are quoting his lines from an iphone review!!
You do understand how big they are? If you exclude NAND and RAM makers , only Intel , Qualcomm and TI would be bigger than them by next year.They are about to overtake , in revenue, folks like Broadcom, Renesas or STMicro and they could actually afford to buy someone like AMD but it doesn't make sense for now (would be too costly for what they would be getting). As for how far they are from it, it remains to be seen, just because we don't know about it ,it doesn't mean they haven't been working on it but it certainly makes sense for them to go there and they have resources to do so. If AMD and Nvidia can while being smaller ,they can too and given their share it makes a lot more sense. To take more and more share from Qualcomm they kinda need to grow up.
Your enthusiasm for Mediatek is noted, but you seem to be out of touch with reality. Mediatek ships lots of parts, yes, and that's valuable to the poor people of the world, yes, but they are in no way generating new ideas and progress in SoCs, at the logic level, at the CPU level, at the SoC level, at the SW level. Their annual revenues are $4.5billion and their annual net income $.9 billion. To put that in context, Apple still has (after handing out oodles in share buybacks and dividends) about $120 billion of cash on hand.
Mediatek is not interesting to 90% of the people who read this blog because no interesting technology is going to come from them. It's not interesting to the other 10% because no interesting business upheaval is going to come from them. They'll continue to provide cheap chips to struggling Chinese manufacturers who are all on the verge of bankruptcy, and nothing will change over the next few years except some mergers and name changes among those companies.
You're like a guy telling us we should get really excited about Nestle because they sell to so many people, have a presence in so many markets, blah blah. All true -- but Nestle just is not a source of the kinds of things readers who read AnandTech find interesting.
I'm nhe one out of touch? Yu can't even convert TWD to USD or notice the MStar buy and not look at older results. Do you even have a clue what income is? How exactly is income higher than revenue? Apple is not relevant in the SoC space sine they are the only ones using it and they are utterly uncompetitive when it comes to price. Apple will have 14% share in smartphones this year and under 10% in 2 to 4 years depending how aggressive the price declines will be around the world. Mediatek was the first one to enable phones and tabs with a cheap quad A7 and it was quite interesting for any less snobish (or xenophobic ,w/e it might be) then you. Mediatek has has an actual wearables platform unlike some others that push phone SoCs into watches. Mediatek has lower prices because they manage to be efficient not because they take low margins. They had over 49% gross margins in Q2 and they managed some very nice chips (not just the SoC but companion chops) over the years. As for your confusion about who uses Mediatek, i imagine you are American and you go to VZW seen 15 options and think that's all there is out there. The actual phone market is starting to be dominated by the new players that might be global, regional or local and they are gainign share and will keep gaining share. Many of those phones are using Mediatek. Go check stores like CDiscont in France, Flipkart in India , eMag in Romania or JD.com in China. Mediatek is in no way China only, it's just not US because that market is dead, killed by regulators failing to do their job and allowing carriers to murder it. And sure the Chines OEMs are all on the verge of bankruptcy , and thousands of other around teh world ,except the few that will be the next giants, it's called consolidation. The next giants are from China and India and maybe Brazil and so on but not the US. In PC Anandtech should review only 1k$ Intel CPUs? Is that your argument? And to hell with AMD since they suck. For better or worse Mediatek is the AMD of the mobile world. Back t o the financials , in the current quarter Mediate will get about 2billiions revenue. In Q2 their R&D was over 370 millions USD and that's more than ARM itself in an entire year.
Apple certainly isn't uncompetitive on price. They have over 60% of the market in the price range they do compete in. That's where all the money is being made anyway.
Huh, STMicro and Broadcom certainly has a larger revenue and business to build on. Renesas has been going downhill ever since Renesas and NEC merged but has similar revenue to ST and Broadcom. STMicro has lots of business, still have some fabs and lots of employees. Mediatek does have some interesting stuff outside mobile application processors and baseband, thanks to stuff like Ralink, but they are still relatively small and has tougher to access bank funds though are in a much better position than the Chinese competitors.
Broadcom has some 8B plus yearly revenue, Mediatek will reach about 2B in Q3 and they are growing way fast . The other mentioned have less revenue. http://www.electronicspecifier.com/cms/images/1H14... So if you exclude the RAM and NAND makers (Samsung, Toshiba, Micron, Hynix) and obviously TSMC since it's a foundry and factor in that Mediatek is very likely to have well over *B revenue next year they are behind very few. Sure some of the NAND and RAM makers make other things too but i don't know how much revenue that is. Mediatek does report revenue each month and an outlook for the quarter so you can check how much they made in the last months and they gained quite a lot of scale/
By any means I'm not disregarding MediaTek, they are growing fast and they surely can't be ignored. However, in terms of performance MediaTek can't touch Qualcomm - a mobile SoC is more than CPU and GPU alone - I think there are some interesting articles on AnandTech about that. Here you can see a high level block diagram of Qualcomm Snapdragon 810: http://telecomtalk.info/qualcomm-announces-64-bit-... Achieving a balance between performance, power draw and heat is not so easy, but is not talked about too much; nonetheless isimportant - what's the point in having a blazing fast SoC on paper, when in real life it eats battery fast and/or gets too hot and real thermal throttling kicks in, drastically limiting performance?
Mediatek never had a fast SoC yet and don't think anyone claimed that but since when is it just about fast ? In PC we don't care just about 1k$ Intel we care about the entire range. And if Mediatek would suck so much,some better tell us why they are doing well and beating so many other, forcing Qualcomm to change roadmaps and drop prices.
Maybe they could afford to but AMD, but they wouldn't, even if they could. If they did, Intel would pull their license to build x86. That would end any purpose to the deal, and kill AMD.
Mediatek is from Taiwan and has grown by buying several other Taiwanese firms. Even AMD reported larger revenue last year than Mediatek at the time. No one would argue that Nvidia basically just has their GPU-chip market though.
You are mistaken. Mediatek originates in Taiwan (Headquarters (Hsinchu), No. 1, Dusing 1st Rd., Hsinchu Science Park, Hsinchu City 30078, Taiwan), not ROC. http://mediatek.com/en/about/contact/ Other than that, I share your view about MTK. They are the new rising star, now also in quality/features and not longer only in respect to price.
What I was missing however in the whole article, is HiSilicon/Huawei's "Kirin 9xx" SoC line.
AMD just as a note in the closing comments? If you go at wikipedia article about SoC the first picture in the right is an AMD Geode! How old is that? AMD is also the only company that I know with a socketed SoC in the form of the AM1 platform. And AMD is just a line in the closing comments? OK...
This article is focused on mobile SoCs, where AMD is has a negligible presence. And if you look only at smartphones SoCs, AMD doesn't even cater for this market.
How long they've been in the game has nothing to do with their presence in modern mobile smartphones and tablets - which is negligible. I hope for them to do well in it, but with them in so few shipping designs they're a marginal player at this point.
Would you folks mind doing a roundup of inexpensive Windows (full) tablets? The Surface Pro 3 is making all the headlines right now, but there is a fast-growing sub-$300 market of Bay Trail/IPS/8-hour/Windows 8.1 (non-RT) tablets out there. Many of them are currently under $200. Seems like something that could be extremely competitive with Android/iOS, given the power and flexibility.
Acer Iconia W3 ($199) Dell Venue 8 Pro ($199) ASUS VivoTab Note ($199) Lenovo ThinkPad Tablet 2 ($299) etc.
"Would you folks mind doing a roundup of inexpensive Windows (full) tablets? The Surface Pro 3 is making all the headlines right now, but there is a fast-growing sub-$300 market of Bay Trail". No these are "on off" flukes since they rely on free Win8 license and throw-away Intel prices. None of these are sustainable for long. It is just Intel's attempt to claim the low-end of the market until they can churn up higher performing parts (then raise prices). Arm really had the tablet markets to themselves with Android/IOS taking 98% of the volume shipped. Win 8.1/9 will relegate to the x86 behemoths the way they always had due to legacy software dependency. With web-based and cloud based apps, the legacy dependency is going away, so it also explains why the cheaper tablets with Android is so popular since their performance no long sucks like 3 years ago. Interestingly Intel made very little headroom in performance last 3 years except for huge jumps in power consumption at the expense of lowering cpu frequency.
“It is sold with an IR remote control, a 5V/2A power adapter, an HDMI cable, a USB cable, a Wi-Fi antenna, and a user’s manual. You can find it for $122 on Aliexpress including shipping. But before you rush to buy the device, understand that Rockchip RK3288 devices are still not really ready, more like beta devices, even Android TV boxes which are expected to come before HDMI sticks had to have their PCB re-laid out to fix some hardware bugs, and the firmware needs more work based on various reports.”
Marvell is still in the business. They have LTE modems too and have announced new products. HiSilicon (Huawei) is also an upcoming player. Plenty more players outside the phone/tablet market.
Not fun for the guys in Finland who developed the Renesas/Broadcom LTE-modem but basically never saw it ship. Wonder who will snap up those people. Ericsson still offers their modems (from the ST-E venture brake up) at least. To bad to see everything just wasted. Broadcom was never big on app processors in handheld devices, but they seem to continue to develop ARM SoC's for other markets.
TI really went down with Nokia early 11 as a result to basically stop making and investing in phones from Nokia. Microsoft's Qualcomm powered platform eventually grew to about 30 million devices per year for them, which is miniscule for a company that had almost 60 000 employees and made about 450 million devices 2010 including more smartphones then than later on. ST-E really went down thanks to it too, they had invested in development and had been cooperating with ST-E but had to end that when they was forced to use Qualcomm exclusively. As well as shutting down the platform they were cooperating on.
Intel at least seems committed. They don't quit on the wireless business and have made more acquisitions lately. Their LTE solutions are finally shipping.
As smartphones and their prices reach commodity level (already happening in markets like China and India due to Xiaomi, Huawei, Micromax, Asus zenphone etc..) the players like Mediatek will slay atleast few giants in the SOC manufacturing sector.
Agreed. Samsung is really looking like the biggest risk, considering they're doing similar designs to MediaTek. But then, Samsung actually has fabs, making them harder to kill. I'm honestly not sure who will be the next big name to fall. TI wasn't all that surprising, but of the "Big Five" who looks weakest? Intel is still trying to become a bigger player, but they've got big pockets that can support Atom while they try to gain market share. The same goes for NVIDIA, though not quite to the same level and NVIDIA is doing more competitive designs IMO. The reality is that anyone who exits the SoC space at this point stands in danger of disappearing altogether in the next decade.
The difficult part seems the be the modem now so unless ARM or someone else licenses a killer baseband,that's what will kill the next one. There is this myth about costs but if you look at ARM itself they spend just 80 million + per quarter on R&D so it's not that much for some players out there. Even rumors about Google making a custom core for server don't seem all that crazy if you realize that costs might not be that huge. As for being out for good, that's not true,there will be transitions to new form factors ( like glasses or w/e will replace the smartphone) and maybe major changes in connectivity (mesh networks maybe) so there can be room for disruption soon-ish. There is also Project Ara that could somewhat liberalize the market and allow for smaller players to compete Maybe it could even go the opposite way of integration, How about a quad SLI phone lol?
Yeah, its a bit hard to predict such things but bound to happen to some major players in few years. Samsung is worried not just for their SoC business but their most important money printing business: Smartphones :)
Already Xiaomi (Big Qualcomm 800/801 and mediatek customer) has dislodged Samsung as the largest smartphones seller in China with Lenovo and Huawei being other big threats and they sell quality stuff at low prices, just check out Xiaomi Mi3, all premium components at low-end prices, selling for roughly $230 in India and is sold out most of the times, hope you can contact and get a review-unit from unofficial resellers like Xiaomishop who gave gsmarena a review unit or maybe contact Hugo Barra directly on Google plus/twitter, he is very active on social media :)
Also, Micromax (big mediatek customer) in India has come very close to Samsung in smartphone marketshare and with the recent launch of Asus Zenfone 4,5,6 and Xiaomi smartphones, the sales and marketshare for Samsung will be a big problem in few years.
Funfact, Intel got a big design win in the form of Asus Zenfone series, the entire series is based on intel atom platform and is selling quite well, they sold 40000 units within 4 days of launch in India and after that they are constantly sold out eventhough they are bringing 15k-20k units every week.
*Agreed that companies like Xiaomi and Motorola are giving nightmares to the more established brands but that’s not to say that these brands will slowly lose all their value. For example despite reporting a serious drop in profits Samsung did manage to improve its revenue in the US and the Samsung Galaxy S5 sales in the US have been 50% more than the Samsung Galaxy S4 sales of yesteryear. The same thing was being said about Apple a few years ago but still despite losing hefty market shares most quarter Apple still manages to post more profits than any other manufacturer. Many a times market share does not mean everything profits also matter. *As I said despite profits being 25% less for Samsung they still manged to earn more than any other Android vendor in the market. In fact other Android vendors hardly earn anything. The total combined profits of Apple and Samsung are almost 105% or so. Yeah its 105% and no I am not drunk the extra 5% come into picture if you consider the losses companies like LG/Sony/HTC made a few quarters ago. Companies like Xiaomi and Micromax hardly manage to earn any profits. Take Xiaomi for example they are selling devices at almost cost price and hardly any profit on hardware sales. You can’t expect companies like Samsung to start competing at those prices because they spend billions on advertising and R&D and their major source of revenue is hardware profits. *In my opinion companies like Samsung and LG will lose market share rather than reducing prices to the level of Xiaomi which are not manageable by them or rather which would not be enough for their expenses. I will readily accept that in future companies like Samsung will lose market share in countries like India and China to some extent but then to the market share which they are losing are to the companies who are hardly making any profit. And even though Samsung loses market share in India and China they will still manage to do well in countries like US/Western Europe/South Korea/Japan etc etc. In developed countries like US for example I don’t think companies like Xiaomi will be able to challenge the might of Samsung. *In the US market the contract system is followed where carriers subsidize the cost of devices and the 900$ or so iPhone or Galaxy S5 over there “”appears”” like 200$ in the US. Although carriers manage to recover the cost over the 2 period contract but still the consumer has a sense of false belief that he is only paying 200$ for the device. Now in such a market if Xiaomi comes with even a 230$ Xiaomi Mi 3 people would “”believe”” its expensive. Although off contract is becoming popular in US the rule of the thumb is contract and as long as contract systems prevail which I guess they will don’t expect iPhones or Galaxy to lose any ground in US. *On the other hand in countries like US the success of a device very much depends on the carrier relations a particular company has with a carrier because of the contract nature followed over there. Carriers don’t openly provide every device on contract. Many a times only devices which they feel have the potential of doing well are sold on contract. In case of Xiaomi as I have said before they really don’t manufacture devices in bulk and given the people of US hardly know about the brand initial demand will as usual be low and not much of the public will warm up to the device initially. Carriers know this and won’t stock the device. If sold on full retail price the Mi 3 would cost at least 230$ in the US which is 30$ less than the iPhone 5S or so and despite how good Mi 3 is its definitely no where near the 5S for the very least and even the most tech savvy wont go for Mi 3 for just saving 30 bucks. I know that overall on contract the iPhone 5S would end being expensive but still how many people over there would actually figure out this and even for contracts several theories are present. Some people say its cheaper some say expensive. But anyhow as of now and the foreseeable future contracts are to remain in US. *Next people in US are not as poor as other developing countries. People over there earn well and can very well afford expensive phones. But still this is more of a personal belief rather than a good point. *Next up is litigation and patent wars. US is a country with very strict patent laws and no matter how much I like Xiaomi I won’t refrain from saying that they copy Apple. In a very worse manner that too. Its visible all over their products. If Xiaomi were to ever enter US and gain traction I am pretty sure Apple is gonna sue the shit out of them. They did it with Samsung and won’t spare Xiaomi. At least Samsung was earning enough to face the 1 billion $ of fines they have seen from Apple I don’t think Xiaomi posses that kind of wealth that they can suffer litigation. What’s even worse is that Xiaomi copies Apple even more than Samsung then the litigation fee can also be equally high. *Next marketing matters a lot in US. Both Samsung and Apple have big marketing budgets and companies like Xiaomi don’t even spend on marketing in a commercial manner let alone putting up an Ad during Super Bowl or something. Companies like Motorola LG and Sony which have less marketing budgets often end up as losers and to be truthful for the past several quarters most manufacturers except Samsung and Apple are losing market share in the US and its not because Samsung and Apple are coming up with some really innovative products. NO!!. In fact the G2 HTC One M7 and even Moto X were far better than S4 but still it all went in vain just because these manufacturers like LG HTC and Motorola did not have a marketing budget as good as Samsung. * As far as I have seen in the electronics Industry the cheapest wins.”jo sasta hai woh bikta hai”. The TV industry and the PC industry are very good examples of the same. The TV once a cash cow for Sony is now Samsung’s field. Same way Lenovo dominated the PC and lapotop industry. But then again PC and TV are mostly individual products sold separately. For example in the TV industry you still don’t have DTH operators subsidizing the cost of TVs same way broadband providers don’t subsidize the cost of laptops even in nations like US that’s why did the once mighty and premium players fell all over the world. In short the product and services did not share any relation in the PC and TV industry. *The smartphone industry however is different in US etc etc as long as carrier subsidies continue I don’t think its gonna be a challenging way for Apple and Samsung.In the smartphone industry of US etc you have services and products linked. In India you do not have any kind of meaningful subsidies and that’s why do I expect the branded players to lose market share. Even in China the operators are starting to cut down subsidies. * Even if we leave all this as I said only few players are making serious money in the smartphone market.The countries like US Japan etc where high end smartphones are sold the profit made on every handset is almost triple the cost of say a 5k handset like Samsung Galaxy Star Duos. Losing market share would not mean much to Samsung and LG as long as they lose it to players who are not making money.Just remember for every China there is a US. For every India there is Canada. For every Vietnam there is Japan. Even if manufacturers lose market share in India CHina etc they still have other developed countries to look forward to.
You are a stupid guy from India right? Look at my comment below to know the actual truth. Losing market share means nothing to Samsung. Even companies like Xiaomi and Micromax are hardly able to make any money on the millions or so smartphones which they happen to sell.
Ultimately it comes down to IP and a bit of market luck/timing. Those with IP in CPU, GPU, and LTE have a head start on those who don't. Fab technology seems to help too. Because time to make a custom CPU is pretty long and costly, and everyone else competes on stock ARM designs at the lower price points. I'm not sure Samsung will be making a custom design any time soon, as they don't have much CPU/GPU design history.
Samsung is a special case, indeed. Not only they have fabs where they manufacture RAM, NAND flash, logic chips (including SoCs - theirs and Apple's), but they are also a major display (basically they own AMOLED market) and battery producer; not to mention they are the top mobile phone manufacturer. :) Intel and Samsung both have deep pockets and various revenue streams which allows them to fight a long term war. By contrast, I think fabless semiconductor companies focused on mobile market, such as Qualcomm and MediaTek, don't have that option; if they fail in short and medium term, they fail for good.
If you want to define term "Fallen" in terms of the SoC market for handsets. Then Nvidia and Intel both certainly has fallen!. ie zero to negligible market share. They might have other business for their SoCs but in phone handsets, they are basically OUT!.
I am much more interested in Modem, Qualcomm, HiSilicon, LG, Nvidia Icera, Mediatek, Broadcom ( even though they decided to quite the modem market ), VIA, Intel, etc.
" We've been saying "wait for the next Atom update" for a while, but on paper at least Cherry Trail looks very promising – it's the first Atom to ship on Intel's latest process technology without waiting a year or more,"
Hmm, we shall about that shipping without waiting a year or more.
Since even Intel's MAINSTREAM (quadcore) processors are currently not scheduled to hit 14nm till about eleven months from now, and since (assuming there aren't more of the mysterious and unexplained delays that are hurting Broadwell) Skylake will at that point be close to shipping, it seems unlikely that good old "negative profits for the past six years and unlikely to change soon" Atom is going to see that limited 14nm fab space any time soon.
Browsers don't run Java and EcmaScript is the only supported scripting language in browsers still when it comes to standards. So whatever you code in needs to compile down to EcmaScript.
the only reliable cpu worth using is the android ffmpeg beta app https://play.google.com/store/apps/details?id=com.... and its interesting that no mainstream site including this one does not bother to run these tests to collate the overall increased performance of the arm products today....
you really need to make a generic script to test with ffmpeg asap anandTech
I may be a fan of Anand, but I am losing interest in the comments section, PLEASE fix it!!!! Give us an edit button, and let the reply show up underneath what we are replying to. My comment above was directed at the guy saying Mediatek is faster than Snapdragon 800 because of a sunspider test....
Apple is the one who continues to impress me, their A7 is almost about to be replaced with the predictable A8 in the next gen of hardware, and yet for single threaded performance I think it's still taking the ARM crown. Sure, others have higher total performance using all four cores, but that doesn't help you in most single to dual thread bound situations you're likely to see on smartphones. If the A8 is still a dual and doubles performance again, I'm going to be mind blown.
It gives a rather nice view as to what is there but that only serves to whet my appetitve for the WHY?
E.g. why is Mediatek so successful?
And my guess is: Because they are cheap!
And while that sounds like an easy answer, if cheap was all that easy, everybody would do it.
So how can they be so much cheaper?
And why doesn't it mean their product is unacceptably bad?
Transistors on a chip carry a certain cost and that cost isn't zero nor is it vastly different for the Chinese manufacturers, than for anybody else.
And 8 "silly" cores might still take up just as much space as 2 Apple A7 cores and on certain multi-core friendly benchmarks even deliver similar throughput.
So what's the major difference?
ARM licensing cost, I believe.
What follows is mostly conjecture, but I hope the general idea is valid so please be tolerant if I'm wrong on the details!
ARM needs money to keep them going and if I have understood the nice licensing article posted here some time ago correctly, they charge top dollar for top performance. Charges keep dropping over the life time of an architecture (e.g. A9 no longer being the top dog) and start already low for low power.
Thus the licensing fee for an A15 core is a lot higher than an A9 these days and a Mali T6xx a lot more than a Mali 4xx... ...but only for the first core on any chip: Any additional core of the same type is peanuts.
Likewise for GPU blocks: A Mali 400MP4 costs very little beyond a Mali 400MP1
The "little" CPUs in the BIG/little pairs A15/A7, A57/A53 etc. cost a lot less in licensing than the BIG brother, again, because ARM charges for performance and basically throws in the little brother for free. In a way the "little" one is the public admission they can't get the BIG one energy efficient enough, which is a bit of an embarassment, after all (I know it's actually a very "clever" solution to the silicon physics problem underneath!)
Now you can't combine ARM CPUs and GPUs in any way you like: First you'll have to match generations!
You can't have Mali 400 on A7, A15, A53 or A57: It would have to be a 6xx (or 7xx) Likewise you can't have Mali 6xx or 7xx on A9.
You'll also have to match CPU cores and GPU "cores" or rather blocks with certain limits of 1-4. Generally it means, if you want more GPU blocks you'll have to throw in more CPU cores. Generally it also means (just to remind you), those blocks and cores will cost very little in terms of ARM licenses. Generally it also means, those blocks and cores will cost transistor real-estate and power when used. Generally even eight A7 cores will use less than 20% of the SoC real-estate: ARM cores are tiny. Generally die-area used for GPU blocks will translate into power until either the DRAM bandwidth, TDP or the battery run out.
Now put all of that together and you have a perfectly good understanding, how Mediatek, Allrunner and all these other clever Chinese manufacturers manage to beat conventional wisdom and the competition:
Sure eight A7 or A53 cores seem "silly" until you're clever enough to realize, that you'll get all eight of them at budget licensing cost from ARM because the A7 and A53 are considered "little". At the same time you'll get access to the BIG T6xx and T7xx Mali GPUs, which might already be cheaper when there is no A15/A53 and in any case only the first block costs real money and with eight CPU cores you can add lots of GPU blocks.
The result are SoCs with "average" scalar CPU performance, but relatively beefy graphics power, which happens to be exactly where power on mobile devices happens to be noticed.
On the other hand it's almost as hard to find any use for a 4GHz x-issue out-of-order 0.6 cpi CPU on a smartphone as it is for a 2GHz A7 octa: It won't help you text faster or do whatever you do with your phone today.
We have reached a certain compute power saturation point, where twice as much power doesn't provide twice the value.
And at that specific point that smart Chinese selection allows to tune the only elastic parameter in the cost of manufacturing a smart phone with sufficient power and endurance: Licensing cost.
Perhaps the Chinese aren't all that innovative when it comes to product design. But they know how to put all their ingenuity into saving cost and with that comes also the economy of scale which then helps for all the less incompressible cost items.
its pretty simple really, ARM inc have provided an interconnect and related IP to allow a generic 4x4 cluster configuration with a single cluster being up to 4 cores, and now they also provide a 4x8 and above cluster interconnect configuration up to 2 Terabit/s per Cache Coherent CoreLink™/AMBA Interconnect for the more demanding SOHO/SME and server class of devices....
so it makes sense to use as much as that provided interconnect as a SoC growth margin etc...
the problem is that currently arm OEM's don't seem to have any way to bring that massive interconnect capability to the outside world where the generic available bridge chips (USB3 to pciE etc) can be interfaced to the core interconnects and gain full data throughput or add things like the existing DDR3 MRAM until ARM incorporate their existing MRAM licences into their future SOC IP etc... and that's holding everyone's progress back
Forget this SoC talk... I want to know where that white Nexus 5 came from! Did you tell the photographer "show me an SoC on a smartphone"? And why doesn't it have a cut-out for the notification LED?
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
94 Comments
Back to Article
dishayu - Tuesday, August 19, 2014 - link
Good read.No word about MIPS based processors? Ingenic's Xburst2 should be out anytime now and they sell a reasonable volume in Chinese and Indian markets.
ZeDestructor - Tuesday, August 19, 2014 - link
There's just no interest in MIPS anymore. It's all ARM or x86 in the "popular" SoC/CPU categories, and as large as the Indian and Chinese markets are, the ARM cores largely dominate in both cases.tuxRoller - Thursday, August 21, 2014 - link
Img is going to be releasing products with the new high-end mips procs. So, I wouldn't say its dead...just dreaming:)stucktrader - Monday, September 1, 2014 - link
The days of Pentium, Alpha, PowerPC, MIPS, Sparc, AMD etc...that was a really exciting time regarding CPUs...
Now all this SOC talk reminds me of those days.... really fun times technology wise...
LiviuTM - Tuesday, August 19, 2014 - link
Actually, the top Mediatek processor shipping today in a phone is MT6592, an octal-core Cortex-A7 with Mali-450 MP4.And MT6595, a octal-core with for A17 + four A7 and a 2 clusters PowerVR 6200 seems just around the corner.
Cheers
Liviu
Rdmkr - Tuesday, August 19, 2014 - link
Was about to comment on this; mediatek is not quite as diminutive as the initial remark suggests. Their current flagship SoCs already match the snapdragon 600 on several benchmarks and the cortex A17 one on the way could seriously compete with Western premium SoCs with a little luck.LiviuTM - Tuesday, August 19, 2014 - link
Indeed, MediaTek is pushing huge volumes in Chinese market.They still have a lot to work before catching Qualcomm, as a SoC is a lot more than CPU and GPU alone, but they are unbeatable on price.
Chinese manufacturers are flooding the Chinese market with cheap smartphones using MediaTek SoCs; they recently started to establish a foothold in European markets as well.
LiviuTM - Tuesday, August 19, 2014 - link
This article is focused on mobile SoCs, where AMD is has a negligible presence. And if you look only at smartphones SoCs, AMD doesn't even cater for this market.LiviuTM - Tuesday, August 19, 2014 - link
Sorry, this comment was meant for @yannigr2JarredWalton - Tuesday, August 19, 2014 - link
Sorry -- somehow I missed the hexa and octa MediaTek options. They're still looking like Exynos levels of performance at best, which isn't the fastest chip around, and octa-core Cortex-A7 just seems... silly. There, I said it: I can't think of many use cases for eight of the same core in a smartphone or tablet. In fact, I think having two really good cores is still a better approach, though there are edge cases where quad-core can still prove useful.Anyway, MediaTek is huge in terms of volume, I agree. That's not even a question. But the MediaTek devices still tend to be cheap, budget offerings. The MT6595, MT6752, and MT6795 will improve performance in some areas, and for $200 devices they should be decent overall, but in most respects the SoCs are still a generation behind.
jjj - Tuesday, August 19, 2014 - link
8 cores is silly but Qualcomm is doing it too with A53. It's silly and it's just marketing but it is what it is.And a generation behind what? Behind the Qualcomm A7 or behind teh fastest SoCs, how many generations are other behind? Intel hasn't even integrated 3D. And the market is cheap and why would anyone sane in China pay 830$ for an iphone when they can get a decent 100$ phone or last year's flagship at 230$? And why would enthusiasts be that wasteful and inefficient?. Are we afraid to get a little bit dirty now? It feels like Anandtech got stuck in a past that is notthing like the real world anymore. How much coverage does Intel gets in mobile and what have they done so far? The US mobile market is just dead because of the carrier monopoly but there is a big world out there with lots of brands and lots of devices. And the next giants are out there too and the next few years will be very bloody with prices tanking and more of the old guard being killed off.
But you guys are just missing it all.
JarredWalton - Tuesday, August 19, 2014 - link
The upcoming MediaTek chips look quite similar to the existing big.LITTLE Samsung Exynos chips, that's what I'm getting at. Okay, so A53 is coming, but the MediaTek has traditionally gone light on the GPU side of the equation, and that doesn't really look to be changing. 2-cluster Rogue is already half of the A7 performance (and that's if they hit equivalent clocks), and 4-cluster Mali-T760 isn't exactly going to set the world on fire either. This is why I say they look like last generation products more than current generation.LiviuTM - Wednesday, August 20, 2014 - link
Hi Jarred, MediaTek is not going to held the performance crown anytime soon, that's for sure. I'm pretty sure high end devices will continue to be based on Qualcomm chips, with the exception of Apple's and, to some extent, Samsung's.However, the big question is whether their chips will be good enough for medium and even medium-high devices in terms of both CPU and GPU performance. As far as I remember, low and medium smartphone market is growing more rapidly than high-end, at least in terms of volume.
In my opinion, 2 major MediaTek's weak points are:
- poor software & developers support - I believe they don't publicly release source code, kernels etc.; this is the reasons you don't see custom ROMs for phones using MediaTek SOCs, as opposed to Qualcomm and Samsung (Exynos) powered models
- poor GPS performance - it really sucks compared to Qualcomm's
Regards
Liviu
melgross - Wednesday, August 20, 2014 - link
Your own prejudices are showing here. Perhaps not everyone thinks as you do.fteoath64 - Sunday, August 24, 2014 - link
An Octa-core with Big.Little is technically a quadcore chip unless it is all A7 in the eight cores which is actually silly. In China, people are practical and do buy low-end devices at low prices just to be able to use them. iPhones are status symbols for the rich and that is about it. Apple cannot make much head-way due to high pricing and huge competition in the mid and low ranges of phones.I am surprised that manufacturers/designers are not making TriCore chip which are purposed such that the third is a Little core dedicated to I/O onlywhile the two large cores can switch up and down in speed or off if not used. This lessens the context switch needed for Big.Little design to a large extend thus saving power while performing close to a quad practically. Nvidia's 5th power saving core is an example of such implementation but there was no dedication to I/O processing as such.
saliti - Tuesday, August 19, 2014 - link
I don't think you know what you are talking about. Have you ever analyzed any of the MTK chips you are talking about or are you just speculating.BMNify - Wednesday, August 20, 2014 - link
"octa-core Cortex-A7 just seems... silly."you try playing any high profile high bit rate 1080p or better with a dual or lower quad core and anything that uses FFMPEG for the audio etc and you soon see the need for an octacore that ffmpeg/vlc and x264 and downstream apps can use to good effect without stuttering
its even hard for lan streaming your DVB-t2 content to your browser too, a good quad or octacore is the way to go for a purely tech POV...
Zoomer - Friday, September 5, 2014 - link
How many minutes would the battery last with the cores going at full blast? A octacore going full on would likely thermal throttle in most designs rather quickly anyway.These tasks need to be offloaded to a dedicated DSP.
Jumangi - Wednesday, August 20, 2014 - link
If that is your benchmark for getting mentioned then Intel has no business on this list either. Regardless of their overall size and talk Atom has gotten zero tracking in the mobile space. As the article itself likes to say its always "wait till X platform is release next year" with them. Meanwhile ARM becomes more dominant every year.fteoath64 - Sunday, August 24, 2014 - link
@Jumangi :"If that is your benchmark for getting mentioned then Intel has no business on this list either. Regardless of their overall size and talk Atom has gotten zero tracking in the mobile space."Totally agree!. Intel is just using the WRONG tech to fight the Arm competition. IF Intel has any sense they ought to be jumping on the RISC-V initiative which actually claims HALF the power consumption of Arm for a given performance level. It is this kind if advantage that x86 can NEVER get to unless Intel has 7nm process node!. Of course Intel can get an Arm license and do their own implementation. But "Not Invented here" syndrome is rampant in Intel. That is for sure. So getting into a competing architecture is the ONLY thing worth doing to counter the Arm attack so far.
So Intel: Are you "Blind or Stupid ?". I am pretty sure they are not stupid since they are still around ....
stucktrader - Monday, September 1, 2014 - link
AMD missed the boat... and to a certain extent INTEL...name99 - Tuesday, August 19, 2014 - link
"already match the snapdragon 600 on several benchmarks" is a meaningless claim.Obviously SoCs that are fields of wimpy cores can do well on a certain class of throughput benchmarks, but that doesn't translate into any sort of value for a cellphone or tablet.
(They could be valuable for a certain type of server, except there are enough other issues there, starting with 32 bit, that that's also a non-starter.)
As for the A17, well, my vaporware can beat up your vaporware. I assume the A57 based Snapdragons will ship around the same time the A17 Mediatek's ship.
vortek - Tuesday, August 19, 2014 - link
It's a bit strange to me that the article skipped Microsemi, whose "SoC" division -- originally Actel -- created the first FPGA with a built-in ARM processor + micro controller system + 50MHz oscillator + DDR controller + CAN bus + I2C + RTC + NVM. The FPGA itself is built out of flash cells so it doesn't lose its configuration on power loss, and is architected for security, low power, and immunity to SEU.londedoganet - Tuesday, August 19, 2014 - link
First of all, nice to see that you're so loyal to your company, coming to post product spec sheets on a public forum! ;-)Second of all, it must be noted that the manufacturers focused on here are ones which are catering to the smartphone/tablet market. Microsemi's offering seems more targeted towards industrial applications, and has features that aren't as useful in a smartphone (FPGA, CAN bus).
ddriver - Tuesday, August 19, 2014 - link
Obviously the article concerns SOCs that are put in mainstream mobile products. Not exotic development boards. There are plenty of those around.BMNify - Wednesday, August 20, 2014 - link
thats all well and godd remembering the old innovators but you forgot to mention and first people to create todays tech that being Ferranti International plc, the UK electrical engineering and equipment firm that operated for over a century from 1885 until it went bankrupt in 1993.among the related tech they produced were the With the invention of the laser in the 1960s the company quickly established itself in the electro-optics arena. From the early 1970s it was delivering the Laser Rangefinder and Marked Target Seeker (LRMTS) for the Jaguar and Harrier fleets, and later for Tornado.
In collaboration with the University of Manchester they built a new version of the famous Mark 1, Some of the technology of the FP 6000 was later used in its Ferranti Argus range of industrial computers which were developed in its Wythenshawe factory. The first of these, simply Argus, was initially developed for military use.
the Semiconductors they produced where the forerunners of the FPGA In the early eighties, Ferranti produced some of the first large uncommitted logic arrays (ULAs), used in home computers such as the Sinclair ZX81, Sinclair ZX Spectrum, Acorn Electron and BBC Microcomputer.
now when Microsemi have a current ARM cortex SOC with octacore ,x4 SATA,x4 micro PCIe and at least x2 1gigabit ethernet in an ODROID form factor AND price point , then we will talk...
what is it with all the ARM oems today limiting their SOC to only phone/tablet specs and noting Quad/Octacore in capable for the SME NAS/SAN range!
Jaybus - Sunday, August 24, 2014 - link
Most so called innovations are incremental improvements of existing technology or combining existing technology in a new way. There are very few paradigm shifting discoveries, such as the laser that you mention. These things take time. After all, Einstein gave the theoretical basis for the laser in 1917.Power envelope is the limiting factor right now, and thus the settling for "good enough". Much of the power used is due to i/o. I expect the next big step to be "new" i/o methods. In particular, on-chip silicon photonics has the potential to drastically increase both the speed and efficiency of chip-to-chip i/o, perhaps even inter-chip i/o, freeing up more power for compute operations. At that point we will define a new level of "good enough", since human wants and desires are endless and we will always want more. ARM is limiting their SOC based on power envelope. Raising that power envelope would pit them against higher performance competition. Intel is trying to enter ARM's low power niche. At the same time, ARM is trying to enter Intel's high performance niche. Neither have yet reached their goals, mostly due to physical limitations that will eventually require replacing electronic interconnects with something else.
R3MF - Tuesday, August 19, 2014 - link
is hope that qualcomm 410 forms the basis on the next Jolla phone.jjj - Tuesday, August 19, 2014 - link
Where is the Snapdrogon 410 in shipping devices? Some showed up in benchmarks but there isn't a single device with it already announced as far as i can tell and i do pay attention to emerging markets since they define the market now more than Samsung or Apple do.For the 810, during their latest results call, they said it will launch late this year, if launch means shipping or devices in store no clue really but with Samsung shipping 20nm parts,they might be in a hurry.
On Mediatek you guys really got to get more informed and it seems you need to reset your definition of first tier phone makers. Folks like HTC, Moto , Nokia, Blackberry and even Sony are fading away and Mediatek has wins with some of those too. They got almost 30% of the smartphone market this year, you can't keep ignoring them. And in revenue they are bigger than AMD and Nvidia and vast majority of semi companies ( now that they bought MStar).
They are also just starting to sell discrete LTE with existing chips and about to launch their quad A17+quad A7 with integrated LTE, the MT6595.
And can you really claim they cut corners or that they are less than others when you guys have reviewed 1 single device with Mediatek- as far as i can remember that single device was the Asus Memo pad HD 7. And you never report on them at all. If you ignore them and belittle then it should be on facts not popular beliefs- if we go by that , Mediatek is from China not Taiwan.
If you ask me Mediatek is very likely to go for a custom core of their own soon too, it makes sense and they are big. They are about to reach 2B revenue in a quarter, they have gross margins just under 50% and R&D budget above 19% of revenue so they actually are on of the few that can make a custom core and makes sense to go there or even buy someone surprisingly big,You like it or not, they are the only real competition to Qualcomm right now and a serious competitor for that matter.
The Allwinner A33 is new i don't think you can see it often in budget tablets as you are saying. I've seen a single tablet with it so far, ofc there could be a handful of them but the thing is new and products with it just starting to show up.
Rockchip RK3288 is in a bunch of tablets like the ifive mini 4 , Pipo P1 , P1, P9, P8 , Teclast P90HD , Chuwi V88 and my list is a couple of weeks old.Not all listed hit retail yet.
2 names that you've missed although they are more significant than most names mentioned are Spreadtrum and Marvell , got the feeling Marvell is working on a custom core too and they are one of the very very few that has been shipping LTE for a while now. Broadcom is also not quite out yet.
Anyway ,it might be time to look at SoCs in far more detail ,not just the CPU, GPU and connectivity and not even just the SoC. You've got a bunch of other compute units and likely more will be added , while companion chips matter too when it comes to price and the financial aspect decides who makes it or not.
LiviuTM - Tuesday, August 19, 2014 - link
Actually, I think Broadcom is out of mobile SoC business for good.No mobile SoC on their site, will close the modem division http://www.eetimes.com/document.asp?doc_id=1323198
jjj - Tuesday, August 19, 2014 - link
Phones with Broadcom are still launching, just today i've seen one announced in India and that's why i said they aren't out quite yet nut yes you are right they are on their way out (unless something changes).jjj - Tuesday, August 19, 2014 - link
but* not "nut" lolLiviuTM - Wednesday, August 20, 2014 - link
They are on the way out, so to speak. You can imagine the exit will be gradual, as they have prior obligations to fulfill. Supply chains deals are made with months or years in advance.Penti - Tuesday, August 19, 2014 - link
Modem business is definitively over, despite having a ready LTE solution. They plan (or have by now) to let staff go in Finland. Renesas/Nokia-LTE team is no more. Doesn't mean Broadcom still won't deliver other products to the mobile market though. They have WiFi, GPS, ARM-SoC's with custom GPU/video-processor for different markets. Even if mobile stuff will get no new app processors stuff like STB's should get new chips. It's 2500 people from the cellular modem divisions they are firing.LiviuTM - Wednesday, August 20, 2014 - link
From what I see, Broadcom is leaving mobile (phones/tablets) for good and is focusing on other markets.There's no mention about mobile SOCs on their site, as opposed to Marvell.
ArthurG - Tuesday, August 19, 2014 - link
custom Mediatek ARM core ? don't think about it, they don't have the right guys to do that. They are only hardware assemblers, putting IP blocks in their SoCs. Their support is nonexistent and their software stack/team is pathetic...LiviuTM - Tuesday, August 19, 2014 - link
You're right, they have a lot to do before even thinking to get into the big league.They software support is really poor, just started to offer Android 4.4 for their SoCs and only for octal-core models, if I'm not mistaken.
However, they are on the rise and who knows what future may hold...
saliti - Tuesday, August 19, 2014 - link
No, KitKat is already available for MTK 6572, (dual core A7, 1.3 GHz) 6582, 6589 (Quad core A9) etc.saliti - Tuesday, August 19, 2014 - link
I meant Quad Core A7Nik_33 - Tuesday, August 19, 2014 - link
And how is that they are the only one with 2.0 Ghz A7 cores?They are also offering KitKat for most of the SoCs - check all the Lenovo phones...
LiviuTM - Tuesday, August 19, 2014 - link
A7 is weaker and much smaller that A15, search in previous AnandTech articles, you'll find more detailed info about that.And by looking on Lenovo site, there are only a handful of phones with KitKat:
http://shopap.lenovo.com/in/en/smartphones
Qualcomm has a lot of expertise in designing custom cores, both discrete and integrated communication modems, memory interfaces and knows a thing or two on how integrate all the blocks into well rounded SoCs; is not by accident they are the leading smartphone SoCs vendor, Currently, even the behemoth which is Intel is struggling to compete with them; this can give you an idea about the engineering expertise Qualcomm owns and MediaTek is still far from touching them in this area.
Cheers
xinthius - Tuesday, August 19, 2014 - link
You're wrong. I suggest you look again, LiviuTM.xinthius - Tuesday, August 19, 2014 - link
In fact, no. I will quote Anand. "The dual-core A7 is now the fastest SoC we've tested under SunSpider, even outpacing Qualcomm's Snapdragon 800 and ARM's Cortex A15"Don't post utter rubbish in the future.
kron123456789 - Tuesday, August 19, 2014 - link
LoL. LiviuTM are talking about Cortex A7, you are talking about Apple A7.LiviuTM - Wednesday, August 20, 2014 - link
Correct :)Apple A7 is a custom ARMv8 (64bit) dual-core design, in a class of its own, for those who don't know. In fact is the first shipping ARM 64bit SoC in a mobile device, which no small feat; it took everyone by surprise, as AnandTech mentioned in Iphone 5S review.
That's why Qualcomm is using standard ARM cores (A57, A53) in their first 64bit products; they have to come up fast with something, but their custom 64bit cores (which I'm sure they are working on) are not ready yet for prime time.
Cheers
Liviu
BMNify - Tuesday, August 19, 2014 - link
@Xinthius: ROFLMAO, you are confusing Apple A7 with ARM Cortex A7, its very clear that @Liviutm is quoting ARM Cortex A7 as generally A7, A8, A11, A57 etc are referred to standard ARM nomenclature and the Apple's stuff is labelled as Apple A6 swift, Apple A7 cyclone etc.. The only reason Anand has mentioned as A7 in that sentence is because you are quoting his lines from an iphone review!!jjj - Tuesday, August 19, 2014 - link
You do understand how big they are? If you exclude NAND and RAM makers , only Intel , Qualcomm and TI would be bigger than them by next year.They are about to overtake , in revenue, folks like Broadcom, Renesas or STMicro and they could actually afford to buy someone like AMD but it doesn't make sense for now (would be too costly for what they would be getting).As for how far they are from it, it remains to be seen, just because we don't know about it ,it doesn't mean they haven't been working on it but it certainly makes sense for them to go there and they have resources to do so. If AMD and Nvidia can while being smaller ,they can too and given their share it makes a lot more sense. To take more and more share from Qualcomm they kinda need to grow up.
name99 - Tuesday, August 19, 2014 - link
Your enthusiasm for Mediatek is noted, but you seem to be out of touch with reality.Mediatek ships lots of parts, yes, and that's valuable to the poor people of the world, yes, but they are in no way generating new ideas and progress in SoCs, at the logic level, at the CPU level, at the SoC level, at the SW level. Their annual revenues are $4.5billion and their annual net income $.9 billion. To put that in context, Apple still has (after handing out oodles in share buybacks and dividends) about $120 billion of cash on hand.
Mediatek is not interesting to 90% of the people who read this blog because no interesting technology is going to come from them.
It's not interesting to the other 10% because no interesting business upheaval is going to come from them. They'll continue to provide cheap chips to struggling Chinese manufacturers who are all on the verge of bankruptcy, and nothing will change over the next few years except some mergers and name changes among those companies.
You're like a guy telling us we should get really excited about Nestle because they sell to so many people, have a presence in so many markets, blah blah. All true -- but Nestle just is not a source of the kinds of things readers who read AnandTech find interesting.
jjj - Wednesday, August 20, 2014 - link
I'm nhe one out of touch? Yu can't even convert TWD to USD or notice the MStar buy and not look at older results. Do you even have a clue what income is? How exactly is income higher than revenue?Apple is not relevant in the SoC space sine they are the only ones using it and they are utterly uncompetitive when it comes to price. Apple will have 14% share in smartphones this year and under 10% in 2 to 4 years depending how aggressive the price declines will be around the world.
Mediatek was the first one to enable phones and tabs with a cheap quad A7 and it was quite interesting for any less snobish (or xenophobic ,w/e it might be) then you. Mediatek has has an actual wearables platform unlike some others that push phone SoCs into watches. Mediatek has lower prices because they manage to be efficient not because they take low margins. They had over 49% gross margins in Q2 and they managed some very nice chips (not just the SoC but companion chops) over the years.
As for your confusion about who uses Mediatek, i imagine you are American and you go to VZW seen 15 options and think that's all there is out there. The actual phone market is starting to be dominated by the new players that might be global, regional or local and they are gainign share and will keep gaining share. Many of those phones are using Mediatek. Go check stores like CDiscont in France, Flipkart in India , eMag in Romania or JD.com in China. Mediatek is in no way China only, it's just not US because that market is dead, killed by regulators failing to do their job and allowing carriers to murder it.
And sure the Chines OEMs are all on the verge of bankruptcy , and thousands of other around teh world ,except the few that will be the next giants, it's called consolidation. The next giants are from China and India and maybe Brazil and so on but not the US.
In PC Anandtech should review only 1k$ Intel CPUs? Is that your argument? And to hell with AMD since they suck. For better or worse Mediatek is the AMD of the mobile world.
Back t o the financials , in the current quarter Mediate will get about 2billiions revenue. In Q2 their R&D was over 370 millions USD and that's more than ARM itself in an entire year.
melgross - Wednesday, August 20, 2014 - link
Apple certainly isn't uncompetitive on price. They have over 60% of the market in the price range they do compete in. That's where all the money is being made anyway.melgross - Wednesday, August 20, 2014 - link
Apple has $176 billion in cash and investments left.Penti - Tuesday, August 19, 2014 - link
Huh, STMicro and Broadcom certainly has a larger revenue and business to build on. Renesas has been going downhill ever since Renesas and NEC merged but has similar revenue to ST and Broadcom. STMicro has lots of business, still have some fabs and lots of employees. Mediatek does have some interesting stuff outside mobile application processors and baseband, thanks to stuff like Ralink, but they are still relatively small and has tougher to access bank funds though are in a much better position than the Chinese competitors.jjj - Wednesday, August 20, 2014 - link
Broadcom has some 8B plus yearly revenue, Mediatek will reach about 2B in Q3 and they are growing way fast .The other mentioned have less revenue.
http://www.electronicspecifier.com/cms/images/1H14...
So if you exclude the RAM and NAND makers (Samsung, Toshiba, Micron, Hynix) and obviously TSMC since it's a foundry and factor in that Mediatek is very likely to have well over *B revenue next year they are behind very few.
Sure some of the NAND and RAM makers make other things too but i don't know how much revenue that is.
Mediatek does report revenue each month and an outlook for the quarter so you can check how much they made in the last months and they gained quite a lot of scale/
LiviuTM - Wednesday, August 20, 2014 - link
By any means I'm not disregarding MediaTek, they are growing fast and they surely can't be ignored.However, in terms of performance MediaTek can't touch Qualcomm - a mobile SoC is more than CPU and GPU alone - I think there are some interesting articles on AnandTech about that.
Here you can see a high level block diagram of Qualcomm Snapdragon 810:
http://telecomtalk.info/qualcomm-announces-64-bit-...
Achieving a balance between performance, power draw and heat is not so easy, but is not talked about too much; nonetheless isimportant - what's the point in having a blazing fast SoC on paper, when in real life it eats battery fast and/or gets too hot and real thermal throttling kicks in, drastically limiting performance?
jjj - Wednesday, August 20, 2014 - link
Mediatek never had a fast SoC yet and don't think anyone claimed that but since when is it just about fast ? In PC we don't care just about 1k$ Intel we care about the entire range.And if Mediatek would suck so much,some better tell us why they are doing well and beating so many other, forcing Qualcomm to change roadmaps and drop prices.
melgross - Wednesday, August 20, 2014 - link
Maybe they could afford to but AMD, but they wouldn't, even if they could. If they did, Intel would pull their license to build x86. That would end any purpose to the deal, and kill AMD.Penti - Tuesday, August 19, 2014 - link
Mediatek is from Taiwan and has grown by buying several other Taiwanese firms. Even AMD reported larger revenue last year than Mediatek at the time. No one would argue that Nvidia basically just has their GPU-chip market though.Kill16by9TN - Thursday, August 21, 2014 - link
You are mistaken. Mediatek originates in Taiwan (Headquarters (Hsinchu),No. 1, Dusing 1st Rd., Hsinchu Science Park, Hsinchu City 30078, Taiwan), not ROC.
http://mediatek.com/en/about/contact/
Other than that, I share your view about MTK. They are the new rising star, now also in quality/features and not longer only in respect to price.
What I was missing however in the whole article, is HiSilicon/Huawei's "Kirin 9xx" SoC line.
tghs - Tuesday, September 2, 2014 - link
Just to clarify: ROC is Taiwan, PRC is China.yannigr2 - Tuesday, August 19, 2014 - link
AMD just as a note in the closing comments? If you go at wikipedia article about SoC the first picture in the right is an AMD Geode! How old is that? AMD is also the only company that I know with a socketed SoC in the form of the AM1 platform.And AMD is just a line in the closing comments? OK...
LiviuTM - Tuesday, August 19, 2014 - link
This article is focused on mobile SoCs, where AMD is has a negligible presence. And if you look only at smartphones SoCs, AMD doesn't even cater for this market.tipoo - Tuesday, August 19, 2014 - link
How long they've been in the game has nothing to do with their presence in modern mobile smartphones and tablets - which is negligible. I hope for them to do well in it, but with them in so few shipping designs they're a marginal player at this point.nathanddrews - Tuesday, August 19, 2014 - link
Would you folks mind doing a roundup of inexpensive Windows (full) tablets? The Surface Pro 3 is making all the headlines right now, but there is a fast-growing sub-$300 market of Bay Trail/IPS/8-hour/Windows 8.1 (non-RT) tablets out there. Many of them are currently under $200. Seems like something that could be extremely competitive with Android/iOS, given the power and flexibility.Acer Iconia W3 ($199)
Dell Venue 8 Pro ($199)
ASUS VivoTab Note ($199)
Lenovo ThinkPad Tablet 2 ($299)
etc.
fteoath64 - Sunday, August 24, 2014 - link
"Would you folks mind doing a roundup of inexpensive Windows (full) tablets? The Surface Pro 3 is making all the headlines right now, but there is a fast-growing sub-$300 market of Bay Trail".No these are "on off" flukes since they rely on free Win8 license and throw-away Intel prices. None of these are sustainable for long. It is just Intel's attempt to claim the low-end of the market until they can churn up higher performing parts (then raise prices). Arm really had the tablet markets to themselves with Android/IOS taking 98% of the volume shipped. Win 8.1/9 will relegate to the x86 behemoths the way they always had due to legacy software dependency. With web-based and cloud based apps, the legacy dependency is going away, so it also explains why the cheaper tablets with Android is so popular since their performance no long sucks like 3 years ago.
Interestingly Intel made very little headroom in performance last 3 years except for huge jumps in power consumption at the expense of lowering cpu frequency.
darkich - Tuesday, August 19, 2014 - link
Jared, did you even do your homework for this article?Rockchip's Rk3288 with 4x Cortex A17 and Mali T760 is already out on the market.
name99 - Tuesday, August 19, 2014 - link
Hmm. Maybe he did not mention it because:“It is sold with an IR remote control, a 5V/2A power adapter, an HDMI cable, a USB cable, a Wi-Fi antenna, and a user’s manual. You can find it for $122 on Aliexpress including shipping. But before you rush to buy the device, understand that Rockchip RK3288 devices are still not really ready, more like beta devices, even Android TV boxes which are expected to come before HDMI sticks had to have their PCB re-laid out to fix some hardware bugs, and the firmware needs more work based on various reports.”
darkich - Tuesday, August 19, 2014 - link
Check out the PiPO P1 tablet.The benchmarks are really top class
kron123456789 - Tuesday, August 19, 2014 - link
Benchmarks like Antutu? Because Tegra K1 really top class in all graphics benchmarks.Penti - Tuesday, August 19, 2014 - link
Marvell is still in the business. They have LTE modems too and have announced new products. HiSilicon (Huawei) is also an upcoming player. Plenty more players outside the phone/tablet market.Not fun for the guys in Finland who developed the Renesas/Broadcom LTE-modem but basically never saw it ship. Wonder who will snap up those people. Ericsson still offers their modems (from the ST-E venture brake up) at least. To bad to see everything just wasted. Broadcom was never big on app processors in handheld devices, but they seem to continue to develop ARM SoC's for other markets.
TI really went down with Nokia early 11 as a result to basically stop making and investing in phones from Nokia. Microsoft's Qualcomm powered platform eventually grew to about 30 million devices per year for them, which is miniscule for a company that had almost 60 000 employees and made about 450 million devices 2010 including more smartphones then than later on. ST-E really went down thanks to it too, they had invested in development and had been cooperating with ST-E but had to end that when they was forced to use Qualcomm exclusively. As well as shutting down the platform they were cooperating on.
Intel at least seems committed. They don't quit on the wireless business and have made more acquisitions lately. Their LTE solutions are finally shipping.
BMNify - Tuesday, August 19, 2014 - link
As smartphones and their prices reach commodity level (already happening in markets like China and India due to Xiaomi, Huawei, Micromax, Asus zenphone etc..) the players like Mediatek will slay atleast few giants in the SOC manufacturing sector.JarredWalton - Tuesday, August 19, 2014 - link
Agreed. Samsung is really looking like the biggest risk, considering they're doing similar designs to MediaTek. But then, Samsung actually has fabs, making them harder to kill. I'm honestly not sure who will be the next big name to fall. TI wasn't all that surprising, but of the "Big Five" who looks weakest? Intel is still trying to become a bigger player, but they've got big pockets that can support Atom while they try to gain market share. The same goes for NVIDIA, though not quite to the same level and NVIDIA is doing more competitive designs IMO. The reality is that anyone who exits the SoC space at this point stands in danger of disappearing altogether in the next decade.jjj - Tuesday, August 19, 2014 - link
The difficult part seems the be the modem now so unless ARM or someone else licenses a killer baseband,that's what will kill the next one.There is this myth about costs but if you look at ARM itself they spend just 80 million + per quarter on R&D so it's not that much for some players out there. Even rumors about Google making a custom core for server don't seem all that crazy if you realize that costs might not be that huge.
As for being out for good, that's not true,there will be transitions to new form factors ( like glasses or w/e will replace the smartphone) and maybe major changes in connectivity (mesh networks maybe) so there can be room for disruption soon-ish.
There is also Project Ara that could somewhat liberalize the market and allow for smaller players to compete Maybe it could even go the opposite way of integration, How about a quad SLI phone lol?
BMNify - Tuesday, August 19, 2014 - link
Yeah, its a bit hard to predict such things but bound to happen to some major players in few years. Samsung is worried not just for their SoC business but their most important money printing business: Smartphones :)Already Xiaomi (Big Qualcomm 800/801 and mediatek customer) has dislodged Samsung as the largest smartphones seller in China with Lenovo and Huawei being other big threats and they sell quality stuff at low prices, just check out Xiaomi Mi3, all premium components at low-end prices, selling for roughly $230 in India and is sold out most of the times, hope you can contact and get a review-unit from unofficial resellers like Xiaomishop who gave gsmarena a review unit or maybe contact Hugo Barra directly on Google plus/twitter, he is very active on social media :)
Also, Micromax (big mediatek customer) in India has come very close to Samsung in smartphone marketshare and with the recent launch of Asus Zenfone 4,5,6 and Xiaomi smartphones, the sales and marketshare for Samsung will be a big problem in few years.
Funfact, Intel got a big design win in the form of Asus Zenfone series, the entire series is based on intel atom platform and is selling quite well, they sold 40000 units within 4 days of launch in India and after that they are constantly sold out eventhough they are bringing 15k-20k units every week.
Rama TT - Sunday, August 24, 2014 - link
*Agreed that companies like Xiaomi and Motorola are giving nightmares to the more established brands but that’s not to say that these brands will slowly lose all their value. For example despite reporting a serious drop in profits Samsung did manage to improve its revenue in the US and the Samsung Galaxy S5 sales in the US have been 50% more than the Samsung Galaxy S4 sales of yesteryear. The same thing was being said about Apple a few years ago but still despite losing hefty market shares most quarter Apple still manages to post more profits than any other manufacturer. Many a times market share does not mean everything profits also matter.*As I said despite profits being 25% less for Samsung they still manged to earn more than any other Android vendor in the market. In fact other Android vendors hardly earn anything. The total combined profits of Apple and Samsung are almost 105% or so. Yeah its 105% and no I am not drunk the extra 5% come into picture if you consider the losses companies like LG/Sony/HTC made a few quarters ago. Companies like Xiaomi and Micromax hardly manage to earn any profits. Take Xiaomi for example they are selling devices at almost cost price and hardly any profit on hardware sales. You can’t expect companies like Samsung to start competing at those prices because they spend billions on advertising and R&D and their major source of revenue is hardware profits.
*In my opinion companies like Samsung and LG will lose market share rather than reducing prices to the level of Xiaomi which are not manageable by them or rather which would not be enough for their expenses. I will readily accept that in future companies like Samsung will lose market share in countries like India and China to some extent but then to the market share which they are losing are to the companies who are hardly making any profit. And even though Samsung loses market share in India and China they will still manage to do well in countries like US/Western Europe/South Korea/Japan etc etc. In developed countries like US for example I don’t think companies like Xiaomi will be able to challenge the might of Samsung.
*In the US market the contract system is followed where carriers subsidize the cost of devices and the 900$ or so iPhone or Galaxy S5 over there “”appears”” like 200$ in the US. Although carriers manage to recover the cost over the 2 period contract but still the consumer has a sense of false belief that he is only paying 200$ for the device. Now in such a market if Xiaomi comes with even a 230$ Xiaomi Mi 3 people would “”believe”” its expensive. Although off contract is becoming popular in US the rule of the thumb is contract and as long as contract systems prevail which I guess they will don’t expect iPhones or Galaxy to lose any ground in US.
*On the other hand in countries like US the success of a device very much depends on the carrier relations a particular company has with a carrier because of the contract nature followed over there. Carriers don’t openly provide every device on contract. Many a times only devices which they feel have the potential of doing well are sold on contract. In case of Xiaomi as I have said before they really don’t manufacture devices in bulk and given the people of US hardly know about the brand initial demand will as usual be low and not much of the public will warm up to the device initially. Carriers know this and won’t stock the device. If sold on full retail price the Mi 3 would cost at least 230$ in the US which is 30$ less than the iPhone 5S or so and despite how good Mi 3 is its definitely no where near the 5S for the very least and even the most tech savvy wont go for Mi 3 for just saving 30 bucks. I know that overall on contract the iPhone 5S would end being expensive but still how many people over there would actually figure out this and even for contracts several theories are present. Some people say its cheaper some say expensive. But anyhow as of now and the foreseeable future contracts are to remain in US.
*Next people in US are not as poor as other developing countries. People over there earn well and can very well afford expensive phones. But still this is more of a personal belief rather than a good point.
*Next up is litigation and patent wars. US is a country with very strict patent laws and no matter how much I like Xiaomi I won’t refrain from saying that they copy Apple. In a very worse manner that too. Its visible all over their products. If Xiaomi were to ever enter US and gain traction I am pretty sure Apple is gonna sue the shit out of them. They did it with Samsung and won’t spare Xiaomi. At least Samsung was earning enough to face the 1 billion $ of fines they have seen from Apple I don’t think Xiaomi posses that kind of wealth that they can suffer litigation. What’s even worse is that Xiaomi copies Apple even more than Samsung then the litigation fee can also be equally high.
*Next marketing matters a lot in US. Both Samsung and Apple have big marketing budgets and companies like Xiaomi don’t even spend on marketing in a commercial manner let alone putting up an Ad during Super Bowl or something. Companies like Motorola LG and Sony which have less marketing budgets often end up as losers and to be truthful for the past several quarters most manufacturers except Samsung and Apple are losing market share in the US and its not because Samsung and Apple are coming up with some really innovative products. NO!!. In fact the G2 HTC One M7 and even Moto X were far better than S4 but still it all went in vain just because these manufacturers like LG HTC and Motorola did not have a marketing budget as good as Samsung.
* As far as I have seen in the electronics Industry the cheapest wins.”jo sasta hai woh bikta hai”. The TV industry and the PC industry are very good examples of the same. The TV once a cash cow for Sony is now Samsung’s field. Same way Lenovo dominated the PC and lapotop industry. But then again PC and TV are mostly individual products sold separately. For example in the TV industry you still don’t have DTH operators subsidizing the cost of TVs same way broadband providers don’t subsidize the cost of laptops even in nations like US that’s why did the once mighty and premium players fell all over the world. In short the product and services did not share any relation in the PC and TV industry.
*The smartphone industry however is different in US etc etc as long as carrier subsidies continue I don’t think its gonna be a challenging way for Apple and Samsung.In the smartphone industry of US etc you have services and products linked. In India you do not have any kind of meaningful subsidies and that’s why do I expect the branded players to lose market share. Even in China the operators are starting to cut down subsidies.
* Even if we leave all this as I said only few players are making serious money in the smartphone market.The countries like US Japan etc where high end smartphones are sold the profit made on every handset is almost triple the cost of say a 5k handset like Samsung Galaxy Star Duos. Losing market share would not mean much to Samsung and LG as long as they lose it to players who are not making money.Just remember for every China there is a US. For every India there is Canada. For every Vietnam there is Japan. Even if manufacturers lose market share in India CHina etc they still have other developed countries to look forward to.
Rama TT - Monday, September 1, 2014 - link
You are a stupid guy from India right? Look at my comment below to know the actual truth. Losing market share means nothing to Samsung. Even companies like Xiaomi and Micromax are hardly able to make any money on the millions or so smartphones which they happen to sell.fm123 - Tuesday, August 19, 2014 - link
Ultimately it comes down to IP and a bit of market luck/timing. Those with IP in CPU, GPU, and LTE have a head start on those who don't. Fab technology seems to help too. Because time to make a custom CPU is pretty long and costly, and everyone else competes on stock ARM designs at the lower price points. I'm not sure Samsung will be making a custom design any time soon, as they don't have much CPU/GPU design history.LiviuTM - Wednesday, August 20, 2014 - link
Samsung is a special case, indeed.Not only they have fabs where they manufacture RAM, NAND flash, logic chips (including SoCs - theirs and Apple's), but they are also a major display (basically they own AMOLED market) and battery producer; not to mention they are the top mobile phone manufacturer. :)
Intel and Samsung both have deep pockets and various revenue streams which allows them to fight a long term war. By contrast, I think fabless semiconductor companies focused on mobile market, such as Qualcomm and MediaTek, don't have that option; if they fail in short and medium term, they fail for good.
fteoath64 - Sunday, August 24, 2014 - link
If you want to define term "Fallen" in terms of the SoC market for handsets. Then Nvidia and Intel both certainly has fallen!. ie zero to negligible market share. They might have other business for their SoCs but in phone handsets, they are basically OUT!.iwod - Tuesday, August 19, 2014 - link
I am much more interested in Modem, Qualcomm, HiSilicon, LG, Nvidia Icera, Mediatek, Broadcom ( even though they decided to quite the modem market ), VIA, Intel, etc.name99 - Tuesday, August 19, 2014 - link
" We've been saying "wait for the next Atom update" for a while, but on paper at least Cherry Trail looks very promising – it's the first Atom to ship on Intel's latest process technology without waiting a year or more,"Hmm, we shall about that shipping without waiting a year or more.
Since even Intel's MAINSTREAM (quadcore) processors are currently not scheduled to hit 14nm till about eleven months from now, and since (assuming there aren't more of the mysterious and unexplained delays that are hurting Broadwell) Skylake will at that point be close to shipping, it seems unlikely that good old "negative profits for the past six years and unlikely to change soon" Atom is going to see that limited 14nm fab space any time soon.
fanofanand - Tuesday, August 19, 2014 - link
Sunspider tests java performance only, and very few people still use Java. Quit spreading FUD, you are obviously an employee paid to post here.Penti - Wednesday, August 20, 2014 - link
Browsers don't run Java and EcmaScript is the only supported scripting language in browsers still when it comes to standards. So whatever you code in needs to compile down to EcmaScript.BMNify - Thursday, August 21, 2014 - link
the only reliable cpu worth using is the android ffmpeg beta apphttps://play.google.com/store/apps/details?id=com....
and its interesting that no mainstream site including this one does not bother to run these tests to collate the overall increased performance of the arm products today....
you really need to make a generic script to test with ffmpeg asap anandTech
fanofanand - Tuesday, August 19, 2014 - link
I may be a fan of Anand, but I am losing interest in the comments section, PLEASE fix it!!!! Give us an edit button, and let the reply show up underneath what we are replying to. My comment above was directed at the guy saying Mediatek is faster than Snapdragon 800 because of a sunspider test....SanX - Tuesday, August 19, 2014 - link
Agree, Anandtech's comment section design circa year 2000 smells.LiviuTM - Wednesday, August 20, 2014 - link
I second you on that, man. :)tipoo - Tuesday, August 19, 2014 - link
Apple is the one who continues to impress me, their A7 is almost about to be replaced with the predictable A8 in the next gen of hardware, and yet for single threaded performance I think it's still taking the ARM crown. Sure, others have higher total performance using all four cores, but that doesn't help you in most single to dual thread bound situations you're likely to see on smartphones. If the A8 is still a dual and doubles performance again, I'm going to be mind blown.LiviuTM - Wednesday, August 20, 2014 - link
Yeah, Apple knows its way.They sure put the assets from P. A. Semi and Intrinsity acquisitions to good use.
paranoised - Wednesday, August 20, 2014 - link
Great post. Reminiscent of the musings of Anand and Brian on the sorely missed podcast. Keep it up!abufrejoval - Wednesday, August 20, 2014 - link
Thanks for the nice overview!It gives a rather nice view as to what is there but that only serves to whet my appetitve for the WHY?
E.g. why is Mediatek so successful?
And my guess is: Because they are cheap!
And while that sounds like an easy answer, if cheap was all that easy, everybody would do it.
So how can they be so much cheaper?
And why doesn't it mean their product is unacceptably bad?
Transistors on a chip carry a certain cost and that cost isn't zero nor is it vastly different for the Chinese manufacturers, than for anybody else.
And 8 "silly" cores might still take up just as much space as 2 Apple A7 cores and on certain multi-core friendly benchmarks even deliver similar throughput.
So what's the major difference?
ARM licensing cost, I believe.
What follows is mostly conjecture, but I hope the general idea is valid so please be tolerant if I'm wrong on the details!
ARM needs money to keep them going and if I have understood the nice licensing article posted here some time ago correctly, they charge top dollar for top performance.
Charges keep dropping over the life time of an architecture (e.g. A9 no longer being the top dog) and start already low for low power.
Thus the licensing fee for an A15 core is a lot higher than an A9 these days and a Mali T6xx a lot more than a Mali 4xx...
...but only for the first core on any chip: Any additional core of the same type is peanuts.
Likewise for GPU blocks: A Mali 400MP4 costs very little beyond a Mali 400MP1
The "little" CPUs in the BIG/little pairs A15/A7, A57/A53 etc. cost a lot less in licensing than the BIG brother, again, because ARM charges for performance and basically
throws in the little brother for free. In a way the "little" one is the public admission they can't get the BIG one energy efficient enough, which is a bit of an embarassment, after all
(I know it's actually a very "clever" solution to the silicon physics problem underneath!)
Now you can't combine ARM CPUs and GPUs in any way you like: First you'll have to match generations!
You can't have Mali 400 on A7, A15, A53 or A57: It would have to be a 6xx (or 7xx)
Likewise you can't have Mali 6xx or 7xx on A9.
You'll also have to match CPU cores and GPU "cores" or rather blocks with certain limits of 1-4.
Generally it means, if you want more GPU blocks you'll have to throw in more CPU cores.
Generally it also means (just to remind you), those blocks and cores will cost very little in terms of ARM licenses.
Generally it also means, those blocks and cores will cost transistor real-estate and power when used.
Generally even eight A7 cores will use less than 20% of the SoC real-estate: ARM cores are tiny.
Generally die-area used for GPU blocks will translate into power until either the DRAM bandwidth, TDP or the battery run out.
Now put all of that together and you have a perfectly good understanding, how Mediatek, Allrunner and all these other clever Chinese manufacturers manage to beat conventional wisdom and the competition:
Sure eight A7 or A53 cores seem "silly" until you're clever enough to realize, that you'll get all eight of them at budget licensing cost from ARM because the A7 and A53 are considered "little".
At the same time you'll get access to the BIG T6xx and T7xx Mali GPUs, which might already be cheaper when there is no A15/A53 and in any case only the first block costs real money and with eight CPU cores you can add lots of GPU blocks.
The result are SoCs with "average" scalar CPU performance, but relatively beefy graphics power, which happens to be exactly where power on mobile devices happens to be noticed.
On the other hand it's almost as hard to find any use for a 4GHz x-issue out-of-order 0.6 cpi CPU on a smartphone as it is for a 2GHz A7 octa: It won't help you text faster or do whatever you do with your phone today.
We have reached a certain compute power saturation point, where twice as much power doesn't provide twice the value.
And at that specific point that smart Chinese selection allows to tune the only elastic parameter in the cost of manufacturing a smart phone with sufficient power and endurance: Licensing cost.
Perhaps the Chinese aren't all that innovative when it comes to product design.
But they know how to put all their ingenuity into saving cost and with that comes also the economy of scale which then helps for all the less incompressible cost items.
BMNify - Thursday, August 21, 2014 - link
its pretty simple really, ARM inc have provided an interconnect and related IP to allow a generic 4x4 cluster configuration with a single cluster being up to 4 cores, and now they also provide a 4x8 and above cluster interconnect configuration up to 2 Terabit/s per Cache Coherent CoreLink™/AMBA Interconnect for the more demanding SOHO/SME and server class of devices....so it makes sense to use as much as that provided interconnect as a SoC growth margin etc...
the problem is that currently arm OEM's don't seem to have any way to bring that massive interconnect capability to the outside world where the generic available bridge chips (USB3 to pciE etc) can be interfaced to the core interconnects and gain full data throughput or add things like the existing DDR3 MRAM until ARM incorporate their existing MRAM licences into their future SOC IP etc... and that's holding everyone's progress back
BMNify - Wednesday, August 20, 2014 - link
MediaTek is spending a lot on R&D, got this info from today's news:http://www.financialexpress.com/news/mediatek-open...
http://economictimes.indiatimes.com/tech/hardware/...
http://online.wsj.com/article/PR-CO-20140820-90679...
LiviuTM - Friday, August 22, 2014 - link
Indeed, MediaTek in on the rise, in terms of sales and they might indeed increase spend in R&D:http://www.icinsights.com/news/bulletins/MediaTek-...
But they still have a long way to come before becoming a truly technology leader. To put that in perspective, take a look here at R&D investments of top 10 world players (note Qualcomm position):
http://www.icinsights.com/news/bulletins/Top-10-Se...
Cheers
fbennion - Friday, August 22, 2014 - link
Forget this SoC talk... I want to know where that white Nexus 5 came from! Did you tell the photographer "show me an SoC on a smartphone"? And why doesn't it have a cut-out for the notification LED?