by 
It's no secret that {complink 444|ARM Ltd.} has been kicking {complink 2657|Intel Corp.}'s butt in smartphones. But ARM is now aiming higher, trying to hit the world's biggest semiconductor company by revenue squarely in the jaw in servers, where 95 percent of the machines are based on Intel microprocessors.
The low power consumption for which ARM processors are famous, and which made them the processor of choice for mobile devices, has become more attractive to server manufacturers as datacenters continue to struggle with the heat generated and the power consumed by traditional server hardware. The situation will only get worse as cloud computing and big-data applications create the need for bigger and bigger server farms.
A series of announcements at ARM's developers' conference in October showed just how serious ARM is about getting into the ring. First, it announced a new interconnect technology for multi-core servers that it claims can deliver up to one terabit of usable system bandwidth a second. Then it launched collaboration with Red Hat and Applied Micro Circuits to develop a 64-bit design platform that could “dramatically lower the total cost of ownership of cloud computing, data centers and enterprises.”
As part of the announcement, Applied Micro described its Micro X-Gene “server on a chip,” aimed at the big-data and cloud server market. And Red Hat pledged to build support within the Fedora community for ARM's 64-bit architecture. That was followed by the introduction of ARM's Cortex-A50 chip, a 64-bit processor aimed at networking, server, and high-performance computing.
There are enough chip and server companies promoting ARM-based processors for so-called microservers (basically server boxes that can hold more and more processors and yet keep energy consumption low) that the press is starting to talk about the datacenter as the next chip battleground. Semiconductor startup Calxeda, originally funded by ARM in 2008, has been promoting the idea ever since and introduced its first chip, the 32-bit EnergyCore SoC, in November 2011.
Marvell has also licensed ARM for a server chip, and in August said it has teamed with flash memory vendor SanDisk Corp. on a 32-bit server reference design incorporating SanDisk's flash-based solid-state-disk modules. Also in August, chip vendor Cavium got a license to ARM's 64-bit v8 server architecture. And in November, AMD said that it would begin shipping 64-bit Opteron server processors based on the ARM architecture in 2014.
System vendors, including Dell, Hewlett-Packard, Penguin Computing, and Boston Ltd., either have introduced ARM-based servers or are working on designs. Until October, most of the activity had been around ARM's 32-bit architecture, but 64-bit systems are what's really needed for the enterprise market. That's why ARM's announcements in October were particularly significant.
Meanwhile Linaro, a non-profit engineering organization that develops open-source software for the ARM architecture, says it's working with a group of companies to accelerate development of Linux for ARM servers. The group includes AMD, Applied Micro Circuits Corporation, Calxeda, Canonical, Cavium, Facebook, HP, Marvell, and Red Hat.
Intel is fighting back. In December it introduced its S1200 Atom processor, which it called the world's first low-power, 64-bit server-class SoC for microservers. “The data center continues to evolve into unique segments and Intel continues to be a leader in these transitions,” said Diane Bryant, vice president and general manager of Intel's datacenter business. “We recognized several years ago the need for a new breed of high-density, energy-efficient servers and other datacenter equipment.”
Apparently, this is not Intel's first foray into the microserver space. According to one news report, Bryant said Intel was responsible for “launching the [microserver] product category” in 2009. But interest in microservers seems to have taken the company by surprise. In a March 2011 fact sheet, Intel said it didn't expect microservers to make up more than 10 percent of the server market.
Judging from the number of chip and system vendors jumping into the microserver market, I bet it's going to be much bigger than that. With the PC market dying and the mobile phone market becoming saturated, it'll be interesting to watch these chip giants fight over it.
These are exciting moves from ARM in their fight against Intel. I am curious what can be the best definition for 'microserver' in this case? Are these products aimed at home users?
Maybe microserver is a low-power, low cost architecture based server focused towards small enterprises. I doubt whether its focused towards home or individuals.
But i do not agree with the statement that smartphone market is saturating.
You are right on their target towards SME but i dont really think microserver architecture would be low as such. It is just a reduction in racking/rack size and hardware components.
From the industry this analyst covers intersection of x86 and ARM with a current micro server focus. Currently ARM 32 bit is an evolving innovation for adopters on performance per watt that does offer a power reduction without performance penalty for the right applications; NAS, small work group, home server and commercial slim workloads.
Right approach includes independent VARS who are software system integrators interfacing with system developers like Boston and Cogent too provide commercial software systems integration addressing customer's working specification. These are not yet off the shelf solutions despite 32 bit OS and applications availability so custom software-system integration is the key.
32 bit ARM with floating point acceleration from TI is becoming available for data analytics. Where again the channel entry point is through system software developer teamed with the platform system producer.
64 bit ARM now under development is currently an interesting option for data center innovator adapters on the potential of working with ARM silicon and system design producer for platform input, resulting in platform design for use. Recall I said innovator adapters with all the prerequisites of co-development that description implies.
While there are examples of compact self contained servers, from MARVELL and Mitac, the idea of low power serves is density in full rack; low power = high density. When combined with switch in SOC to enable multi core communications across switch fabric; in VM mode is aimed to resolve Xeon utilization issue and to knock the switch out of the rack equation for cost/price.
On Intel v ARM, currently there are no suitable low power Intel offerings available with the first scheduled mid 2013 that is octa Atom. Presently E3 1220v2 dual core offering is discontinued unavailable through channels. New S1200 Atom a dual core is performance limited to toy server.
What's important to understand is this market is not about single processors on board or even four quads on board that do exist for Calxeda and Marvell today This market is about many ARM multi cores on blades that can achieve performance per watt advantage into current Xeon E5 24/26xx and higher price rungs.
Mike Bruzzone, Camp Marketing
I don't think so. Home users don't have a concept for how to use servers.
I think it might be used to consolidate in small businesses.
Yes a reduction in rack size, but also increase in the density of compute nodes per full rack, plus the consolidation of hardware components into silicon for switch fabric.
The first ARM server from ZT Systems designed by Phytek including discrete ECC was introduced November 2010 on ST Spear 1310; SOC discontinued 2011. Like many nascent ARM platform implementations there was no OS. Was aimed for home Network Appliance; flicks, photos, files.
Currently there is Network Appliance example from Huawei Systems Group, based on Marvell Armada, chosen over Intel Atom for performance and low power.
Yes commercial server is the aim targeting Xeon product and price voids. Think many ARMs on blade meaning focused applications and system management.
Today for 32 bit ARM moving into 64 bit (think validation site preparedness) there is a green field business opportunity, for VARs and Consultants engaged in software system integration; Canonical, Red Hat, others. Where application system management written for the type of work loads ARM platform's can do well on v Xeon is a unique and differentiated business opportunity.
There is also a green field opportunity for benchmark authors that are not Intel SPEC or Intel Hadopp. Rather ARM SPEC and ARM Hadoop on these platform operating system and management workloads that ARM platform's do well on aimed toward 64 bit.
Focus on what ARM does well on v Intel is a unique applications business in the new computing paradigm. Develop to create an ARM server paradigm on unique application requirements, addressing sales requirement, that is not the typical Intel requirement.
ARM server is a viable high margin business.
Mike Bruzzone, Camp Marketing
Targeting multi-task industrial applications with 16bits and 32 bits processors are other alternative to explore. Even with Internet -of – things, cyber physical systems/Real time OS and M2M massive markets are emerging for them to explore. But if looking at these areas, what about Microchip's PICs which also capturing other complex applications?
This has a direct link with the other so the war will definitely spread to its sub branches.
Yes of course. The pointer – markets are open and the game would probably be won by the best knight movers.
Doug, the “Front of the house” uses for extended RFID you point out are the exact uses that will raise the greatest objections from the privacy advocates.
I belive greater benifits will be gained in “Back of the house” uses. Exended RFID technology could allow for faster unloading and loding of product by increasing the read zone entering and exiting a dock and improving the quality of the reads of fast moving product.
It could also allow reading of product tags through a trailer wall allowing a whole truck to be read leaving or entering the yard.
Warehouse cycle counting and inventorying would also benifit by allowing an auditor to perform an inventory from the ground without assitance of lift equipment.
The battery operated (i.e. active) tags do extend the range of applications in RFID. However, because of the ease of access to user/consumer data, the advertisers will be bombarding us with more and more ridiculous promotions and sales that we are not interested in. I also agree that the privacy advocates will be furious when they see the true potential of active tags.
As long as the applications stay within the supply chain, there will be no problem. However, if this extended range gets into our everyday life, it will bring more problems than good use in my opinion.
@Cryptoman…you are right. Of course there is no way that this technology will not be exploited by marketeers, spammers, and Homeland Security. Pandora's box has been opened and now we will continue to experience greater distortions of good technology gone bad through applications designed to invade our personal lives.
@All…does anybody know how or where to join a privacy advocacy group that has some real clout? I know there are individuals writing books, but who is making the big noises in Washington DC?
Alexander, though it seems a bit futuristic but i will not bet that it will not happen in next 2-3 years as this will immensely be focused towards customer loyalty. What i think will happen is that there will be apps and programs that will integrate the technology with smartphones instead as they are ubiquitous. Moreever, data analytics companies will be looking like hungry crocodiles for this opportunity.
@Cryptoman, though i agree with you that this is just another tool for advertisers to send more ads which in most cases are more spams than useful. But i am sure that with more advanced data analytics, companies will make themselves smarter and will not waste their resources but be more focussed on their targets.
@Himanshugupta…Good point. This may reduce spamming if this becomes a ubiquitous practice. If my smartphone gets zapped with an email every time I walk by a reader, I may surrender my loyalty card. Coupons on the go. Yikes! I guess we will have to see where this moving train takes us.
@Himanshugupta: Smartphone apps are already in the picture when it comes to targeted marketing. Companies are using their apps to give discounts to loyal customers and also sending them out different promotions. Meanwhile they're able to get feedback on different surveys and polls from the customers through the apps.
Rather than annoying, I would find these targeted ads more useful. Instead of random ads advertising about products I may never use available at places I may never go to, these location-specific marketing will send me relevant ads that I can take advantage of. Of course people can always unsubscribe if they feel annoyed but I just feel this type of advertising will be more useful to most consumers.
@TaimoorZ…also a good point. I look at my recommendations on Amazon to see how close they get. Since it is based on history of purchases, I am targeted pretty well. It is a combination of push and pull marketing. They push one URL and I have the choice to scroll multiple pages. As long as I don't get one email per product, it might be OK.
Home servers may be 'disguised' into some products such as home network storage, or some products for storing photos, music, videos to be shared throughout the house.
In fact Apple's iphone Passbook app and ecosystem is trying to target this. Things such as boarding pass, loyalty cards are stored digitally in the phone and they have smart features such as when you walk near the store, the phone will notify you. Sounds great however this is only for iPhone users. I think adopting RFID for cards can bring in good benefits for users though privacy is of great concern here. How can they possibly turn off RIFD detection for their card?
I agree with the comments that smart phone apps are already coming up whereby the targeted ads, identity and all such things are implemented.
In my opinion the real use of RFID tags is not for people and related applications but for those inventories lying in heaps and stacks in the warehouses, on the store racks . RFID tags will help better management of inventory, stock taking, finding an item buried somewhere in the corner of a ware house and so on.
Any innovation in the RFID should be aimed at making these tags smaller, cheaper, rugged and smarter for such inventory related applications.
Trying to enter into personal information and targeted marketing may actually reduce the focus of using RFID tags as smartphones are anyway going to have these features as standard in the near future.
@Douglas: I agree. Rather than having individual SMS alerts from each individual store as soon as you step inside a mall, you can have one alert which gives you details of all the offers and discounts applicable on all outlets. This would make life much easier I believe. Also, as you mentioned, it could tell you about what most people like you are buying so you have an idea of what items are popular.
@t.alex: I think the next development in this area would be how to integrate RFID technology with the smartphone apps without having to rely on external RFID cards. Perhaps things like NFC technology or Bluetooth would come in handy here to control the process entirely through smartphones and not have any other RFID chips involved.
@t.alex, There is currently a case in the courts trying to resolve a privacy invasion claim filed by a student in a Texas school where the students , (4000) have been required to wear RFID badges as locators. The school says it in case of an emergency if the school gets locked down and also since the government subsidy is based upon attendance, the school wants to be able to find students who may not be in their seats but are on campus somewhere when attendance is taken. Another school in Maryland is using palm scanning as an accounting method for students who are buying their school lunches. The Texas case will be one worth following. The outcome will establish a legal precedent for other schools and public institutions.
@TaimoorZ NFC would not be able to achieve the applications discussed in the article because of its extremely limited read range. UHF RFID could work, though. There is a company in NYC that has built a loyalty program for stores based on UHF RFID. Brouha is the company name. Here is an article that talks more about what they are doing: http://www.rfidjournal.com/article/view/8943
@jmcfarland… I read the article and I think what is catchy about this particular RFID application is that the apartment/condo access cards are the same cards that double for customer loyalty. If the housing access section is secured from copy or read rights, then I can see where there would be no issues with security as far as the apartment access is concerned. Do you know if the housing access side is encrypted to avoid the possibility of a disloyal strore employee from lifting the access data for nefarious purposes?
@TaimoorZ,
What is the advantage of integrating RFID chip into a smartphone?
@t.alex: Integrating RFID chip with the smarphone would make lives easier for people as they won't have to carry the card and the phone all the time. Also, instead of carrying multiple RFID cards, you can have all the information integrated into that smartphone (assuming your smartphone is capable of writing information on the RFID chip).
@jmcfarland: The reason I mentioned NFC was because it's the only radio-waves related technology which is being built inside the smartphones. But you're right – because of the limited range NFC might not be so useful for this application. Do you reckon any other form of technology based on radio waves will be built inside smartphones in the coming future? I see a lot of potential uses if that happens.
@t.alex,
“What is the advantage of integrating RFID chip into a smartphone?”
I think that integrating RFID chip into any device without the user being able to remove it will violate user privacy and freedom. When the item is already sold, the tag should be desactivated.
@jmcfarland,
Thanks for sharing the information about Brouha. I have the impression that the loyalty program has some limitations. It is not because I visit a shop that I will necessary buy items there. Am I missing anything?
@Douglas,
It is indeed good to know if users stand any chance to fight against privacyu issues brought about by the use of RFID tagging. I don't know which advocacy group is making much noise, but I am sure that manufacturers will find ways to circumvent their actions.
@Rich, i think cost will be a problem going from passive to active as most of the cards that we carry now-a-day are passive. Also, i would not like to carry another card in my pocket.
@Hospice, i think the advantage will be for the targeted ads as store will know which sections you visited and which items you checked. So they will send you relevant offers and schemes on your smartphone or next time you visit the shop.
That I get but not for 'true' server purposes. It will most likely be media server usage.
Agreed. RFID technology is yet to scope out its full potential in supply chain. Its extremely beneficial for supply chain security. I dont know if in international supply chain business its already in action.
Nobody likes to remain an underdog forever. This was a battle that everyone expected but most thought ARM would never be able to win. I disagree. It may not win outright but the constant nibbling on Intel market share and the desire of OEM customers for more than one source, technology advancement and lower costs will position ARM IP for more than just the mobile market.
RFID seems look very popular. However, when there is problem and suggested solution is RFID, an organization does implement it and are reluctant. What can be reason for this? May be RFID need more work to do – hardware or software.
@_hm, I think we are still in the early stages of RFID development. There are specific areas that RFID was originally targeted to address. Back in 1989 Savi Technology began with logisitics tagging products. Back then, I talked with them as they were developing a hand held, walk-around polling device for area inventory management. As a person walked up and down the aisles of stacked inventory, the hand-held reader would gather the counts for rapid inventory accounting. I was so excited about their company, I almost asked for an interview for employment. So that was 23 years ago and they are still in the asset management/tracking business. This is still the primary application for RFID. The key term here is “Asset.” If you consider humans as assets that need to be tracked for location and information data gathering purposes, then the next step is to ask how important are those assets to track and what are the consequences of losing track of them. So if those assets are tanks and very expensive military gear, then the asset becomes a liability if lost. I think _hm, we will see greater and greater deployments proportional to the cost of the RFID as measured against the considered worth of the asset. As development continues, the lower the cost of the RFID components, the lower the value of the asset will have to be to merrit the installation of a tracking system. Animals, People, Things, and Documents are all prospective candidates for massive tracking and data gathering technologies. That doesn't leave much out. My guess is that in the next couple of years we will be living with RFID as part of our daily experience as much as we now are adjusted to stop signs, traffic signals, and Starbucks on every corner.
@Rich, You are correct. There will be an upcharge from passive to active and I think the decision to implement will be similar to the smartcard now used in Cable TV systems to ensure against piracy. These smartcards have microprocessors built in with encryption technologies that require a 1:1 subscriber to operator's authorization relationship. I was buying smartcards in quantities of 10,000 for $9.75 each and charging the subscriber to cover my cost. If we begin to see loyalty cards issued in the same fashion, then we will probably see some unique marketing techniques designed to defray the cost of the cards so the retailer does not take the entire hit. I don't think customer's will pay for the cards nless there is an added incentive in promised savings over X period of time. I think of people buying coupon books like Diner's Club in order to realize greater savings at each meal. The idea being that the card pays for itself after Y number of uses.
What typical data is contained in RFID tag? Is it barcode type product code and serial number? Or date of manufacturing and best before date? Is there international standard for RFID tag data?
_hm, Instead of going into the details here, I am including a PDF URL that will let you review the standard for yourself. There is a basic information standard called EPC described in the PDF, but when RFID carriers also include both read and write capabilities via EEPROM associated add-ons, then the data is defined by the developer of the software and is only limited to the size of the memory. In real-time applications for tracking, the data can be forwarder via Wi-Fi in real time. There are lots of variations but here is the PDF that starts with the standard EPC. http://www.gs1.org/gsmp/kc/epcglobal/tds/tds_1_5-standard-20100818.pdf
@Himanshugupta,
I see, if the ads/offers are relevant that may entice potential customers to commit to buying. I was just skeptical about the loyalty of the targeted audience.
@SP,
” Its extremely beneficial for supply chain security. “
The Supply Chain security can be enhanced by RFID for sure and it can also help optimize the supply chain management. It can help track product flow and identify inbound and outbound product in the warehouse.
RFID is not dying for certain. It still has some tricks on its bag. I think integating with SCM and CRMwould be great.
Surely there are conflictions when it comes for industry based servers and home based servers but that wont be much of a problem since those 2 vary based on their requirements and data which needs to be handled.
RFID isn't dying and it's not getting some great play either relative to the way it's been hyped in recent years. It's making a splash, one drop at a time!
@Bolaji, I like the turn of phrase making a big splash one drop at a time. That implies some big drops as opposed to a steady flow. If we say the drops are implementations, I would agree as the infrastructure investment for a system beyond a handheld reader and RFID tags can be quite extensive and therefore somewhat costly. But once the infrastructure is in place and more chips come with integrated on board storage with read/ write functionality, then we will see those drops getting closer together accelerated by converging technologies using the cloud, sophisticated polling techniques, and extensive adoptions of universal standards like GS1's Gen 2 RFID with EPC and eEPC now agreed upon by 111 committee members impacting over 2 million companies worldwide. Though not always the case, EPC bar codes and data bars can be included in the RFID carrier so if you use the printed bar code as a model for volume predictions, then RFID is about to turn from a drop system into a firehose.
@Bolaji, I like the turn of phrase making a big splash one drop at a time. That implies some big drops as opposed to a steady flow. If we say the drops are implementations, I would agree as the infrastructure investment for a system beyond a handheld reader and RFID tags can be quite extensive and therefore somewhat costly. But once the infrastructure is in place and more chips come with integrated on board storage with read/ write functionality, then we will see those drops getting closer together accelerated by converging technologies using the cloud, sophisticated polling techniques, and extensive adoptions of universal standards like GS1's Gen 2 RFID with EPC and eEPC now agreed upon by 111 committee members impacting over 2 million companies worldwide. Though not always the case, EPC bar codes and data bars can be included in the RFID carrier so if you use the printed bar code as a model for volume predictions, then RFID is about to turn from a drop system into a firehose.
@Rich, You were right about the rambling but when you consider it was first posted in 2008, this is a good general overview of the areas of academic concern and potential further areas for development. But in 4 years, we have come a long long way and Sanjay seemed to anticipate the potential technologies that would come to play in both the readers and the tags themselves. I also thought it was to his credit that he was able to see back in 2006 that it would be MIT that would take the bull by the horns to instigate the bulk of the research papers and open the door to advancements in the various applications for RFID. I was impressed with his foresight. Thanks for the link.
I envisage a future like the scene from the movie “Minority Report ” where Tom Cruise walks through a store and holographic service personnel recognize him and greet him, trying to sell him stuff they imagine he would be interested in. The background info is gathered from mining data on his shopping or surfing history and I find this whole idea annoying and frankly quite scary.
There must be great opportunity in both ends of the pipe: low cost mobile processors for client based apps and super powerful processors for the host based servers managing cloud based solutions. I would look at companies like ARM and Qualcomm to do very well.
@flyingscot annoying it may be, but it is already here. If you carry a smartphone, you can get offers not only based on your purchasing history but on your location.
I can see great applications for this technology in supply chain management. On a personal level, I am not a big fan of revealing my purchasing history for someone else to use – it feels like a loss of control. Time to start wrapping my cards in tin foil!
Yep.. ARM is eating slowly. I remember somewhere mention internet companies like Facebook and some others are having there own lowcost server farms. That is these servers are cheap, easy to replace and consume less power, and of course a good candidate is using ARM processor.
Exactly Wale and I think this is a good sign for most of the markets including consumers, since there will be competition all around and the features will get upgraded in no time to match up its opponent.
To be honest Bolaji I dont belive in splashes. I feel one drop at a time is not at all enough.