The Helio X20 is expected to make its way into devices in early 2016, and will “revolutionise” mobile processors, according to MediaTek.
This is down to its ability to reduce power consumption significantly by altering the number of cores working at any one time depending on the power needed to complete tasks.
MediaTek said that this has been made possible by the firm’s new Tri-Cluster CPU architecture that has three processor clusters each designed to handle different types of workloads more efficiently.
“If a user needs heavy performance, [the Helio X20] will invoke 2, 4, 8 cores, intelligently looking at the workload to decide how many it needs,” said MediaTek’s senior director of corporate sales for EMEA, Chet Babla, in a briefing with The INQUIRER.
“There will be a dramatic drop in power consumption compared to big.LITTLE architecture because of this.”
The Tri-Cluster CPU consists of one cluster of two ARM Cortex-A72 cores running at 2.5GHz for high performance, and two clusters of four ARM Cortex-A53 cores, one running at 2GHz for medium loads and one running at 1.4GHz for light activities.
MediaTek has also integrated a CorePilot 3.0 heterogeneous computing scheduling algorithm which controls which threads are allocated to the cores.
CorePilot 3.0 schedules the tasks for all CPUs and GPUs while managing power and thermal effects so that extreme performance can be attained while creating less heat.
This is said to reduce power consumption by 30 percent compared with conventional dual-cluster architectures on top of the increase in energy efficiency thanks to Helio X20′s supported ARM Mali-T880 GPU.
“With the integration of MediaTek’s WorldMode Category 6 LTE modem with carrier aggregation and upgraded CorePilot 3.0 advanced scheduling algorithm, the Helio X20 is set to revolutionise the mobile processor industry and address the global demand for flagship mobile devices,” MediaTek said.
The Helio X20 also has several features designed to increase device display performance and multimedia experiences.
These include support for dual main cameras with a built-in 3D depth engine for a faster shot-to-shot experience, multi-scale de-noise engines for higher quality images, a 120Hz mobile display refresh rate for crisper and more responsive browsing, and an integrated ARM Cortex-M4 low power sensor processor to support always-on applications such as MP3 playback and voice activation.
Instant Articles will let stories load more than 10 times faster than standard mobile web articles and will include content from publishers such as the New York Times, BuzzFeed and National Geographic, Facebook said in a blog post on its website.
“Instant Articles lets them deliver fast, interactive articles while maintaining control of their content and business models,” said Facebook Chief Product Officer Chris Cox.
The news publishers can either sell and embed advertisements in the articles and keep all of the revenue, or allow Facebook to sell ads.
The Internet social networking company will also let the news companies track data and traffic through comScore and other analytics tools.
The other launch partners for Instant Articles are NBC, The Atlantic, The Guardian, BBC News, Spiegel and Bild, Facebook said.
TSMC looks certain to keep its foot firmly on the throat of 28nm chip production, according to a new study.
Digitimes said that it looks like competition from Samsung Electronics, Globalfoundries, UMC and SMIC will not be enough to upset TSMC’s rule.
TSMC started its commercial 28nm process five years ago and now accounts for over 75 per cent of the global 28nm foundry market. In addition, the 28nm products also contributed 30 per cent to TSMC’s total revenues in 2014.
Initially 28nm capacity was triggered by rising demand for application processors for smartphones and tablets. TSMC had clients including Qualcomm, MediaTek, Apple and Nvidia.
Vlients from the mobile device sector will continue to fill TSMC with 28nm orders in 2015, but the SSD market is expected this year is likely to give another wave of demand for 28nm.
Controller chips for SSDs are to officially migrate into the 28nm process in 2015, and TSMC has an advantage to solicit more orders from SSD vendors.
TSMC has reportedly snapped up SSD controller chips orders from Marvell Technology, Phison Electronics and more recently from Apple. Thus, TSMC will see its share in the 28nm process market remain high and contribution of the 28nm products to its sales further increase in 2015.
MediaTek has established itself as the world’s second-largest maker of Long-Term Evolution (LTE)-enabled cellular baseband processors in 2014.
Beancounters at market research firm Strategy Analytics have added up the numbers and divided by their shoe size and worked out that the industry has a new number two.
While everyone knows that Qualcomm, has near total dominance of the high-growth LTE baseband segment in the past and had a 95 per cent share in 2013 a battle has been going on behind the scenes.
Other LTE baseband suppliers had too little of a share to be ranked behind Qualcomm, MediaTek had enough of an impact in the market in 2014 to get a second-place ranking from Strategy Analytics.
The research firm predicted that MediaTek will continue to gain shares in the LTE baseband segment thanks to increased traction in China, the world’s biggest smartphone market.
“Growing revenue contributions from LTE basebands will lift MediaTek’s baseband revenue share over the next few quarters,” said Christopher Taylor, director of the Strategy Analytics RF and wireless component service.
In 2014 revenue from LTE baseband sales overtook revenue from 3G baseband sales for the first time, thanks to a strong push from the industry, the research firm said.
The global market for cellular baseband processors, which are used in mobile devices to process wireless communication, grew an impressive 14.1 per cent year-over-year to reach $22 billion in 2014.
Qualcomm, MediaTek, Spreadtrum, Marvell and Intel grabbed the top-five cellular baseband revenue share spots in 2014, the research company said.
Qualcomm had a 66 per cent revenue share of the cellular baseband processor market, followed by MediaTek with a 17 per cent share and Spreadtrum with a 5 per cent share, according to Strategy Analytics.
While everyone is rushing to 10nm process technology for smartphones, fabless chipmaker MediaTek is about to create a 10 core SoC using TSMC’s 20nm process tech.
According to Digitimes the outfit is about to enter volume production of its 10-core SoC series for smartphones in the third quarter of 2015.
Dubbed Helios X20, the SoC will be targeted at Chinese based smartphone makers who want to upgrade their flagship devices.
Marketing will begin in the middle of the second quarter. When it gets into the shops it will be the world’s first 10-core chip.
The Helios X20 uses a 2+4+4 design, delivering 40 per cent more performance than eight-core chips. While this will give a lot of power to a smartphone, it is not clear what it will do for battery life or the size of the beast.
Still it is nice to see that someone has found a new way of getting more life out of the 20 nm process and do something good with it.
The Internet company has hired advisers to help it evaluate options for the stake, Chief Executive Officer Marissa Mayer told investors on a conference call on Tuesday. It will not be included in the planned spin-off of its stake in China’s Alibaba Group Holding Ltd, she said.
Investors have been urging Mayer to monetize the Yahoo Japan stake separately, after she announced plans to spin off the Alibaba stake in January, which could be worth $40 billion.
The advisers will help Yahoo “determine the most promising opportunities to maximize value” for the Yahoo Japan stake, said Mayer.
But Wall Street remained broadly cautious about the plan.
“They are taking the slow train, stressing the process,” said Colin Gillis, an analyst BGC Partners, who warned that a deal, if any, could be a long way down the line. “Engaging advisers doesn’t mean spinning it out.”
Yahoo owns about 35 percent of Yahoo Japan Corp, which has a market value of almost $25 billion on the Tokyo Stock Exchange. Japanese internet company SoftBank Corp is the biggest shareholder, with about 36 percent, according to Thomson Reuters data.
Last month Yahoo shareholder Starboard Value LP said the Alibaba spin-off was a “good first step” but urged Yahoo to also spin off its Yahoo Japan stake in a tax-efficient manner. Starboard did not reply to a request for comment.
Chip designer ARM reported a 36 per cent rise in first-quarter net profit amid strong demand for its technology.
The British company said that expects 2015 revenue to meet the expectations of the cocaine nose jobs of Wall Street.
ARM recorded net profit of $126.7 million for the three months to March 31 and revenue rose 22 percent.
Shares in ARM, which makes money by licensing its designs to chip makers, then collecting royalty revenue when the chips ship, were up by more than 5 per cent on the back of the news.
Processor-royalty revenue in dollar terms, a much-watched figure, rose 31 per cent on the year, the company said, adding that it has signed 30 processor licenses for a broad range of applications.
ARM CEO Simon Segars said: As the world becomes more digital and more connected, we continue to see an increase in the demand for ARM’s smart and energy-efficient technology, which is driving both our licensing and royalty revenues.@
Processor-licensing revenue was down 2 per cent in the quarter, which was in line with expectations following strong growth previously. Chief Financial Officer Tim Score told journalists he expects it to grow in future quarters.
Aside from smartphones and tablets, ARM said it is also seeing demand for its processors to be used for servers and networking and for the “Internet of Things”, a term used for the growing tendency for more items to be wirelessly connected.
ARM expects to benefit from the growth of the Internet of Things in areas such as health and in cars, Score said.
Twitter.com has been redesigned to make content posted to the site more accessible to people who do not have accounts with the service. Those people can visit the site, and as of Wednesday they will find 18 tabs leading to streams of content on various topics, whether it be country singers, general news, or travel guides.
There’s also dozens of other curated streams of content accessible from links on the home page, with the content organized around more granular topics like U.S. federal agencies, art museums and wedding guides.
Previously, visitors to Twitter.com who did not have accounts were greeted with a sign-up page.
The changes come as Twitter faces continuing pressure to grow the number of people who use its site, and find new ways to make money off people who see its tweets and interact with them. Twitter ended the last quarter of 2014 with 288 million users who log in monthly — a 20 percent increase from the previous year, but the smallest annual growth rate Twitter ever reported.
One of Twitter’s biggest problems is that many people still don’t understand what it’s for.
With the redesigned home page, the company is trying to address this, by highlighting the site’s value as a source of real-time information and news. The tweets Twitter has selected for its new streams, the company says, come from some of the most popular accounts posting on those topics.
People without accounts still can’t do much to interact with the content. To reply to, re-tweet or “favorite” one of the tweets, the visitor is prompted to create an account. But with the redesign, Twitter hopes it might give the uninitiated enough bait to sign up.
And even without a flood of new sign-ups, Twitter’s new home page is likely to get more tweets in front of more people. That could give rise to new advertising methods around those tweets.
The new home page is available first in the U.S. on the desktop, Twitter said on Wednesday, though it will be arriving “to more places over time.”
MediaTek is working on two new tablet SoCs and one of them is rumored to be a $5 design.
The MT8735 looks like a tablet version of Mediatek’s smartphone SoCs based on ARM’s Cortex-A53 core. The chip can also handle LTE (FDD and TDD), along with 3G and dual-band WiFi. This means it should end up in affordable data-enabled tablets. There’s no word on the clocks or GPU.
The MT8163 is supposed to be the company’s entry-level tablet part. Priced at around $5, the chip does not appear to feature a modem – it only has WiFi and Bluetooth on board. GPS is still there, but that’s about it.
Once again, details are sketchy so we don’t know much about performance. However, this is an entry-level part, so we don’t expect miracles. It will have to slug it out with Alwinner’s $5 tablet SoC, which was announced a couple of months ago
According to a slide published by Mobile Dad, the MT8753 will be available later this month, but we have no timeframe for the MT8163.
But there’s nothing to see here as far as Torvalds is concerned. It’s just another day in the office. And all this in “Back To The Future II” year, as well.
Meanwhile under the bonnet, the community are already slaving away on Linux 4.1 which is expected to be a far more extensive release, with 100 code changes already committed within hours of Torvalds announcement of 4.0.
But there is already some discord in the ranks, with concerns that some of the changes to 4.1 will be damaging to the x86 compatibility of the kernel. But let’s let them sort that out amongst themselves.
After all, an anti-troll dispute resolution code was recently added to the Linux kernel in an effort to stop some of the more outspoken trolling that takes place, not least from Torvalds himself, according to some members of the community.
It has apparently has taped out multiple 14nm designs and is tweaking its equipment using a lead product at the moment. The company is on track to start high-volume shipments of 14nm chips this year.
Jason Gorss, a spokesman for Global Foundries said that the outfit’s 14nm FinFET technology is maturing and on schedule at our Fab 8 facility in Malta, New York,.
“The early version (14LPE) is qualified in our fab and our lead product is yielding in double digits. Since 2014, we have taped multiple products and testchips and are seeing rapid progress, in yield and maturity, for volume shipments in 2015.”
The comment follows a statement from Mubadala Development last week week which claimed htat Global Foundries had begun ramping manufacturing of 14nm chips for customers.
Mubadala, which owns GlobalFoundries, did not provide any details. Even though production is currently not in high volume, it is clear that GlobalFoundries ships certain chips to clients.
It is not clear what 14nm chips GlobalFounfries produces at present, but it is highly likely that the company makes Samsung Exynos 7420 application processors for its process tech partner.
Another early partner of GlobalFoundries with its 14nm FinFET production could be Apple, which is expected to use Samsung’s 14nm process tech to make its upcoming A9 system-on-chip.
Global Foundries licensed Samsung’s 14LPE (low-power early) and 14LPP (low-power plus) process technologies last year.
This process uses FinFET transistors and rely on back-end-of-line (BEOL) interconnects of 20nm manufacturing technology. The14nm FinFET transistors allow a performance boost for chips by 20 per cent at the same power or cut power consumption by 35 per cent without decreasing performance or complexity.
The deal that helped Crytek recover from its recent financial difficulties was Amazon, according to a report from Kotaku.
The online retail giant signed a licensing deal for CryEngine, Crytek’s proprietary game engine. Sources within the company put the deal’s value at between $50 million and $70 million, and suggested that Amazon may be using it as the bedrock for a proprietary engine of its own.
However Amazon uses the technology, though, the importance of the deal for Crytek cannot be overstated. Last year, during the summer, it became apparent that all was not well at the German developer. Employees hadn’t been fully paid in months, leading to an alleged staff walkout in its UK office, where a sequel to Homefront was in development. Koch Media acquired the Homefront IP and its team shortly after.
When the company’s management eventually addressed the rumors, it had already secured the financing necessary to take the company forward. No details of the deal were offered, but it’s very likely that Crytek got the money it needed from Amazon.
We have contacted Crytek to confirm the details, but it certainly fits with the perception that Amazon could emerge as a major creator of game content. It has snapped up some elite talent to do just that, it acquired Twitch for a huge sum of money, and it has been very open about where it plans to fit into the overall market.
An upcoming MediaTek SoC has been spotted in GFXbench and this tablet-oriented chip has created a lot of speculation thanks to the choice of GPU.
The Cortex-A53 based MediaTek MT8163 was apparently tested on a dev board with 2GB of RAM and the benchmark failed to identify the GPU. GFXbench identified the GPU as a part coming from “MediaTek Inc. Sapphire-lit”.
Spinning up the rumour mill
This is where the speculation starts, as many punters associated the GPU with AMD, and the presence of the word “Sapphire” also prompted some to conclude that AMD’s leading GPU add-in-board partner had something to do with it.
The Sapphire word association doesn’t look like anything other than clutching at straws, because it’s highly unlikely that an AIB would have much to do with the process of licensing AMD IP for mobile graphics.
However, this does not necessarily mean that we are not looking at a GPU that doesn’t have anything to do with AMD. The fact that MediaTek’s name is on it is perhaps more important, because it suggests an in-house design. Whether or not the part is indeed an in-house design, and whether it features some AMD technology, is still up for debate.
Why would MediaTek need AMD to begin with?
MediaTek relies on ARM Mali GPUs, although it uses Imagination GPUs on some designs. So where does AMD fit into all this?
As we reported last month, the companies have been cooperating on the SoC graphics front for a while, but they are tight lipped about the scope of their cooperation.
MediaTek is a supporter of HSA and a founding member of the HSA Foundation, but this doesn’t prove much, either, since the list of founding members includes ARM, Imagination, Texas Instruments, Samsung and Qualcomm.
Using AMD technology on SoCs would have to be a long-term strategy, built around the concept of using AMD IP to boost overall SoC performance rather than just GPU performance. This is why we do not expect to see the fruits of their cooperation in commercial products anytime soon.
Improved compute performance is one of the reasons MediaTek may be inclined to use AMD technology, but another angle is that “Graphics by AMD” or “Radeon Graphics” would sound good from a marketing perspective and allow MediaTek to differentiate its products in a saturated market.
In a move that could produce even more automated suggestions and tips for LinkedIn users, the professional network has purchased California startup Refresh, the maker of an app that gathers news and insights about participants in meetings.
Launched three years ago, Refresh is designed to be a “digital briefing book” that can call up online information related to people that users are scheduled to meet. The information can be anything from blog posts, news articles or Facebook posts to personal notes or favorite sports teams.
The Refresh mobile and desktop app is aimed at helping people relate to one another more quickly, but it can also be used to refresh one’s memory when running into acquaintances unexpectedly.
The details of the deal were not disclosed. Refresh has stopped taking on new users and its app will shut down April 15.
“Refresh has surfaced insights associated with hundreds of millions of meetings, and has been central to countless connections and closed deals,” co-founder Bhavin Shah wrote on the Refresh blog in announcing the deal.
LinkedIn already has an app called Connected that was somewhat of a rival to Refresh. It can log the people users have met and offer updates and information about interests shared with “connections,” which are acquaintances in the LinkedIn lingo. It’s unclear whether Refresh features will be added to Connected or the LinkedIn website itself.
“Our team will focus its efforts on providing LinkedIn members with more insights to help them better do their jobs,” Shah wrote.
AMD has been working closely with Microsoft on the upcoming DirectX 12 to create something which it calls “Asynchronous Shaders,” which are more efficient way of handling task queues.
In DirectX 11 synchronous task scheduling is handled with multi-threaded graphics with pre-emption and prioritization.
GPU’s shaders do the drawing of the image, computing of the game physics, post-processing and more, and they do this by being assigned various tasks. These are handled through the command stream which is generated through merging individual command queues.
However the queue has empty bits because they are not generated in order in multi-threaded graphics. This means the shaders don’t reach their full potential and get stuck in dead end jobs, with ugly demanding partners who demand huge sacrifices for their screaming kids..
DirectX 12 enters
In DirectX 12, Asynchronous Shaders provide asynchronous multi-threaded graphics with pre-emption and prioritization. The Asynchronous Compute Engines (ACE) on AMD’s GCN-based GPUs will interleave the tasks, filling the gaps in one queue with tasks from another.
Despite that, it can still move the main command queue to the side to let priority tasks pass by when necessary. It probably goes without saying that this leads to a performance gain.
AMD’s GCN GPUs, each ACE can handle up to eight queues, with each one looking after its shaders. Basic GPUs will have just two ACEs, while more elaborate GPUs carry eight.
AMD demonstrations show that it does not cost much in frame rate to run Asynchronous Shaders and post-processing but it will improve performance.
All the increased parallelism and will ensure that more frames make their way to the screen even faster, which can be especially interesting for purposes such as VR.
Intel has released details of its next -generation Xeon Phi processor and it is starting to look like Intel is gunning for a chunk of Nvidia’s GPU market.
According to a briefing from Avinash Sodani, Knights Landing Chief Architect at Intel, a product update by Hugo Saleh, Marketing Director of Intel’s Technical Computing Group, an interactive technical Q&A and a lab demo of a Knights Landing system running on an Intel reference-design system, Nvidia could be Intel’s target.
Knights Landing and prior Phi products are leagues apart and more flexible for a wider range of uses. Unlike more specialized processors, Intel describes Knights Landing as taking a “holistic approach” to new breakthrough applications.
The current generation Phi design, which operates as a coprocessor, Knights Landing incorporates x86 cores and can directly boot and run standard operating systems and application code without recompilation.
The test system had socketed CPU and memory modules was running a stock Linux distribution. A modified version of the Atom Silvermont x86 cores formed a Knights Landing ’tile’ which was the chip’s basic design unit consisting of dual x86 and vector execution units alongside cache memory and intra-tile mesh communication circuitry.
Each multi-chip package includes a processor with 30 or more tiles and eight high-speed memory chips.
Intel said the on-package memory, totaling 16GB, is made by Micron with custom I/O circuitry and might be a variant of Micron’s announced, but not yet shipping Hybrid Memory Cube.
The high-speed memory is similar to the DDR5 devices used on GPUs like Nvidia’s Tesla.
It looks like Intel saw that Nvidia was making great leaps into the high performance arena with its GPU and thought “I’ll be having some of that.”
The internals of a GPU and Xeon Phi are different, but share common ideas.
Nvidia has a big head start. It has already announced the price and availability of a Titan X development box designed for researchers exploring GPU applications to deep learning. Intel has not done that yet for Knights Landing systems.
But Phi is also a hybrid that includes dozens of full-fledged 64-bit x86 cores. This could make it better at some parallelizable application categories that use vector calculations.