Sign up for our Technology Newsletter 
nVidia NVLINK 2.0 Going In IBM Servers
Comments Off on nVidia NVLINK 2.0 Going In IBM Servers
On Monday, PCWorld reported that the first servers expected to use Nvidia’s second-generation NVLINK 2.0 technology will be arriving sometime next year using IBM’s upcoming Power9 chip family.
IBM launched its Power8 lineup of superscalar symmetric multiprocessors back in August 2013 at the Hot Chips conference, and the first systems became available in August 2014. The announcement was significant because it signaled the beginning of a continuing partnership between IBM and Nvidia to develop GPU-accelerated IBM server systems, beginning with the Tesla K40 GPU.
The result was an HPC “tag-team” where IBM’s Power8 architecture, a 12-core chip with 96MB of embedded memory, would eventually go on to power Nvidia’s next-generation Pascal architecture which debuted in April 2016 at the company’s GPU Technology Conference.
NVLINK, first announced in March 2014, uses a proprietary High-Speed Signaling interconnect (NVHS) developed by Nvidia. The company says NVHS transmits data over a differential pair running at up to 20Gbps, so eight of these differential 20Gbps connections will form a 160Gbps “Sub-Link” that sends data in one direction. Two sub-links—one for each direction—will form a 320Gbps, or 40GB/s bi-directional “Link” that connects processors together in a mesh framework (GPU-to-GPU or GPU-to-CPU).
NVLINK lanes upgrade from 20Gbps to 25Gbps
IBM is projecting its Power9 servers to be available beginning in the middle of 2017, with PCWorld reporting that the new processor lineup will include support for NVLINK 2.0 technology. Each NVLINK lane will communicate at 25Gbps, up from 20Gbps in the first iteration. With eight differential lanes, this translates to a 400Gbps (50GB/s) bi-directional link between CPUs and GPUs, or about 25 percent more performance if the information is correct.
NVLINK 2.0 capable servers arriving next year
Meanwhile, Nvidia has yet to release any NVLINK 2.0-capable GPUs, but a company presentation slide in Korean language suggests that the technology will first appear in Volta GPUs which are also scheduled for release sometime next year. We were originally under the impression that the new GPU architecture would release in 2018, as per Nvidia’s roadmap. But a source hinted last month that Volta would be getting 16nm FinFET treatment and may show up in roughly the same timeframe as AMD’s HBM 2.0-powered Vega sometime in 2017. After all, it is easier for Nvidia to launch sooner if the new architecture is built on the same node as the Pascal lineup.
Still ahead of PCI-Express 4.0
Nvidia claims that PCI-Express 3.0 (32GB/s with x16 bandwidth) significantly limits a GPU’s ability to access a CPU’s memory system and is about “four to five times slower” than its proprietary standard. Even PCI-Express 4.0, releasing later in 2017, is limited to 64GB/s on a slot with x16 bandwidth.
To put this in perspective, Nvidia’s Tesla P100 Accelerator uses four 40GB/s NVLINK ports to connect clusters of GPUs and CPUs, for a total of 160GB/s of bandwidth.
With a generational NVLINK upgrade from 40GB/s to 50GB/s bi-directional links, the company could release a future Volta-based GPU with four 50GB/s NVLINK ports totaling of 200GB/s of bandwidth, well above and beyond the specifications of the new PCI-Express standard.
Courtesy-Fud
Will Intel Release It’s 10nm Processors By 2017?
Comments Off on Will Intel Release It’s 10nm Processors By 2017?
Intel has said that a job advert which implied that it would not be using the 10nm process for two years was inaccurate and confirmed that it is on track for a 2017 release.
The advert, which was spotted by the Motley Fool has since been taken down, said the company’s 10-nanometer chip manufacturing technology would begin mass production “approximately two years” from the posting date.
Intel has said that the advert was wrong and confirmed that its “first 10-nanometer product is planned for the second half of 2017.”
It is not expected that Intel will roll out server chips in 2017. At the moment the plan appears to be introducing its second-generation 14-nanometer server chip family in early to mid-2017. But instead Intel will be trying to get its process ramped at high yields experimenting on the PC market so that 10-nanometer server processors will be ready for the first half of 2018.
This follows Intel’s traditional pattern of a having a few parts released as it experiments with the new tech. This is what happened in the first year of Intel’s 14-nanometer availability.
Courtesy-Fud
Enterprise Needs Driving Cloud Sales Boom
Comments Off on Enterprise Needs Driving Cloud Sales Boom
The cloud continues to gain major ground, driven by enterprise storage needs.
Sales are way up for little-known manufacturers that sell directly to big cloud companies like Google and Facebook, while the market for traditional external storage systems is shrinking, according to research company IDC.
Internet giants and service providers typically don’t use specialized storage platforms in their sprawling data centers. Instead, they buy vast amounts of capacity in the form of generic hardware that’s controlled by software. As users flock to cloud-based services, that’s a growing business.
Revenue for original design manufacturers that sell directly to hyperscale data-center operators grew by 25.8 percent to more than US$1 billion in the second quarter, according to the latest global IDC report on enterprise storage systems. Overall industry revenue rose just 2.1 percent from last year’s second quarter, reaching $8.8 billion.
These so-called ODMs are low-profile vendors, many of them based in Taiwan, that do a lot of their business manufacturing hardware that’s sold under better known brand names. Examples include Quanta Computer and Wistron.
General enterprises aren’t buying many systems from these vendors, but the trends at work in hyperscale deployments are growing across the industry. Increasingly, the platform of choice for storage is a standard x86 server dedicated to storing data, according to IDC analyst Eric Sheppard. Sales of server-based storage rose 10 percent in the quarter to reach $2.1 billion.
Traditional external systems like SANs (storage area networks) are still the biggest part of the enterprise storage business, logging $5.7 billion in revenue for the quarter. But sales in this segment were down 3.9 percent.
Overall demand for storage capacity continued to grow strongly, with 37 percent more capacity shipped in the quarter compared with a year earlier.
Source-http://www.thegurureview.net/aroundnet-category/enterprise-storage-needs-driving-cloud-sales-boom.html
Is Intel Supporting Open-Source?
Intel has suddenly made some interesting hardware less interesting to open sourcers by insisting that its i915 DRM kernel graphics driver for upcoming Skylake and Broxton hardware demands some binary-only firmware blobs.
According to Phoronix these first i915 DRM firmware blobs are for Skylake and Broxton for the GuC and DMC.
DMC is the Display Microcontroller used by Skylake (Gen9) within the display engine to save and restore its state when entering into low-power states and then resuming. It saves and restores display registers across low-power states separate of the kernel.
Intel said that the firmware blobs are required by the DRM driver rather than being an optional add-on.
The license of these firmware blobs also indicate that redistribution is only allowed in binary form without modification. Beyond that, “no reverse engineering, decompilation, or dis-assembly of this software is permitted.”
Basically this will kill off any desire for Open Source enthusiasts to touch Skylake, although we doubt Intel will be too worried – they are a very nice couple. In any event AMD apparently uses something similar to protect bits of its operation.
Still Intel is shipping these firmware files early so everyone knows they are there.
Intel Talks More About Skylake
A new Intel roadmap suggests the first Broadwell LGA parts will launch in Q2, while Skylake-S parts will come in Q3.
The roadmap was published by PC Online and points to two Broadwell LGA launches this quarter – the Core i7-5775C and Core i5-5675C. These two parts will be joined by a total of four Skylake-S products in Q3, the Core i7-6700K, Core i7-6700, Core i5-6600K, Core i5-6600 and the Core i5-6500.
Both Skylake-S and Broadwell LGA will replace the current crop of Haswell parts, including Devil’s Canyon products. However, Broadwell LGA sits one tier above Skylake-S and Haswell-based products.
Starting in Q4, we should see more Broadwell LGA parts, but we don’t have any names yet. In the first quarter of 2016, we can also expect new Skylake-S parts.
Speaking of 2016, Intel plans to unleash the Broadwell-E in the first quarter of 2016. Little is known about Broadwell-E, but the new 14nm flagship is expected to sport eight cores. Clocks remain unknown, although the 14nm node promises substantial gains.
Intel Sends Braswell SoC To Partners
Intel announced that it is now shipping the Bay Trail system on a chip (SoC) successor codenamed Braswell to OEM partners.
Announced almost exactly a year ago at Intel’s Developer Forum in Beijing, Braswell is a more powerful version of Bay Trail running on the 14nm fab process, designed to power low-cost devices like Chromebooks and budget PCs.
The chip maker said that devices will hit the market sometime in late summer or autumn.
“We expect Braswell-based systems to be available in the market for the back to school 2015 selling season,” an Intel representative told The INQUIRER. “Specific dates and options will be announced by our OEM partners.”
That’s all Intel will give us for now, but we were told that full details regarding the upcoming chip will be revealed at IDF in Shenzhen next week.
Braswell was expected to arrive at the end of 2014 when it was originally unveiled last year.
Kirk Skaugen, general manager of Intel’s PC Client group, said that it will replace Bay Trail as part of the Atom line, and will feature in over 20 Chromebook designs.
“Last year, we had only four designs on Chrome. Today I can announce that we will have over 20 designs on Chrome,” said Skaugen at the time.
Intel recently announced another 14nm chip, the Atom x range, previously codenamed Cherry Trail, although this will be focused on tablets rather than the value PC market segment and Chromebooks like Braswell.
In terms of power, Braswell is likely to fit snuggly above the Atom x5 and x7 Cherry Trail SoCs and beneath the firm’s recently announced 5th-generation Core products, previously codenamed Broadwell.
Unveiled at Mobile World Congress earlier this year, Intel’s Atom x5 and x7 chips, previously codenamed Cherry Trail, are also updates to the previous Bay Trail Atom line-up, being the first Intel Atom SoCs on 14nm.
These higher-powered SoCs are designed to bring improved 3D performance to mainstream premium handheld devices running full versions of Windows and Android, such as 7in to 10.1in tablets and 2-in-1 hybrid laptops priced at around $119 to $499.
For example, Microsoft quietly announced on Tuesday that the upcoming Surface 3 tablet-laptop hybrid will be powered by an Intel Atom x7. The device is priced at $500.
Intel Debuts The N3000 Series SoC
Intel has launched Intel N3000 series systems on a chip (SoCs), which will kill off Bay Trail-M and Bay Trail-D SoCs on the desktop and mobile PCs.
CPU World also has spotted some other chips which have been revealed to the world.
Intel has also launched desktop and mobile Core i3 and Pentium microprocessors. New mobile models are Pentium 3825U, Core i3-5015U and i3-5020U. These ones are based on Broadwell 14nm.
Core i3-5015U and i3-5020U are dual-cores with Hyper-Threading technology, HD 5500 graphics and ultra low 15 Watt TDP. The processors run at 2.1 GHz and 2.2 GHz. This is 100 MHz higher than the i3-5005U and i3-5010U models, that were launched three months ago.
The i3-5015U and i3-5020U chips offer a 50 MHz higher graphics boost. Official prices of these SKUs are $275 and $281.
The Pentium 3825U incorporates a couple of enhancements on the older Pentium 3805U. It supports Hyper-Threading that allows it to process twice as many threads. It also has base and maximum graphics frequencies increased to 300 MHz and 850 MHz.
The 3805U and 3825U operate at 1.9 GHz and have 2 MB L2 cache. The 3825U processor is rated at 15 Watt TDP, and priced at $161.
Will Intel Release Skylake This Year?
Intel has confirmed that it will release Core M processors this year based on its new Skylake chip design.
Intel CEO Brian Krzanich said at the Goldman Sachs Technology and Internet conference that the the new Core M chips are due in the second half of the year and will also extend battery life in tablets, hybrids, and laptop PCs.
The new chips will mean much thinner tablets and mobile PCs which will make Apple’s Air look decidedly portly. Intel’s Core M chips, introduced last year, are based on the Broadwell but the Skylake chips should also improve graphics and general application performance.
The Skylake chips will be able to run Windows 10, as well as Google’s Chrome and Android OSes, Krzanich said. But most existing Core M systems run Windows 8.1, and Intel has said device makers haven’t shown a lot of interest in other OSes. So most Skylake devices will probably run Windows 10. Chipzilla is expected to give more details about the new Core M chips in June at the Computex trade show in Taipei.
Skylake systems will also support the second generation of Intel’s RealSense 3D camera technology, which uses a depth sensor to create 3D scans of objects, and which can also be used for gesture and facial recognition. The hope is that the combination of Skylake and a new Windows operating system will give the PC industry a much needed boost.
In related news, Intel announced that socketed Broadwell processors will be available in time for Windows 10.
IBM Goes Bare Metal
IBM has announced the availability of OpenPower servers as part of the firm’s SoftLayer bare metal cloud offering.
OpenPower, a collaborative foundation run by IBM in conjunction with Google and Nvidia, offers a more open approach to IBM’s Power architecture, and a more liberal licence for the code, in return for shared wisdom from member organisations.
Working in conjunction with Tyan and Mellanox Technologies, both partners in the foundation, the bare metal servers are designed to help organisations easily and quickly extend infrastructure in a customized manner.
“The new OpenPower-based bare metal servers make it easy for users to take advantage of one of the industry’s most powerful and open server architectures,” said Sonny Fulkerson, CIO at SoftLayer.
“The offering allows SoftLayer to deliver a higher level of performance, predictability and dependability not always possible in virtualised cloud environments.”
Initially, servers will run Linux applications and will be based on the IBM Power8 architecture in the same mold as IBM Power system servers.
This will later expand to the Power ecosystem and then to independent software vendors that support Linux on Power application development, and are migrating applications from x86 to the Power architecture.
OpenPower servers are based on open source technology that extends right down to the silicon level, and can allow highly customised servers ranging from physical to cloud, or even hybrid.
Power systems are already installed in SoftLayer’s Dallas data centre, and there are plans to expand to data centres throughout the world. The system was first rolled out in 2014 as part of the Watson portfolio.
Prices will be announced when general availability arrives in the second quarter.
Intel Investing Big In Isreal
Intel is investing a further $550 million in Israel, more specifically in the upgrade of its Fab 28 in Kiryat Gat.
According to Israel21c, this will bring the total scope of Intel investments in Israel to over $6 billion since 2006. The Kiryat Gat facility is likely to be one of the first Intel 10nm fabs.
Israeli Ministry of Economy official Ziva Eger said the investment will help create thousands of jobs and reinforce the country’s standing as a world leader in technology.
“The agreement signed today between the Industrial Cooperation Authority and Intel is another expression of Intel’s contribution by way of its purchase of equipment, new technologies and Israeli products developed together with Intel,” said CEO of Intel Israel Maxine Fassberg.
Fab 28 currently churns out 22nm silicon for Intel. The fab was passed over for the 14nm upgrade. A source familiar with the matter told us that Israel competes with Ireland for every node upgrade.
“We lost 14nm to Ireland and won 10nm,” the source said.
Israel is currently in a better position to offer incentives and subsidies for such investments, as Ireland’s ‘business-friendly’ tax policies are being scrutinized by the European Union.
Intel is expected to launch the first 10nm CPU in 2016, followed by 7nm parts a couple of years later.