NAND flash prices have been inflating excessively lately and Phison Electronics chairman Khein Seng Pua warned that prices are set to go up again in the third quarter as end-market demand surges.
He told Digitimes that while prices might decrease a little in the second quarter, Chipmakers’ ongoing transition from 2D to 3D NAND memory has led to tight supply and inflated the chip prices.
System OEMs are reluctant to deliver their products as the more they sell the more they lose due to soaring NAND flash costs, Pua warned.
Meanwhile, chipmakers’ supply to channel distributors has been falling short of demand prompting the distributors to promote lower-capacity storage devices.
“Channel distributors particularly those in China have turned to promote 96GB SSDs instead of 128GB ones due to insufficient chip supply,” Pua said.
Distributors have even experienced tight supply of 8GB and 4GB eMMC devices.
Pua believes NAND flash prices will soon see correction following excessive gains but Apple’s new iPhone will take a lot of NAND flash from the market and push prices up again.
Chipmakers’ transition to 3D NAND memory will become smooth in general between May and June, which will help ease the supply shortages, Pua indicated.
The industry’s output of 64-layer 3D NAND will account for more than half of the total output in the fourth quarter of 2017 Pua said.
Originally scheduled to launch sometime in Q3 2017, with August as the most obvious target, it appears that Intel has pushed the launch for its new HEDT (High-end Desktop) X299 platform date forward to the 25th and 27th week of this year, which puts it somewhere in the end of June, beginning of July, timeframe.
As detailed earlier, Intel Skylake-X lineup will include six- and ten-core CPUs, have support for quad-channel DDR4-2667 memory and should be the core of Intel’s HEDT X299 LGA2066 platform.
The Kaby Lake-X lineup will only include quad-core SKUs, have a TDP of 112W and lack the integrated GPU, but also come dual-channel DDR4-2667 memory support and 16 PCIe 3.0 lanes, down from up to 44 PCIe 3.0 lanes with the Skylake-X SKUs.
If the rumor, coming from Benchlife.info, proves out to be true, we should hear and see more about Intel’s new X299 HEDT platform at the Computex 2017 show, kicking off on May 30th in Taipei.
You might seen we’ve writing about millimeter waves several times. and we usually attributed this term to 5G. AMD has just acquired Nitero, a millimeter wave company that wants to use this technology to cut the cord on your VR and AR headset.
AMD has figured out that cables are a very limiting factor in a Virtual Reality or Augmented Reality. This is not a big secret as even if you only had a few minutes to play with one, you quickly realize that making things wireless is more comfortable.
The acquisition provides AMD with a broader portfolio of IP capable of enabling VR headset and solution providers with key technology required to create more immersive computing experiences.
Mark Papermaster, AMD chief technology officer and senior vice president said:
“Unwieldly headset cables remain a significant barrier to drive widespread adoption of VR. Our newly acquired wireless VR technology is focused on solving this challenge, and is another example of AMD making long-term technology investments to develop high-performance computing and graphics technologies that can create more immersive computing experiences.”
Nitero has designed a phased-array beamforming millimeter wave chip to address the challenges facing wireless VR and AR. This is the same frequency that Intel and Qualcomm will use for Wi-Gig. This enables very fast speeds within a room, but due to its high frequency the signal won’t really penetrate any walls.
This is not that important for the VR and AR markets as we don’t see a case where you need to leave an office or a room with the VR / AR headset on.
The 60GHz technology has the potential to enable multi-gigabit transmit performance with low latency in room-scale VR environments. It will rely heavily on the beamforming characteristics to solve the requirement for line-of-sight associated with traditional high-frequency mm-wave systems. The main goal is potentially eliminating wired VR headsets and letting users to become more easily immersed in virtual and augmented worlds.
Nitero co-founder and CEO Pat Kelly said:
“Our world class engineering team has been focused on solving the difficult problem of building wireless VR technologies that can be integrated into next-generation headsets. We are excited to play a role in furthering AMD’s long-term technology vision.”
Pat joined AMD as corporate vice president, Wireless IP highlighting the importance of the whole acquisition and the whole technology potential. Fudzilla calls this a step in the right direction.
Despite promises that Intel made to Donald (Prince of Orange) Trump that it was building new factories in the Land of the Fee, it seems that Intel is still downsizing its US workforce.
The number of full-time workers directly employed by Intelnear Rio Rancho fell by 37 percent in 2016 – from 1,900 salaried workers in 2015 to 1,200 as of December – according to the company’s latest annual report to the Sandoval County Commission.
Intel spokeswoman Liz Shipley told the Albuquerque Journal in an email that its head count is down from what it reported last spring in its 2015 report.
This is the sharpest annual decrease to date in direct, full-time employment at the plant since the company began laying off workers and reducing its head count through attrition in 2013. In that year, the Sandoval County plant employed 3,300 people, meaning its salaried workforce has fallen by nearly two thirds over the last four years.
The company still employs about 1,000 contract workers, about half of whom are generally on site daily to work on specific projects, Shipley said. But it’s not clear how many of those are full or part-time workers.
Intel announced in April 2016 that it planned to lay off about 12,000 people worldwide, or about 11 percent of its global workforce. That restructuring affected the Sandoval County site, according to the Intel report.
Sandoval County Commissioner Jay Block, who represents Corrales and Rio Rancho, said: “It’s hard for people here to find high-paying jobs like those at Intel. That’s why this plant is so critical to our community.”
The plant still makes 32-nanometer chips, while Intel’s newer plants are producing 22- and 14-nanometer chips. It was apparently creating newer plants that Intel promised Trump that it would create.So on one hand it is creating jobs in one part of the country and in the other gutting them.
The company is now preparing to produce next-generation 10-nanometer chips, putting the Sandoval County plant far behind the curve. Intel has not given the plant any major upgrades since 2009, when the plant went from 45- to 32-nanometer technology.
AMD has released a new custom “balanced” power plan for those using Ryzen CPU on Windows 10 OS.
Until today, AMD Ryzen CPU users were limited to using the “high performance” plan in Windows 10 OS, at least if they want to get most performance out of their Ryzen CPU. Now, AMD has released a new tweaked “balanced” power plan that should provide a compromise between performance and power efficiency which “automatically balances performance with energy consumption on capable hardware”.
According to the explanation posted by AMD’s Robert Hallock, the new power plan reduces the times and thresholds for P-state transition in order to improve clockspeed ramping as well as disables core parking for “more wakeful cores”.
These tweaks are apparently enough for the new plan to provide similar performance to the Microsoft’s “high performance” power plan setting, at least according to AMD’s own slides. As far as power is concerned, the new balanced power plan does not change how the processor handles low-power idle states, so basically, you’ll get additional performance without compromising the power efficiency.
The new balanced plan is quite simple to install and you can find both the download link as well as check out further explanation over at AMD’s community blog. AMD will also include the final power plan with next AMD chipset drivers for Ryzen CPUs.
After launching the Ryzen 7 CPU lineup, AMD will launch its mainstream Ryzen 5 lineup in just under a week, but today we have additional information about an entry-level Ryzen 3 SKU, the Ryzen 3 1200.
Scheduled to launch sometime in the second half of this year, the Ryzen 3 lineup will compete well against Intel’s Core i3 dual-core lineup. It is still not clear if AMD will include dual-core SKUs in its Ryzen 3 lineup, but it is most likely that all will be quad-core SKUs with and without SMT-enabled. Earlier rumors also suggest that there will be a Ryzen 3 1200X SKU that should be similar but with support for XFR (eXtended Frequency Range) technology, which may give it a further overclocking boost.
According to details leaked by ASRock’s support page and originally spotted by Computerbase.de, the Ryzen 3 1200 SKU works at 3.1GHz frequency (most likely 3.4GHz Turbo) and has a 65W TDP.
Last week, the JEDEC Solid State Technology Association announced that it now has a full standard for the widely anticipated DDR5 memory that is expected to arrive in June 2018, based on new Hybrid DIMM technologies such as NVDIMM-P, which is intended to give servers the ability to store RAM data in between reboots.
DDR5 to use 3D chip stacking with TSVs
As with any new forecasted memory standard, the association says that DDR5 (Double Data Rate 5) memory will offer improved performance with greater power efficiency as compared to previous generation DRAM technologies. DDR5 will provide double the bandwidth and density of DDR4 along with improved channel efficiency, making it ideal for high performance combined with improved power management and cost savings.
DDR5 is expected to become the industry’s first DIMM approach that will include 3D chip stacking using through-silicon vias (TSVs), similar to what Toshiba has been doing with NAND flash since 2015. Since TSVs can be placed anywhere on the chip rather than just at the edge, it’s easy to implement a wide data bus with higher performance and low-power through shorter distances.
NVDIMM-P: Combining NVM or NAND flash with DRAM memory space
Mian Quddus, Chairman of the JEDEC Board of Directors, said that increasing server performance requirements are driving the need for more advanced technologies, and the standardization of next generation memory such as DDR5 and the new generation persistent modules NVDIMM-P will be essential to fulfilling those needs.
The organization has announced that it is also working on a standard for non-volatile hybrid memory called NVDIMM-P, or “Non-Volatile Dual Inline Memory, Persistent,” that basically would map DRAM and NAND to the same memory space. The proposed standard effectively provides both byte- and block-level drive access.
In a move which will get the Nvidia fanboys jolly cross, AMD has said that its new Vega line-up will actually compete with their favourite chip maker in the notebook market.
Nvidia is not used to competition and has been jacking up prices lately for somewhat disappointing chips, but AMD is saying that it thinks its new offerings can force Nvidia to pull its finger out.
At the AMD Tech Summit in Beijing this weekend, AMD vice president Scott Herkelman took the stage to discuss the upcoming Vega-powered graphics cards. He didn’t give anyone a release date but said that AMD’s plans were to put a bit of competition in the notebook GPU market.
AMD plans to decrease the overall footprint of the upcoming mobile GPUs by stacking VRAM dies and freeing up more internal space without sacrificing performance. Size is an important consideration for notebook manufacturers, but this announcement was light on details.
Sadly it was not clear if he was talking about rolling out Vega to discrete mobile or if it will be included in AMD’s “APUs” — a CPU/GPU combo that delivers a smaller overall footprint but a lot less graphical performance.
Herkelman said Vega-powered mobile chips will provide notebook manufacturers with the horsepower they’ll need for their products to drive virtual reality and “the latest and greatest AAA games.” This hints at discrete GPUs powered by the new Vega architecture.
AMD’s previous architecture, code-named “Fiji,” never made much impact in the notebook market, in part due to its power demands. However this could not have been the only reason. Nvidia did well bringing its 10-series GPUs to notebooks despite causing the city lights to dim when anyone plugged it in.
AMD’s Vega-powered GPUs will be available in 4GB and 8GB options, on account of the way the new chips will stack memory. Herkelman told Beijing throngs that Vega-powered chips were “just around the corner.”
Intel has announced the availability of the Intel Xeon processor E3-1200 v6 product family, which includes eight new versions that Intel claims provide key improvements in processor speed, enhanced memory capabilities, hardware-enhanced security and reliability features with support for the latest Intel graphics.
Designed for entry-level servers used by small-business customers and powering entry-level workstations used by business professionals worldwide, the Intel Xeon processor E3-1200 v6 product family supports the increasing workloads and workflows of today’s server and workstation customers.
Intel claims it can provide workstation users with VR capabilities and improved security functions including Intel Software Guard Extensions (Intel SGX) to help protect sensitive data like encryption keys used for communications over the internet from unauthorized access or modification. Intel Authenticate Solution offers customisable multifactor authentication to help protect access to the most sensitive company data.
Intel claims the Intel Xeon processor E3-1200 v6 product family delivers up to 56 per cent performance improvement compared with the Intel Xeon processor E3-1200 v2, which is widely used by small businesses in entry-level workstations today.
The Intel Xeon processor E3-1280 v6 supports up to 3.90 GHz base frequency with Intel Turbo Boost Technology 2.0 frequency up to 4.20 GHz. Additionally, the Intel Xeon processor E3-1200 v6 product family supports enhanced memory speeds, with configurations of robust Error-Correcting Code (ECC) memory up to 64 GB DDR at 2400 MHz.
In addition to enhanced processor performance and support for fast memory, the latest Intel HD graphics P630 is available on some of the range.
The Intel Xeon Processor E3-1200 v6 product family started shipping last month and available for purchase from worldwide OEM and channel providers.
TSMC is currently manufacturing the MediaTek 10nm, deca core based Helio X30 and it looks like in 2018, TSMC might be ready for 7nm and twelve core SoCs from the same house.
Samsung and Qualcomm are already pumping out millions of 10nm SoCs as we speak, and it all looks ready for the March 29 date, ot shall we say today’s introduction of the Samsung Galaxy S8 phone. Samsung’s usual strategy is to ship the phone in the following month making the actual shipping happening at the beginning of Q2 2017. Samsung’s Galaxy S8 and S8+ are just the first of many 10nm based phones to come. Samsung uses both the Snapdagon 835 and the Exynos 8895 both 10nm SoCs for its phones.
MediaTek, on the other hand, is getting ready to start shipping its Helio X30 10nm deca-core and we saw prototype devices at the Mobile World Congress 2017. The SoC looks like it’s much improved compared to last year’s flagship Helio X20.
Recent information implies that the new SoC planned for next year might be the 7nm twelve core and it remains to be seen what the core configuration will be. MediaTek had a hard time fighting the Samsung’s Exynos 8890, the Huawei Kirin 955 / 960 and the Snapdragon 820 last year. The upcoming Exynos 8895 and Snapdragon 835 in 10nm look like very solid performers too. For MediaTek, it remains to be seen if the additional two cores and twelve core design can make a difference. It seems that everyone else in the spectrum stops at eight cores but MediaTek decides that more is better. Just remember, MediaTek was the first company to push the industry from the usual four cores to now a standard eight cores, so it had some good vision at the right time.
The mobile industry managed to get ahead of the rest of the chip industry, at least when it comes to the transition to new manufacturing processes. Snapdragon 835 is the world’s first 10nm SoC developed by the Samsung’s fab while Samsung and MediaTek took their time and only officially announced their 10nm offering roughly a month ago.
Time will tell if it will be realistic to expect 7nm SoC ready in 1H 2018 for the next generation refresh. With Xiaomi making its own, codenamed Pinecone, a Surge S1 branded core and possibly even a higher end core, the pressure will be on manufacturers like MediaTek which relies on China based phone manufacturers. Huawei has been manufacturing a few varieties of Kirin for a while now and sells a lot of phones. There are still a few big names in China including Oppo and Vivo which don’t have their own in house SoC.
Xiaomi wants to compete with some big names such as Huawei, as the big players have their own chips. But regardless of the fact that Xiaomi did a great job appealingto a world population with Hugo Bara’s hire, Oppo is the company which is using mostly Qualcomm and some MediaTek and it remains the number one in China market.
US, Europe and most other developed markets will embrace the big players including Apple, Samsung, Qualcomm and Huawei’s SoC solutions, with little space for the rest of the SoC competition. It will be tough to compete in the future but it will definitely bring some much needed innovation.
It is expecting to make 50 million of the chips before July, according to the Chinese-language Economic Daily News.
For those who came in late the “A11” chips is what is going to be powering the next iPhone which will be launched in September. If you believe the Tame Apple Press the chip and the phone are going to be super, game changing and exciting. Those hacks who have not sold their press credibility are saying that the chip will probably be the only thing that Apple has changed since the iPhone 6.
The chip will will be built on a 10nm FinFET process. TSMC has been making these since the middle of last year and has been shipping them to other customers in the first calendar quarter of 2017.
According to the unreliable Tame Apple Press TSMC is to deliver 100 million “A11” chips before the end of 2017. This figure is important because it slightly ahead of what was delivered by the end of 2017 for the A10 Fusion. This all implies that Apple is expecting to sell more iPhone 8s than iPhone 7s. At this point, we find it extremely doubtful.
Analyst at IDC have added up some numbers and divided them by their shoe size and reached the conclusion that global shipments of augmented and virtual reality headset devices are expected to reach 99.4 million units in 2021.
To put this number into perspective that would be a 10-fold increase from the 10.1 million units shipped in 2016.
The shipment value of AR headsets during the forecast period will grow from $209 million in 2016 to $48.7 billion in 2021. Meanwhile, VR headsets will expand from $2.1 billion in 2016 to $18.6 billion in 2021, IDC tells us.
Most AR headsets are expected to cost well over $1000 which means that the tech is far less accessible to consumers initially, though that’s probably for the best as the AR ecosystem and wide social acceptance are still a few years away, IDC added.
Software king of the world Microsoft is locking down system updates for those using AMD’s Ryzen and Intel’s Kaby Lake processors on Windows 7 and 8.1.
Users are now starting to encounter the following error message: “Your PC uses a processor that isn’t supported on this version of Windows.”
This message appears when a user attempts to update their OS and a quick look at Microsoft’s support page reveals upgrading to Windows 10 is the only way to fix the problem.
Microsoft’s support page on the matter says that Windows 10 is the ‘only’ OS to support these updated hardware configurations. You will need Windows 10 if running Kaby Lake or newer, AMD’s Bristol Ridge or newer (this includes Ryzen), or the Qualcomm 8996 and want to receive important updates to remain secure.
Those who own these chips should not be surprised, and indeed those who spend money on getting the latest chips should probably not be using Windows 7 or 8 anyway. AMD warned that this would be happening in February.
At the time, it said it would not be releasing drivers for Ryzen running on Windows 7. Intel hinted that something similar would happen for Kaby Lake support last year.
The question really is one of ethics. Windows 8.1 won’t hit its end of life until next year, Vole is switching off its support early for new chips.
It has been quite some time since Qualcomm announced Snapdragon X16, the world’s first Gigabit LTE modem. The same GigabitLTE Snapdragon X16 modem is now part of the Snapdragon 835 – a 10nm SoC that is about to debut in a dozen high end phones.
Many people who are not close to the matter are having a hard time to understand why it’s important to get faster modems in an everyday device. Many moan that the speeds they are getting from their carriers are not even touching the Cat 4 maximum speed of 150 Mbps on a download but they are forgetting that these are the best case scenario speeds for Cat 4. What happens is that the average speed increases with new technology as most carriers are now using the Cat 6 300 Mbps maximum speed network.
Today, Telstra in Australia, Sprint in the USA, EE in the UK and a few others have announced or have already deployed their versions of the Cat 16 category GigabitLTE capable of sub 1 Gbps speeds.
It’s a typical technology cat and mouse game. We need faster phones to get the faster internet from carriers. What many people need to understand is that they won’t really get 1 Gbps download speeds as this is a maximum, but the average speed might increase for many.
If you are getting – let’s say – 30 to 60 Mbps today with Cat 6, a Gigabit LTE could increase your speeds to 60 Mbps to 120 Mbps. In our case, in Vienna Austria, we see around 80 Mbps to 100 Mbps, and GigabitLTE could double the speed to 160 Mbps to 200 Mbps. You would need a GigabitLTE phone as well as a GigabitLTE capable network to get to the GigabitLTE speeds. There are two options – the Snapdragon 835 powered phone or the Samsung Exynos 8895. They both support GigabitLTE speeds and the launch of GigabitLTE phones will speed up the deployment of this technology worldwide.
Don’t forget that Samsung Galaxy S8 is likely to ship with both Exynos 8895 and Snapdragon 835, both supporting GigabitLTE speeds.
With the mass introduction of the Snapdragon 835 and Exynos 8895 phones starting with the Samsung Galaxy S8, followed by GigabitLTE deployment by the carriers, we expect that the average download and upload speed will increase, enabling the next generation of content and applications. It looks likely that AT&T, T-Mobile and Sprint are already committed to the GigabitLTE, likely coming this year. Worldwide, there are 15 companies who plan to launch GigabitLTE this year.
If you are one of the skeptical ones that say we don’t need faster internet on the phone, I can remember one very rich man that goes by the name of Bill Gates who wasn’t convinced in the success of the internet. That definitely doesn’t mean that he was right about it, as now even Gates and the rest of the world have the capability of 100s of Mbps speeds on a smartphone device, something that didn’t really exist just a decade ago.
The same performance delta can be associated with internet speed as 3G stopped at 3.6 Mbps / 7.2Mbps. Speed eventually got to 21.6 Mbps with HSPA+. That was some ten years ago and today it is normal to have a Cat 6 LTE 4K network capable of 300 Mbps and, in some cases, advanced carriers get to 600 Mbps, and in the case of Telstra, it even gets to 1Gbps speeds. Qualcomm is planning to ship Snapdragon X20 with 1.2 Gbps maximum speeds in early 2018 and it is already sampling a modem that exceeds GigabitLTE’x magical number.
GigabitLTE with 1Gbps speed is just an introduction to 5G speeds, and it can be viewed as a gateway to 5G. 5G is a new communication technology that will enable a huge technology leap. One of the things that may become a reality is 4K or even 4K 360 video as the default. This will push the need for more and higher resolution VR capable Head Mounted Devices (HMD) and enable new games and applications that we cannot even imagine today.
Think about Facebook live with 360 VR capabilities? We don’t think that this is far off.
Client-grade SSDs of mainstream capacities continue to see rising contract prices in the PC-OEM market during this first quarter.
Beancounters at DRAMeXchange have added up some numbers and divided by their shoe size and worked out that average, contract prices of MLC-based client-grade SSDs are projected to go up by 12-16 percent compared with the fourth quarter of last year, while prices of TLC-based products are expected to increase by 10-16 percent sequentially.
Second quarter, end device sales are anticipated to be relatively flat. Furthermore, PC-OEMs are reaching their limits on SSD costs. While the average prices of mainstream client-grade SSDs will keep climbing, the increase in the second quarter will likely be more moderate.
Alan Chen, senior research manager of DRAMeXchange said that the average contract prices of client-grade SSDs in the PC-OEM market are rising this first quarter because not only PC clients are aggressively stocking up their inventories, smartphone clients are also maintaining strong demand for storage components.
“At the same time, the industry-wide transition to 3D-NAND and 2D-NAND TLC production has sharply reduced the supply of Flash memory of the 2D-NAND MLC type. Thus, the price increase of MLC-based SSDs is outpacing that of TLC-based SSDs.”
Chen added that SSDs are increasingly preferred by consumers due to having faster read/write speed than HDDs, so PC-OEMs will keep up their SSD purchases despite tight supply for NAND Flash and SSDs. In the global notebook market, the SSD adoption rate is estimated to arrive at 45 percent this year.
Additionally, the growth in the notebook SSD adoption will be higher in the consumer-class notebook segment than the business-class segment. On the other hand, the tight NAND Flash supply and sharp price hikes for SSDs will likely discourage PC.
OEMs from raising storage capacity. Therefore, the storage specifications for mainstream PC-OEM SSDs are expected to remain in the 128GB and 256GB options”.
Shipments of client-grade SSDs will drop by seven to eight percent sequentially in the first quarter.