65 nanometers is the next goal of the Zelenograd plant Angstrem-T, which will cost 300-350 million euros. The company has already submitted an application for a preferential loan for the modernization of production technologies to Vnesheconombank (VEB), Vedomosti reported this week with reference to the chairman of the board of directors of the plant, Leonid Reiman. Now Angstrem-T is preparing to launch a production line for microcircuits with a 90nm topology. Payments on the previous VEB loan, for which it was purchased, will begin in mid-2017.
Beijing crashes Wall Street
Key American indices marked the first days of the New Year with a record drop; billionaire George Soros has already warned that the world is facing a repeat of the 2008 crisis.
The first Russian consumer processor Baikal-T1, priced at $60, is being launched into mass production
The Baikal Electronics company promises to launch into industrial production the Russian Baikal-T1 processor costing about $60 at the beginning of 2016. The devices will be in demand if the government creates this demand, market participants say.
MTS and Ericsson will jointly develop and implement 5G in Russia
Mobile TeleSystems PJSC and Ericsson have entered into cooperation agreements in the development and implementation of 5G technology in Russia. In pilot projects, including during the 2018 World Cup, MTS intends to test the developments of the Swedish vendor. At the beginning of next year, the operator will begin a dialogue with the Ministry of Telecom and Mass Communications on the formation of technical requirements for the fifth generation of mobile communications.
Sergey Chemezov: Rostec is already one of the ten largest engineering corporations in the world
The head of Rostec, Sergei Chemezov, in an interview with RBC, answered pressing questions: about the Platon system, the problems and prospects of AVTOVAZ, the interests of the State Corporation in the pharmaceutical business, spoke about international cooperation in the context of sanctions pressure, import substitution, reorganization, development strategy and new opportunities in difficult times.
Rostec is “fencing itself” and encroaching on the laurels of Samsung and General Electric
The Supervisory Board of Rostec approved the “Development Strategy until 2025”. The main objectives are to increase the share of high-tech civilian products and catch up with General Electric and Samsung in key financial indicators.
Long years Intel company was ahead of the rest in terms of the pace of implementation of advanced technical processes for the production of complex microcircuits (read processors). It stalled during the introduction of the 14nm process technology. Problems with the introduction of the 14-nm process technology were aggravated by the fact that the PC market stopped showing positive dynamics. For several years now, instead of annual growth, we have seen a decrease in sales volumes. A decrease in revenue automatically leads to a reduction in funding for development and modernization of production, which causes the manufacturer to have an irresistible desire to exploit what has already been created and not rush into innovation.
The lack of adult competition also does not contribute to moving forward, for which we can say “thank you” to you know who. Everything taken together at this stage leads us to the fact that the 14nm process technology for Intel is workhorse for a year. The expected 10nm process technology and, in particular, Cannonlake processors will not make a difference in the market. Having been burned by the difficult implementation of 14-nm technological standards, Intel will be struggling for a long time - slowly and limitedly switching to the production of 10-nm solutions. We were disappointed that Intel was unable to begin releasing 10nm processors in the middle of this year, as the company's tick-tock strategy and early plans dictated. Now it looks like we'll have to get used to the idea that there won't be a 10nm Intel CPU in 2017 (which has already been decided) or even in 2018.
Japanese sources, citing OEM manufacturers, shared the news that Intel is developing another 14 nm processors. As you know, in the fourth quarter of this year the company will bring the second generation of 14-nm to the market Skylake processors- processors Kaby Lake(third 14nm after Broadwell). Kaby Lake processors will replace Skylake in all product categories. In the fourth quarter of 2017, the company's first 10nm processors are expected to be released - solutions based on the Cannonlake architecture. But these processors, if you believe the latest leaks, will not become mass solutions anytime soon. Most likely this will happen no earlier than 2019. Because in 2018, Kaby Lake is promised to be replaced by 14nm processors Coffee Lake.
We first heard about Coffee Lake processors in April of this year according to information from the profile of one of the Intel employees in one of social networks to search for vacancies. Then there was speculation that this was the name of one of the 10nm or even 7nm Intel processors. Today we can say with some degree of confidence that these will be the company’s next “optimized” 14nm processors.
Coffee Lake processors will be on the market at the same time as 10nm Cannonlake processors. The latter will be released for thin laptops and tablet-like systems in low-end configurations in the U and Y series with TDPs from 15 W to 4.5 W. Everything above - U to H - will be processors based on the Coffee Lake architecture. These are massive and productive systems with the number of cores from two to six. The integrated graphics core of Coffee Lake processors will also be a class higher than that of Cannonlake: GT3e instead of GT2 for Cannonlake. This information makes you imagine that the 14nm process technology for Intel is a long-term process. However, we are repeating ourselves. Just like Intel...
]| Process Name | |
|---|---|
| 1st Production | |
| Lithography | Lithography |
| Immersion | |
| Exposure | |
| Wafer | Type |
| Size | |
| Transistor | Type |
| Voltage | |
| Fin | Pitch |
| Width | |
| Height | |
| Gate Length (Lg) | |
| Contacted Gate Pitch (CPP) | |
| Minimum Metal Pitch (MMP) | |
| SRAM bitcell | High-Perf (HP) |
| High Density (HD) | |
| Low-Voltage (LV) | |
| DRAM bitcell | eDRAM |
| Intel | Samsung Alliance | IBM (Now GlobalFoundries) | UMC | Common Platform Alliance | |||||
|---|---|---|---|---|---|---|---|---|---|
| P1272 (CPU) / P1273 (SoC) | 14LPE 1st generation; 14 nm Low Power Early , 14LPP2nd generation; 14 nm Low Power Performance , 14LPC3rd generation; 14 nm Low Power Cost , 14LPU4th generation; 14 nm Low Power Ultimate |
14HP 14nm High Performance |
14FDSOI | ||||||
| 2014 | 2015 | 2017 | 2Q 2017 | ||||||
| 193 nm | 193 nm | 193 nm | 193 nm | 193 nm | |||||
| Yes | Yes | Yes | Yes | Yes | |||||
| SADP | LELE | SADP | D.P. | ||||||
| Bulk | Bulk | SOI | Bulk | SOI | |||||
| 300 mm | 300 mm | 300 mm | 300 mm | 300 mm | |||||
| FinFET | FinFET | FinFET | FinFET | Planar | |||||
| 0.70 V | 0.80 V | 0.80 V | 0.80 V | ||||||
| Value | 22 nm Δ | Value | 20 nm Δ | Value | 22 nm Δ | Value | 28 nm Δ | Value | 28 nm Δ |
| 42 nm | 0.70x | 48 nm | N/A | 42 nm | N/A | N/A | |||
| 8 nm | 1.00x | 8 nm | 10 nm | ||||||
| 42 nm | 1.24x | 37 nm | 25 nm | ||||||
| 20 nm | 0.77x | 30 nm | 18-26 nm | 0.72-0.79x | 20 nm | 0.71x | |||
| 70 nm | 0.78x | 78 nm | 1.22x | 80 nm | 0.80x | 90 nm | 0.79x | ||
| 52 nm | 0.65x | 64 nm | 1.00x | 64 nm | 0.80x | 64 nm | 0.71x | ||
| 0.0706 µm² | 0.54x | 0.080 µm² | 0.78x | 0.0900 µm² | 0.63x | 0.090 µm² | 0.59x | ||
| 0.0499 µm² | 0.54x | 0.064 µm² | 0.79x | 0.0810 µm² | 0.81x | 0.081 µm² | 0.68x | ||
| 0.0588 µm² | 0.54x | ||||||||
| 0.0174 µm² | 0.67x | ||||||||
Composition [edit]
It"s important to note that not all processes compete with each other. The process should cater to the products that will make use of the underlying technology. The composition of the actual integrated circuit also varies by manufacturer and by design due to different goals. For example, the cache on Apple "s 14 nm (manufactured by Samsung) accounts almost 1/3 of the entire chip whereas Intel "s Broadwell cache accounts for only 10% of the entire chip. Likewise, Intel "s Broadwell and Skylake target high-performance and incorporate a large amount of higher-speed elements which are inherently sparse. Tall cells account for almost 30% Skylake's composition and less than 1% on Apple's or . Those numbers are somewhat expected given tall logic cells are generally optimized for performance and high frequency (e.g., high-switching circuitry in the CPU) whereas short cells are optimized for density (e.g., GPU shader arrays).
A third improved process, "14nm++", is set to begin in late 2017 and will further allow for +23-24% higher drive current for 52% less power vs the original 14nm process. The 14nm++ process also appear to have slightly relaxed poly pitch of 84 nm (from 70 nm). It "s unknown what impact, if any, this will have on the density.
| Intel 14nm Design Rules | ||
|---|---|---|
| Layer | Pitch | Scale Factor |
| Fin | 42 nm | 0.70 |
| Contacted Gate Pitch | 70 nm | 0.78 |
| Metal 0 | 56 | - |
| Metal 1 | 70 | 0.78 |
| Metal 2 | 52 | 0.65 |
IBM [edit]
IBM developed their own "14HP" (14nm High-Performance) process at their East Fishkill, NY plant. Note that the plant AND the process, along with numerous semiconductor technology IPs, were sold to GlobalFoundries in late 2014. GF still operates the plant (also by ex-IBM semiconductor engineers) and the process which is used by IBM for their various processors. This process was designed by IBM for their very large chips with effective power supply and clock distribution capable of producing dies as large as 700 mm² and larger with a hierarchical BEOL of 17 levels of copper interconnect for high performance wire-ability. It should be noted that GlobalFoundries had no such capabilities prior to their acquisition of IBM's plant, semiconductor manufacturing group, and IP portfolio.
UMC [edit]
UMC announced the start of 14nm process mass production in February 2017. The 14nm process is their first process to use FinFET, and provides up to 55% higher performance and twice the gate density compared to their 28nm process.
All modern computing technologies are based on semiconductor electronics. For its production, silicon crystals are used - one of the most common minerals in our planet. Since the demise of bulky tube systems and the development of transistor technology, this material has occupied an important place in the production of computer equipment.
Central and graphic processors, memory chips, various controllers - all this is produced on the basis of silicon crystals. For half a century, the basic principle has not changed, only chip creation technologies are being improved. They are becoming thinner and more miniature, energy efficient and productive. The main parameter that will be improved is the technical process.
Almost all modern chips consist of silicon crystals, which are processed by lithography to form individual transistors. A transistor is a key element of any integrated circuit. Depending on the state of the electric field, it can transmit a value equivalent to a logical one (passes current) or zero (acts as an insulator). In memory chips, data is written using combinations of zeros and ones (transistor positions), and in processors, calculations are performed when switching.
In 14-nm technology (compared to 22-nm), the number of barriers is reduced, their height is increased, and the distance between the dielectric fins is reduced
A technological process is a procedure and procedure for manufacturing any product. In the electronics industry, in its generally accepted meaning, this is a value that indicates the resolution of the equipment used in the production of chips. The size of the functional elements obtained after silicon processing (that is, transistors) also directly depends on it. The more sensitive and accurate the equipment used for processing crystals for processor blanks, the finer the technical process will be.
What does the numerical value of a technical process mean?
In modern semiconductor manufacturing, the most common method is photolithography - etching elements on a chip coated with a dielectric film using light. It is the resolution of the optical equipment that emits light for etching that is the technical process in the generally accepted interpretation of the word. This number indicates how thin the feature on the chip can be.
What does the technical process affect?
The technical process directly affects the quantity active elements semiconductor chip. The thinner the technical process, the more transistors will fit on a certain area of the chip. First of all, this means increasing the number of products from one piece. Secondly, reducing energy consumption: the thinner the transistor, the less energy it consumes. As a result, with an equal number and arrangement of transistors (and therefore an increase in performance), the processor will consume less energy.
The downside of switching to a fine technical process is that the equipment will become more expensive. New industrial units make it possible to make processors better and cheaper, but they themselves increase in price. As a consequence, only large corporations can invest billions of dollars in new equipment. Even such well-known companies as AMD, Nvidia, Mediatek, Qualcomm or Apple do not make processors themselves, entrusting this task to giants like TSMC.
What does reducing the technical process give?
By reducing the technological process, the manufacturer has the opportunity to increase performance while maintaining the same chip dimensions. For example, the transition from 32 nm to 22 nm made it possible to double the transistor density. As a result, on the same chip as before, it became possible to place not 4, but already 8 processor cores.
For users, the main benefit is reduced energy consumption. Chips using a thinner process technology require less energy and generate less heat. Thanks to this, you can simplify the power system, reduce the cooler, and pay less attention to blowing components.
Processor technology on smartphones
Smartphones are demanding on hardware resources and quickly drain battery power. Therefore, to slow down the discharge consumption, processor developers for mobile devices We are trying to introduce the latest technical processes into production. For example, the once popular dual-core MediaTek MT6577 was produced using a 40 nm process technology, and Qualcomm Snapdragon 200 early series were manufactured using 45 nm technology.
In 2013-2015, 28 nm became the main technological process for chips used in smartphones. MediaTek (up to and including Helio X10), Qualcomm Snapdragon S4, 400 series, as well as models 600, 602, 610, 615, 616 and 617 are all 28 nm. It was also used in the manufacture of Snapdragon 650, 652, 800, 801, 805. The “hot” Snapdragon 810, interestingly, was made using a thinner 20 nm process technology, but this did not help it much.
Apple also used 20nm technology in its A7 (iPhone 5S). The Apple A8 for the sixth iPhone used 20 nm, and the A9 model (for 6s and SE) already uses the new 16 nm technological process. In 2013-2014, Intel made their Atom Z3xxx using 22-nanometer technology. Since 2015, chips with 14 nm have been launched into production.
The next step in the development of processors for smartphones is the widespread development of 14 and 16 nm technological processes, and then we can expect 10 nm. The first copies on it may be Qualcomm Snapdragon 825, 828 and 830.
Another leaked Intel processor roadmap hints that the company will not be able to begin production of 10 nm CPUs until at least the end of 2020.
According to this roadmap, the company plans to release two model range desktop processors. These will be Core S and X series processors.
High-end Cascade Lake X models will offer up to 18 cores. Now there is already a similar processor - Core i9-9980XE, but the new generation is expected to receive some improvements.
The regular S series will be represented by Comet Lake S. It will appear at the end of this year and the beginning of next year. These processors will have up to 10 cores, which will be useful for content creators and gamers.
Despite the increase in the number of cores, and therefore performance, the new processors will not be able to provide greater energy efficiency, since they will continue to be produced using 14 nm technology. Apparently, desktop chips manufactured at 10 nm can only be expected from Intel in 2021. Meanwhile, AMD will release 7 nm processors in the summer.
MRAM memory ready for production
2nd of MarchIntel is ready to start producing MRAM memory in large volumes. This type of memory was developed by Intel and is non-volatile memory, meaning it can be used for storing data and not just as RAM.
Magnetoresistive random access memory was created as a universal replacement for DRAM (volatile) and NAND (non-volatile) memory. It has now become very difficult to reduce element sizes when producing these types of memory, and MRAM does not have such strict restrictions. In addition, MRAM has a much higher yield of usable chips during production. Thus, with 22 nm production technology, the level of usable microcircuits on a pancake is 99.9% - the amazing reliability of the technology.
Additionally, MRAM already exhibited a setup time of 1 ns, which is higher than the theoretical limit for DRAM. Write speeds are also several thousand times faster than NAND. MRAM also guarantees 10 years of data storage at a temperature of 200° C and reliability of 10 6 switching cycles. All this was reported by Ligion Vei, an Intel engineer.
It seems that the production of MRAM memory will begin using 22 nm technology, although it is noted that Intel has already begun to unload 14 nm factories, so a quick transition to a thinner technical process is possible.
Intel is preparing Comet Lake-S using a 14 nm process
November 28, 2018Rumors have appeared on the Internet about the continued use of the 14 nm process technology by Intel and the new Comet Lake-S processor architecture.
Intel is fighting hard against AMD with its multi-core processors, the third generation of which is already being manufactured using 7 nm technology.
However, Intel still doesn't have 10nm technology and is forced to squeeze all the juice out of existing hardware. The new Comet Lake-S architecture will be produced using 14 nm++ technology. The flagship chip will have a 10C/20T formula. Given this formula, frequencies should be reduced compared to the 8-core i9-9900K, which works great above 5 GHz, although it consumes a lot of power.
Time will tell how successfully Comet Lake-S will be able to counter Zen 2.
AMD Radeon RX 590 appeared in the 3DMark database
October 17, 2018Lately there have been rumors on the Internet that AMD is about to release new Polaris 30 video cards. Some are talking about the 600 series of accelerators, but now that the Radeon RX 590 card has appeared in 3DMark, many are talking about what it is Polaris 30.
The Radeon RX 590 video card has appeared in the database of the popular 3DMark benchmark. Its GPU is no different from the Polaris 20, made according to 14 nm standards. The only difference is the 12 nm process technology. The move to thinner elements allows AMD to create some thermal headroom by overclocking the processor. The GPU in the Radeon RX 590 runs on clock frequency 1545 MHz, which is 205 MHz higher than the Boost frequency of the Radeon RX 580.
When the Radeon RX 590 will go on sale is still unknown, but it is obvious that AMD simply needs to respond in some way to NVIDIA’s releases.
Intel is forced to return to the 22 nm process
October 13, 2018Trying to fulfill all orders for 14 nm production, Intel is forced to make compromises. Considering that the 10 nm process is far from ready, the company simply has no alternative but to transfer some products to outdated technologies.
These products include the H310 chipsets, which will now become larger. This decision is quite understandable. The fact is that H310 is the simplest system logic chip designed to work with 8th and 9th generation Core processors. Motherboards built on these chipsets are used in office machines and simple consumer machines, for which its modest capabilities are sufficient. Considering the low requirements for the chip, Intel decided to produce them using 22 nm technology.
According to Chinese sources, the new chipset is called H310C. Its dimensions are 10x7 mm, while the usual 14 nm H310 chip has dimensions of 8.5x6.5 mm. The heat dissipation of the original chip was 6 W, and due to the change in production technology, its increase is not expected. It is also not expected that changing the chip will affect the design motherboards.
Intel expands 14 nm production
October 4, 2018Faced with a shortage of production capacity caused by the failure of the 10 nm production launch, Intel nevertheless decided to expand its production, subject to pressure from AMD.
Considering that the company's new flagship processors Core i9-9900K, Core i7-9700K and Core i5-9600K will be released on October 8, Intel saw no other way to open another production site in Vietnam.
The company's press release states: “In order to ensure continued supply of processors...Intel will add additional manufacturing facilities for pre-production/finished products. The new zone is located in Vietnam. The new production area has become a certified equivalent (in shape, size, function and reliability) to the company’s products and technologies.”.
September 11, 2018Amid difficulties with the production volumes of chips using the 14 nm process, Intel decided to look for third-party manufacturers.
The information resource Digitimes reports that Intel has decided to maintain production of its highly profitable processors, in the main servers, and chipsets for them. Other, inexpensive products, such as chipsets entry level H310 and other desktop chips of the 300 series will be outsourced to TSMC.
Intel has determined that the volume of shortfalls is now 50%, and therefore the only way out of this situation will be outsourced production, since the company does not want to increase its own capacity.
Now TSMC is already a contract manufacturer for Intel, producing for it the SoFIA series of system-on-chips and some FPGA products, as well as communication chips for the iPhone.
Motherboard manufacturers believe that the shortage of 14 nm chipsets from Intel will decrease by the end of 2018.
GlobalFoundries stops working on 7nm
September 8, 2018Microprocessor manufacturer GlobalFoundries has made an important announcement about a change in company strategy.
The contract chip manufacturer has announced it is winding down work on the 7LP (7nm) process. Instead, it will focus on the 14LPP/12LP platform, and will produce RF devices, embedded memory and other low-power devices. Against the background of the closure of 7 nm development, GloFo will reduce 5% of its staff, and will also terminate licensing agreements with AMD and IBM.
GlobalFoundries CTO Gary Patton reported that the first 7 nm chips will be available to customers in the fourth quarter of this year, however "A few weeks ago" the company decided to make a sharp strategic pivot.
He noted that the decision was not based on technical difficulties faced by the company, but on exploring the business opportunities that opened up for the 7LP platform, as well as financial considerations.
Against the backdrop of this announcement, AMD announced that it was transferring all of its 7 nm processors, CPU and GPU, to production at TSMC.
Intel will not release 10 nm processors until the winter holidays of 2019
July 30, 2018Let's imagine 2019. AMD is selling 7 nm processors with all its might, while Intel still only offers 14 nm solutions. This is approximately how sad Intel's prospects for 10 nm technology look.
In a question and answer session following financial results for the second quarter of 2018, Intel stated that the first product based on the 10 nm process will not appear until the winter holidays of 2019. This means that Intel will retain the 14 nm process not only until the end of 2018, but also for almost the entire 2019.
In the client segment, Intel is preparing to release the 9th generation Core processors called Whiskey Lake. This will be the 5th generation of processors manufactured using the 14 nm process. The first four were Broadwell, Skylake, Kaby Lake and Coffee Lake.
Most likely, Intel will move into 2019 with Whiskey Lake, competing with AMD's 12 nm Pinnacle Ridge and increasing the number of cores. Intel is also losing in the HEDT segment, releasing 20 and 22-core processors with LGA2066, as well as preparing a 28-core chip. By the end of 2019, AMD is preparing to release the 3rd generation of EPYC processors, already using 7 nm technology. Ryzen processors will be produced according to 10 nm standards.
In the picture you can see an Intel slide dated 2013. It shows Intel's plans to release 10 nm technology in 2015. This image perfectly demonstrates how plans can diverge from reality.
Intel refused to move Ice Lake to 14 nm process technology
May 21, 2018It is now known to everyone that Intel cannot get the 10 nm process up and running. However, management has been aware of the difficulties for a long time. Former company engineer Francois Pidnoeul said that the company had a chance to port the Ice Lake architecture to the 14 nm++ process. Pidnoyol proposed such a step two years ago, but management rejected it. The lack of this backup plan has led to a noticeable slowdown in processor development as difficulties in the manufacturing division delay Intel's entire evolution.
Management had the opportunity to provide the Ice Lake design (ICL in the tweet) on 14 nm technology, but they decided that it was not necessary. Perhaps they did this because they were confident that two years was enough to set up 10 nm production technology. Management made the wrong bet, and now Intel's product portfolio is suffering because of it.