AMD and Intel scientists outline future chip tech

Tri-gate, fully depleted SOI, metal gates and strained silicon lead the way


AMD's boffins continue to explore ways to create ever more powerful processors, and the company this week outlined a couple of the avenues its research is taking. So did its arch-rival, Intel.

Both firms' work is focused on the transistor the key CPU building block. AMD has long touted the benefit of silicon-on-insulator (SOI) designs, and this week said it had measured the fastest PMOS (P-channel metal-oxide semiconductor) transistor speed ever published, equivalent to a 30 per cent speed increase over previously published transistor designs.

AMD's new transistor uses fully depleted SOI (FDSOI), a technique it announced it was working on earlier this year.

SOI works by placing an insulating oxide between a silicon substrate and the silicon that makes up transistor switch. It improves over traditional MOS transistors by reducing the switch's capacity to hold charge - its 'capacitance' - and thus reducing the time it takes for the switch to flip on and off.

When a transistor is switched on, it has to charge up before the current will flow through the switch. The greater the switch's capacitance, the longer that takes. Equally, when the transistor is switched off, current continues to flow until the switch has been emptied of charge. SOI reduces the switch's capacitance significantly, so the transistor operates much more quickly.

SOI transistors are harder to make than traditional MOS transistors because the different physical structures of the silicon and the insulator can lead to electrical imperfections where the two materials meet, slowing the switch down and allowing current to leak. Essentially, the silicon ceases to have a pure crystalline structure, degrading its performance as a semiconductor.

SOI implementations typically use a thick, 'partially depleted' silicon films to overcome these defects - a sort of half-way house between the traditional transistor and the perfect SOI component, which would use thin, 'fully depleted' silicon films.

AMD also said researching strained silicon - a favourite of Intel researchers - but in conjunction with metal gate technology to deliver a 20-25 per cent NMOS (N-channel metal-oxide semiconductor) performance gain over existing strained-silicon transistors.

The metal gates are constructed from Nickel Silicide rather than conventional polysilicon. Essentially, the new material conducts charge better than the old one, speeding the flow of electrons through the transistor. That allows the chip maker to reduce the gate thickness and make it easier to scale transistors down as fabrication processes move to 65nm and below.

Strained silicon involves depositing silicon onto a substrate whose atoms are spaced further apart than silicon atoms usually are. Because atoms inside compounds tend to line up with one another, the silicon atoms move further apart to align themselves with substrate's atoms. This essentially stretches and 'strains' the silicon, the upshot of which is that the resistance to the flow of electrons is reduced, speeding electron flow by up to 70 per cent. Transistors made from strained silicon get a 35 per cent performance boost over regular transistors of the same size.

Adding metal gate technology provides a further 20-25 per cent increase in performance, AMD's scientists say.

They also applied metal gates to their FDSOI transistor, contributing to the 30 per cent performance gain they experienced there.

Intel, meanwhile, has demonstrated a 30nm tri-gate transistor. The transistor's gate controls the flow of current through the device - it's what allows the transistor to act as a switch and thus combine with other transistors to form logic gates, from which computational devices can be built.

Traditional 'planar' transistors use a single gate sitting on top of the silicon (which itself sits on top of the substrate). Intel's tri-gate transistor has gate material on each side of the silicon as well as on top, like a saddle over the silicon's back. Because the gate only controls current through the layers of silicon it's closest to, having gate material on three sides of the silicon activates current through more of the material than on one side alone, the top. That improves the ability of the transistor to switch on quickly, and thus its performance.

Intel announced tri-gate transistors last year, but this week it said it had scaled the device down to 30nm. The company was demonstrating a 60nm device this week, but it can show the device can be easily reduced to half that size. Work on the transistor has reached the point where the company can begin developing it for commercial use, it said, though it put no timescale on it. But fabbing tri-gate transistors can be done using the same processes used to fab today's transistors. ®


Other stories you might like

  • HPE unveils Arm-based ProLiant server for cloud-native workloads
    Looks like it went with Ampere – which means a certain Reg writer lost a bet

    Arm has a champion in the shape of HPE, which has added a server powered by the British chip designer's CPU cores to its ProLiant portfolio, aimed at cloud-native workloads for service providers and enterprise customers alike.

    Announced at the IT titan's Discover 2022 conference in Las Vegas, the HPE ProLiant RL300 Gen11 server is the first in a series of such systems powered by Ampere's Altra and Altra Max processors, which feature up to 80 and 128 Arm-designed Neoverse cores, respectively.

    The system is set to be available during Q3 2022, so sometime in the next three months, and is basically an enterprise-grade ProLiant server – but with an Arm CPU at its core instead of the more usual Intel Xeon or AMD Epyc X86 chips.

    Continue reading
  • US weather forecasters power up latest supercomputers to keep you out of the rain
    NOAA makes it rain for HPE, AMD

    Predicting the weather is a notoriously tricky enterprise, but that’s never held back America's National Oceanic and Atmospheric Administration (NOAA). After more than two years of development, the agency brought a pair of supercomputers online this week that it says will enable more accurate forecast models.

    Developed and maintained by General Dynamics Information Technology (GDIT) under an eight-year contract, the Cactus and Dogwood supers — named after the fauna native to the machines' homes in Phoenix, Arizona, and Manassas, Virginia, respectively — will support larger, higher-resolution models than previously possible. The cost to build, house, and support and operate these machines, now operational, will cost $150 million over the next five years, we understand.

    “People are looking for the best possible weather forecast information that they can get,” Brian Gross, director of the Environmental Modeling Center for the National Weather Service, told The Register.

    Continue reading
  • Google said to be taking steps to keep political campaign emails out of Gmail spam bin
    Just after Big Tech comes under fire for left and right-leaning message filters

    Google has reportedly asked the US Federal Election Commission for its blessing to exempt political campaign solicitations from spam filtering.

    The elections watchdog declined to confirm receiving the supposed Google filing, obtained by Axios, though a spokesperson said the FEC can be expected to publish an advisory opinion upon review if Google made such a submission.

    Google did not immediately respond to a request for comment. If the web giant's alleged plan gets approved, political campaign emails that aren't deemed malicious or illegal will arrive in Gmail users' inboxes with a notice asking recipients to approve continued delivery.

    Continue reading
  • China is trolling rare-earth miners online and the Pentagon isn't happy
    Beijing-linked Dragonbridge flames biz building Texas plant for Uncle Sam

    The US Department of Defense said it's investigating Chinese disinformation campaigns against rare earth mining and processing companies — including one targeting Lynas Rare Earths, which has a $30 million contract with the Pentagon to build a plant in Texas.

    Earlier today, Mandiant published research that analyzed a Beijing-linked influence operation, dubbed Dragonbridge, that used thousands of fake accounts across dozens of social media platforms, including Facebook, TikTok and Twitter, to spread misinformation about rare earth companies seeking to expand production in the US to the detriment of China, which wants to maintain its global dominance in that industry. 

    "The Department of Defense is aware of the recent disinformation campaign, first reported by Mandiant, against Lynas Rare Earth Ltd., a rare earth element firm seeking to establish production capacity in the United States and partner nations, as well as other rare earth mining companies," according to a statement by Uncle Sam. "The department has engaged the relevant interagency stakeholders and partner nations to assist in reviewing the matter.

    Continue reading
  • California's attempt to protect kids online could end adults' internet anonymity
    Websites may be forced to verify ages of visitors unless changes made

    California lawmakers met in Sacramento today to discuss, among other things, proposed legislation to protect children online. The bill, AB2273, known as The California Age-Appropriate Design Code Act, would require websites to verify the ages of visitors.

    Critics of the legislation contend this requirement threatens the privacy of adults and the ability to use the internet anonymously, in California and likely elsewhere, because of the role the Golden State's tech companies play on the internet.

    "First, the bill pretextually claims to protect children, but it will change the Internet for everyone," said Eric Goldman, Santa Clara University School of Law professor, in a blog post. "In order to determine who is a child, websites and apps will have to authenticate the age of ALL consumers before they can use the service. No one wants this."

    Continue reading
  • Is computer vision the cure for school shootings? Likely not
    Gun-detecting AI outfits want to help while root causes need tackling

    Comment More than 250 mass shootings have occurred in the US so far this year, and AI advocates think they have the solution. Not gun control, but better tech, unsurprisingly.

    Machine-learning biz Kogniz announced on Tuesday it was adding a ready-to-deploy gun detection model to its computer-vision platform. The system, we're told, can detect guns seen by security cameras and send notifications to those at risk, notifying police, locking down buildings, and performing other security tasks. 

    In addition to spotting firearms, Kogniz uses its other computer-vision modules to notice unusual behavior, such as children sprinting down hallways or someone climbing in through a window, which could indicate an active shooter.

    Continue reading

Biting the hand that feeds IT © 1998–2022