SamMobile has affiliate and sponsored partnerships. If you buy something through one of these links, we may earn a commission.

Notifications
    News For You

    Samsung becomes the world’s first brand to demonstrate MRAM technology for in-memory computing

    General
    By 

    Last updated: January 13th, 2022 at 10:06 UTC+01:00

    Samsung, the world's biggest memory chip maker, has announced that it has become the first in the world to demonstrate the MRAM (Magnetoresistive Random Access Memory) technology for in-memory computing. The company's paper on its innovation, titled ‘A crossbar array of magnetoresistive memory devices for in-memory computing,' was published by Nature on its website, and its print edition is coming next.

    The research was a collaboration between SAIT (Samsung Advanced Institute of Technology), Samsung Foundry, and Samsung Semiconductor R&D. Dr. Seungchul Jung (Staff Researcher at SAIT), Dr. Donhee Ham (Fellow of SAIT and Professor of Harvard University), and Dr. Sang Joon Kim (Vice President of Technology at SAIT) worked on the research.

    Usually, data is stored in memory chips (DRAM) and processed by a CPU or an AP (Application Processor). However, data storage and computing happen on the same chip with in-memory computing. Since there is no need to transfer data from memory to the processor and vice versa, a lot of time is saved. Data processing inside the memory happens in a highly parallel manner, resulting in substantial power savings. Samsung claims that MRAM technology will be great for things like AI processing.

    Other in-memory computing solutions like PRAM (Phase-change Random Access Memory) and RRAM (Resistive Random Access Memory) have been demonstrated worldwide, but MRAM has been difficult to prove. Despite MRAM's non-volatile nature, high operational speed, and endurance, it was difficult to use it for in-memory computing due to its low resistance. It was also impossible to use MRAM's power efficiency advantage in a standard in-memory computing architecture.

    Samsung has now created a solution to the problem through its new architecture innovation. The South Korean firm developed an MRAM array chip that uses ‘Resistance Sum' instead of the standard ‘Current-Sum.' The new solution counters the problem of small resistances of individual MRAM devices.

    The company's new MRAM in-computing was tested by running AI applications, and it offered 98% accuracy in the classification of handwritten digits and 93% accuracy in detecting faces in images. Samsung says that its new technology can be used to make highly power-efficient AI chips.

    Dr. Seungchul Jung, the first author of the paper, said, “In-memory computing draws similarity to the brain in the sense that in the brain, computing also occurs within the network of biological memories, or synapses, the points where neurons touch one another. In fact, while the computing performed by our MRAM network for now has a different purpose from the computing performed by the brain, such solid-state memory network may in the future be used as a platform to mimic the brain by modeling the brain's synapse connectivity.

    Join SamMobile's Telegram group and subscribe to our YouTube channel to get instant news updates and in-depth reviews of Samsung devices. You can also subscribe to get updates from us on Google News and follow us on Twitter.

    General MRAMSamsung FoundrySamsung Semiconductor
    Galaxy AI summarized

    Scroll for more related content
    News For You

    You might also like

    Samsung will be ready to make 2nm mobile chips in 2025

    Samsung will be ready to make 2nm mobile chips in 2025

    Samsung Foundry unveiled its updated semiconductor chip process roadmap today. During the ongoing SFF 2024 expo in the US, the South Korean semiconductor chip giant unveiled its process roadmap for 2nm and 3nm chips and plans for specialized nodes for AI and automotive chips. Samsung Foundry plans to make 2nm chips in 2025 and 1.4nm […]

    • By Asif Iqbal Shaik
    • 2 days ago
    Qualcomm interested in Samsung’s 2nm process for Snapdragon 8 Gen 5

    Qualcomm interested in Samsung’s 2nm process for Snapdragon 8 Gen 5

    Over the past few years, Qualcomm has shifted away from Samsung Foundry for the fabrication of its high-end Snapdragon chips. However, the company is keen on collaborating with Samsung Foundry again for future chips, such as the Snapdragon 8 Gen 5, which will likely be released in 2025. Qualcomm CEO keen on using Samsung Foundry […]

    • By Asif Iqbal Shaik
    • 1 week ago
    Samsung could announce plans for 1nm chips next month

    Samsung could announce plans for 1nm chips next month

    Samsung Foundry could announce its plans for 1nm chips next month during its SAFE Forum event in the USA. According to Taiwanese media, the South Korean firm is also expected to announce mass production plans for 2nm chips during the event. Samsung could start production of 1nm chips as soon as 2026 Samsung Foundry is […]

    • By Asif Iqbal Shaik
    • 2 weeks ago
    New Samsung Semiconductor chief calls for unity among employees

    New Samsung Semiconductor chief calls for unity among employees

    A couple of weeks ago, Samsung appointed a new head of its semiconductor chip business. Jun Young-hyun, who is now heading the South Korean firm's most important business arm, has called for unity among employees to stay on top of the competition. Jun Young-hyun, Samsung Device Solutions' new chief, calls for unity between employees and […]

    • By Asif Iqbal Shaik
    • 2 weeks ago
    AMD rumored to use Samsung Foundry for its 3nm chips

    AMD rumored to use Samsung Foundry for its 3nm chips

    While Samsung Foundry became the first company in the world to start manufacturing 3nm chips in late 2022, no big-name client used its process node. Some rumors even claimed it to be a doomed technology. However, things might be improving for the company, as AMD is rumored to use Samsung's 3nm GAA process node for […]

    • By Asif Iqbal Shaik
    • 2 weeks ago
    Google drops Samsung to go with TSMC for Pixel 10’s Tensor chip

    Google drops Samsung to go with TSMC for Pixel 10’s Tensor chip

    Since Google made its first Tensor processor for Pixel devices, its chips have been manufactured by Samsung Foundry. After four generations, though, the company is ditching Samsung Foundry to go with TSMC to fabricate its next-generation Tensor chip, the Tensor G5, which will be used in the Pixel 10 series. More information about Google's switch […]

    • By Asif Iqbal Shaik
    • 3 weeks ago