{"id":515,"date":"2024-03-22T15:34:44","date_gmt":"2024-03-22T15:34:44","guid":{"rendered":"http:\/\/www.spintronicfactory.eu\/?p=515"},"modified":"2024-03-22T15:34:44","modified_gmt":"2024-03-22T15:34:44","slug":"a-cmos-integrated-spintronic-compute-in-memory-macro-for-secure-ai-edge-devices","status":"publish","type":"post","link":"http:\/\/www.spintronicfactory.eu\/?p=515","title":{"rendered":"A CMOS-integrated spintronic compute-in-memory macro for secure AI edge devices"},"content":{"rendered":"\n

A group of researchers from Taiwan has developed a cutting edge spintronic non-volatile compute-in-memory (nvCIM) macro compatible with CMOS technology. CIM consists in performing computational operations in the memory units themselves, thus avoiding the need to transfer data between the memory and the processing units. The aim is to combine CIM architecture with spintronics in order to provide a protection against cyber-attacks and data theft for future AI-powered edge computing devices.<\/p>\n\n\n\n

The use of spintronic elements in the device allows non-volatile memory, meaning that it is able to store the data without any power supply. The data protection is ensured by spintronic-based random-and-reliable physically unclonable function (SRR-PUF) which is embedded in the nvCIM macro.<\/p>\n\n\n\n

<\/div>\n\n\n\n
\n
\"\"<\/figure>\n\n\n\n
<\/div>\n\n\n\n
\"\"<\/figure>\n<\/div>\n\n\n\n

Left: die photo of the nvCIM macro. Right: diagrams detailing the structure of the macro and 22 nm spin-transfer torque magnetic random-access memory (STT-MRAM)<\/em>. Credit: National Tsing Hua University (NTHU).<\/p>\n\n\n\n

<\/div>\n\n\n\n

“The 6.6\u2009megabit complementary metal\u2013oxide\u2013semiconductor (CMOS)-integrated macro uses 22\u2009nm spin-transfer torque magnetic random-access memory technology,” the Taiwanese research team explained in their paper. “The macro achieves high randomness (inter-Hamming distance, 0.4999) and high reliability for physically unclonable functionality (intra-Hamming distance, 0), as well as a high energy efficiency for dot-product computation (between 30.1 and 68.0\u2009tera-operations per second per watt).”<\/p>\n\n\n\n

One of the most important feature of this system is its CMOS-compatibility allowing it to co-exist with current semiconductor technology and simplifying its practical integration. The embedding of spintronics into CMOS technology for CIM applications offers significant potential to strengthen safety, enhance performance and drive innovation in AI processing, leading to the development of more advanced and higher-performing devices.<\/p>\n\n\n\n

More information:<\/strong>\u00a0A CMOS-compatible spintronic compute-in-memory macro to secure AI edge devices (techxplore.com)<\/a>. <\/p>\n\n\n\n

Original article:<\/strong>\u00a0A CMOS-integrated spintronic compute-in-memory macro for secure AI edge devices | Nature Electronics.<\/a><\/p>\n","protected":false},"excerpt":{"rendered":"

A group of researchers from Taiwan has developed a cutting edge spintronic non-volatile compute-in-memory (nvCIM) macro compatible with CMOS technology. CIM consists in performing computational operations in the memory units themselves, thus avoiding the need to transfer data between the memory and the processing units. The aim is to combine CIM architecture with spintronics in… Read More »A CMOS-integrated spintronic compute-in-memory macro for secure AI edge devices<\/span><\/a><\/p>\n","protected":false},"author":1,"featured_media":0,"comment_status":"closed","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"passster_activate_protection":false,"passster_protect_child_pages":"","passster_protection_type":"password","passster_password":"","passster_activate_overwrite_defaults":"","passster_headline":"","passster_instruction":"","passster_placeholder":"","passster_button":"","passster_id":"","passster_activate_misc_settings":"","passster_redirect_url":"","passster_hide":"no","passster_area_shortcode":"","neve_meta_sidebar":"","neve_meta_container":"","neve_meta_enable_content_width":"","neve_meta_content_width":0,"neve_meta_title_alignment":"","neve_meta_author_avatar":"","neve_post_elements_order":"","neve_meta_disable_header":"","neve_meta_disable_footer":"","neve_meta_disable_title":"","_themeisle_gutenberg_block_has_review":false,"_ti_tpc_template_sync":false,"_ti_tpc_template_id":"","footnotes":""},"categories":[1],"tags":[20,12,5],"class_list":["post-515","post","type-post","status-publish","format-standard","hentry","category-uncategorized","tag-ai","tag-computing","tag-magnetic-memories"],"_links":{"self":[{"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/posts\/515","targetHints":{"allow":["GET"]}}],"collection":[{"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=515"}],"version-history":[{"count":23,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/posts\/515\/revisions"}],"predecessor-version":[{"id":592,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=\/wp\/v2\/posts\/515\/revisions\/592"}],"wp:attachment":[{"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=515"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=515"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/www.spintronicfactory.eu\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=515"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}