<\/p>\nfrom Huawei\u2019s presentation:<\/p>\n
<\/a><\/p>\n
The Ascend series is Huawei\u2019s main competitive offering to Nvidia\u2019s artificial intelligence (AI) chips, and its current chip on the market is the Ascend 910C.<\/p>\n
Huawei is planning three new series over the next three years, namely the Ascend 950, 960 and 970, with the 950 to be launched in the first quarter of next year.<\/p>\n
There will be two types of Ascend 950 chips, the 950PR and 950DT, with the former suitable for the prefill stage of inference and to run recommendations, while the latter will be optimised for the decode stage of inference and model training.<\/p>\n
The Ascend 960 will have twice the computing power and memory capacity of the 950, and the company aims to go even higher with the 970.<\/p>\n
Huawei, which is barred from working with the world\u2019s top chip foundry Taiwan Semiconductor Manufacturing Co by US export controls, did not say who will manufacture its chips. But analysts say the company works with chip equipment suppliers and China\u2019s largest foundry, Semiconductor Manufacturing International Corp.<\/p>\n
Huawei said it now has its own proprietary high-bandwidth memory, technology currently dominated by South Korea\u2019s SK Hynix and Samsung Electronics.<\/p>\n
The Ascend 950PR will integrate Huawei\u2019s HBM chip, called HiBL 1.0.<\/p>\n
Huawei says its HBM chip is more cost-effective than industry-leading HBM3E and HBM4E, providing 128GB of memory and delivering 1.6Tbps in memory bandwidth.<\/p>\n
Ascend 950DT will feature a higher-performance HBM chip called HiZQ2.0, providing 144GB of memory and 4Tbps memory bandwidth.<\/p>\n
Huawei has been using cluster computing, which puts multiple computers or \u201csupernodes\u201d to work together in order to improve the performance of its chips.<\/p>\n
Its current main product is the Atlas 900 A3 SuperPoD, a system that uses 384 of Huawei\u2019s latest 910C chips and which, one industry expert says, rivals some of Nvidia\u2019s most advanced offerings. In June, Huawei showed off the system and the cloud service that runs on it, CloudMatrix 384, at a key industry show.<\/p>\n
Huawei plans to introduce the Atlas 950 SuperPod in the fourth quarter of 2026. The system will pack 8,192 Ascend 950DT chips and comprise 160 cabinets \u2013 128 compute cabinets and 32 communications cabinets \u2013and be deployable in a 1,000 sq m space.<\/p>\n
Mr Eric Xu, Huawei\u2019s rotating chairman, said the Atlas 950 SuperPod will have 6.7 times more computing power and 15 times more memory capacity than the NVL144 system that Nvidia intends to launch in 2026, and would continue to beat a successor system Nvidia is planning for 2027.<\/p>\n
Huawei is also planning a second supernode product, called the Atlas 960 SuperPoD, which it said would pack up to 15,488 Ascend 960 chips and comprise 220 cabinets across a 2,200 sq m space.<\/p>\n
Besides the Ascend series, Huawei has another central processing unit chip product line called Kunpeng, which is used for general servers. Huawei first announced the Kunpeng 920 chipset in 2019.<\/p>\n
The company will roll out new versions, the Kunpeng 950 and 960, in 2026 and 2028, Mr Xu said, adding that this would be accompanied by another new cluster computing product called the TaiShan 950 SuperPod, which he said would be focused on general-purpose computing. REUTERS<\/p>\n
ChinaHuaweiSemiconductorsAI\/artificial intelligence<\/p>\n","protected":false},"excerpt":{"rendered":"BEIJING \u2013 China\u2019s Huawei has ended years of secrecy to detail its product road map for chips and…\n","protected":false},"author":2,"featured_media":157597,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[21],"tags":[49,48,285,61],"class_list":{"0":"post-157596","1":"post","2":"type-post","3":"status-publish","4":"format-standard","5":"has-post-thumbnail","7":"category-computing","8":"tag-ca","9":"tag-canada","10":"tag-computing","11":"tag-technology"},"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/157596","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/comments?post=157596"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/157596\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media\/157597"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media?parent=157596"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/categories?post=157596"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/tags?post=157596"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}