News
Alibaba Cloud has revealed a modular datacenter architecture it claims will motivate it to satisfy question for AI infrastructure by making improvements to efficiency and build times for ticket new facilities.
Equipped at its annual Apsara conference the day gone by, the “CUBE DC 5.0” architecture used to be described as using “prefabricated modular designs” plus “advanced and proprietary technologies such as wind-liquid hybrid cooling system, all-direct current power distribution architecture and smart management system.”
Alibaba Cloud hasn’t explained these techs extensive, but claims the modular arrangement reduces deployment times by up to 50 percent when put next to feeble datacenter building ways.
The Register has requested Alibaba to display the workings of a “wind-liquid hybrid cooling system.” For what or no longer it’s worth, machine translation of the term into Mandarin, then fed into engines like google, produced outcomes describing wintry plate cooling – a approach that sees thin reservoirs of cooled liquids positioned on hardware, with cooling accomplished by circulating liquid and/or blowing air throughout the plates.
Whatever the term describes, CEO of Alibaba Cloud Intelligence Eddie Wu told the conference his firm “is investing heavily in building an AI infrastructure for the future.”
“These enhancements are not just about keeping up with AI demands but about setting a global standard for efficiency and sustainability.”
Other steps in the direction of that honest contain a scheduler stated to greater organize hardware sources so that they acquire up to 90 percent utilization rates.
Alibaba Cloud’s IaaS offering, the Endeavor Elastic Compute Service (ECS), has reached its ninth era. Conference attendees were told it’s greater geared up for AI applications as it has improved recommendation engine speeds by 30 percent and database learn/write queries per 2nd by 17 percent.
Also at the conference, Alibaba Cloud announced an “Open Lake data utility” that integrates a pair of enormous records engines so they is also feeble by generative AI applications. One other new offering, “DMS: OneMeta+OneOps,” it sounds as if combines and manages metadata from 40 diversified records sources.
It’s 2024, so Alibaba Cloud moreover announced some AI news: the discharge of its Qwen 2.5 multimodal items, readily accessible in sizes from 0.5 to 72 billion parameters, supporting 29 languages and tuned for the wants of sectors including car and gaming. The new items are stated to acquire “enhanced knowledge [and] stronger capabilities in math and coding.”
A text-to-video AI mannequin that works with both Chinese language and English prompts, Tongyi Wanxiang, used to be moreover released.
“The new model is capable of generating high-quality videos in a wide variety of visual styles from realistic scenes to 3D animation,” boasted Alibaba Cloud execs. ®
