AMD demonstrated its latest breakthroughs in the field of artificial intelligence to the world at the "Advancing AI 2024" conference held in San Francisco. This event brought together many technology giants to witness AMD’s ambitions in the field of AI computing. The editor of Downcodes will take you to have an in-depth understanding of AMD's major products released this time, as well as its strategic layout in the field of AI, and experience the wonderful moments of this technological feast.
Under the spotlight of the technology world, AMD once again became the center of hot discussion at the "Advancing AI2024" conference in San Francisco, demonstrating its latest achievements in the field of artificial intelligence. This grand event on October 10th is not only a technology carnival, but also the opening of a new chapter in the field of AI computing.
At the conference, AMD unveiled five core products, covering multiple fields from flagship AI chips to server CPUs, to AI network cards, DPUs and mobile processors, demonstrating its ambitions in the AI battlefield. The scene brought together industry giants such as Google, OpenAI, and Microsoft to witness this shining moment in the AI industry.
AMD's flagship AI chip released this time, Instinct MI325X GPU, uses advanced HBM3E high-bandwidth memory and provides powerful computing power of up to 21PFLOPS. Compared with NVIDIA's H200 GPU, MI325X is superior in memory capacity and bandwidth, especially in the peak computing power of FP16 and FP8, showing obvious advantages.
AMD's AI chip development blueprint is also eye-catching. The MI350 series to be launched next year will be based on the new CDNA4 architecture. It is expected that the peak AI computing power of the 8-card MI355X will reach an astonishing 74PFLOPS. This expectation has the industry full of expectations.
As the demand for computing power in data centers continues to grow, AMD has also launched the first AI network card Pensando Pollara400 that supports the UEC Super Ethernet Alliance, and the Pensando Salina400DPU with doubled performance to ensure that it can meet the needs of every data center.
Also attracting much attention is AMD's fifth-generation EPYC server CPU. This product, known as the "world's best CPU", uses an advanced 3/4nm process technology and supports up to 192 cores and 384 threads. The top-of-the-line EPYC9965 has a thermal design power consumption of 500W, and the starting price of a single unit is as high as US$14,813. Compared with Intel's processors, it has achieved significant improvements in multiple performance indicators.
Since returning to the data center market in 2017, AMD's market share has soared from 2% in 2018 to 34% in the first half of this year, and now covers more than 950 cloud instances and more than 350 OxM platforms globally.
AMD also launched a new generation of commercial AI mobile processors, the Ryzen AI PRO300 series. It is expected that more than 100 Ryzen AI PRO PCs will be launched by 2025, showing AMD's ambition in the AI PC field.
At this event, AMD not only demonstrated the powerful performance of its flagship AI chip, but also revealed its future development plans, demonstrating its determination to continue to innovate in the AI field and continue to promote technological progress. Whether through high-performance chips or comprehensive network solutions, AMD aims to meet all the needs of the modern data center and drive AI computing to new heights. For technology enthusiasts, AMD’s conference is undoubtedly a technology feast. Let us look forward to more surprises that AMD will bring in the AI era.
AMD's strong performance in the AI field indicates that competition in the AI computing field will become more intense in the future, and it will also bring us more technological innovations worth looking forward to. Let us wait and see what surprises AMD will bring us next!