In addition to Weibo, there is also WeChat
Please pay attention
WeChat public account
Shulou
2025-02-04 Update From: SLTechnology News&Howtos shulou NAV: SLTechnology News&Howtos > IT Information >
Share
Shulou(Shulou.com)12/24 Report--
Meta, OpenAI and Microsoft said at an AMD investor event on Wednesday that they will all use AMD's newly developed artificial intelligence chip Instinct MI300X. This suggests that while Nvidia's GPU chip is crucial for developing and deploying artificial intelligence programs such as ChatGPT, many technology companies are looking for alternatives.
If AMD's latest high-end chip, Instinct MI300X, can start shipping early next year and meet the needs of artificial intelligence companies and related cloud service providers, it could reduce the cost of developing artificial intelligence models and put competitive pressure on Nvidia's soaring artificial intelligence chip business.
Su Zifeng, chief executive of AMD, said on Wednesday: "all the attention is focused on the large processors and large GPU used in cloud computing."
AMD says MI300X is based on a new architecture that significantly improves performance. The most important feature of this chip is that it has the high-performance HBM3 memory of 192GB, which transmits data faster and can adapt to larger artificial intelligence models.
Su Zifeng directly compared MI300X and its system with Nvidia's mainstream artificial intelligence GPU chip H100.
"this performance can be translated directly into a better user experience," Su said. "when you ask a question to the model, you always want it to respond more quickly, especially if the response is becoming more and more complex."
The main question for AMD is whether companies that have been based on Nvidia will invest more time and money in another GPU supplier. "it takes effort to accept AMD," says Su Zifeng.
AMD told investors and partners on Wednesday that the company had improved its ROCm software suite. In order to compete with Nvidia CUDA software, which is the benchmark of the industry, a key defect has been solved. This original flaw in the ROCm software suite has been one of the main reasons why artificial intelligence developers now prefer Nvidia.
Chip price is also important. AMD did not disclose the pricing of MI300X on Wednesday. Nvidia GPU chips currently sell for about $40,000 each, and Su Zifeng said that AMD chips must be cheaper and cheaper to operate than Nvidia to persuade customers to buy them.
AMD also said on Wednesday that it had signed agreements with some of the companies that needed GPU chips most. According to a recent report by market research firm Omidia, Meta and Microsoft are the two biggest buyers of Nvidia H100 GPU chips in 2023.
Meta said it will use MI300X chips to handle artificial intelligence reasoning workloads, such as artificial intelligence stickers, image editing and the operation of voice assistants.
Microsoft Chief Technology Officer Kevin Scott (Kevin Scott) said the company will provide access to MI300X chips through Azure network services.
Oracle's cloud computing will also use this chip.
OpenAI said it would support AMD's GPU chip in a software product called Triton. Triton is not a large language model like GPT, but it also uses chip functions to carry out artificial intelligence research.
AMD has not yet predicted large-scale sales of the chip, and only expects total revenue of the data center GPU to be about $2 billion in 2024. By contrast, Nvidia's data center revenue exceeded $14 billion in the most recent quarter alone, but that figure also includes other chip businesses other than GPU.
However, AMD said the market for artificial intelligence GPU chips could climb to $400 billion over the next four years, twice as much as the company had expected. This shows that the industry has high expectations for high-end artificial intelligence chips, which is why AMD is now focusing investors' attention on the product line.
Su Zifeng also said that AMD does not think it is necessary to beat Nvidia to achieve better performance in the market.
Speaking of the artificial intelligence chip market, Su Zifeng said: "I think it is clear that Nvidia now has the vast majority."We think the figure could exceed $400 billion by 2027. We can totally get a piece of the pie."
Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.
Views: 0
*The comments in the above article only represent the author's personal views and do not represent the views and positions of this website. If you have more insights, please feel free to contribute and share.
Continue with the installation of the previous hadoop.First, install zookooper1. Decompress zookoope
"Every 5-10 years, there's a rare product, a really special, very unusual product that's the most un
© 2024 shulou.com SLNews company. All rights reserved.