preface
As we all know, the hottest technology product in the first half of this year must be ChatGPT. Not only in foreign fire, Microsoft, Google, Meta, Amazon, etc. all participated in the war, but also in the domestic fire mess, Baidu, iFLYTEK, Ali, Huawei, etc. are also all into the game, which can be described as a hundred regiments in the AI industry.
OpenAI, the parent company of ChatGPT, successfully trained the GPT-3 language model with 10,000 NVIDIA GPU chips, which shocked the world. The powerful NVIDIA AI chip is more sought after by the market, occupying a monopoly position in the market, this AI model of the fire, but also make NVIDIA earn a lot of money, become the world's first chip company with a market value of more than one trillion, NVIDIA said that this is the era of AI iPhone. ，
US News: Increase the ban on sales to China!
In October last year, NVIDIA was banned from selling A100/H100 chips for AI training to China, and then its castration performance launched the A800/H800 that meets export controls to China, and the castrated 800 series chip performance is only 50-70%, but the series of products are still used by domestic manufacturers such as Alibaba, Baidu, Tencent and ByteDance
According to the latest news of the US media on June 28, A800 and H800 are also going to be banned from sale, and the United States does not allow NVIDIA to sell these powerful GPU chips to China. It is expected to be closed in July at the earliest.
Nvidia Chief Financial Officer Colette Kress said that the US restrictions on the export of artificial intelligence chips to China "will permanently lose opportunities for the US industry" to warn the United States not to ban sales. But it's hard to predict whether there will be a ban next, and industry insiders say it's highly likely.
Can domestic GPUs pick the beam?
High-performance GPUs have become one of the most direct factors limiting the development of China's AI industry. Once NVIDIA A800 and H800 chips are banned, how can China's AI models still play? Can domestic GPUs pick the beam?
GPU chips can be divided into three types, one is called AI, which is used for artificial intelligence, and AI large models are also used; One is called FP, which is used for supercomputing acceleration; There is also a kind called rendering, which is ordinary game rendering and graphics applications.
At present, domestic GPUs have made breakthroughs in rendering this piece, AI large model training this block, mainly need to process high-granularity information, need double-precision floating point computing capabilities, at present, domestic GPU chips, mainly support single-precision floating-point computing.
As NVIDIA said, if the United States really wants to ban sales, then it will force the development of Chinese GPU chips, and the future will inevitably make gradual progress, and then let the US GPU chips lose the opportunity forever.
It is in the context of the surge in computing power demand, the cutting off of hardware supply, and the difficulty of real "marketization" that a number of GPU "tinder" companies have emerged in China. Through independent research and development, Haiguang Information has successfully mastered core technologies such as high-end coprocessor microstructure design, and launched DCU products with excellent performance on this basis, with powerful computing power and high-speed parallel data processing capabilities.
Compared with the company's Shensuan No. 1 product and the high-end GPU products (A100) and AMD high-end GPU products (MI100) of the leading GPU manufacturer NVIDIA, in typical application scenarios, the indicators of Haiguang Information Shensuan No. 1 single chip basically reach the level of the same type of high-end products in the world.
Compared with the NVIDIA A100 products currently used by international mainstream artificial intelligence companies, Haiguang DCU single-chip products can basically reach 70% of its performance level, and at the same time, the interconnection performance of the company's DCU products still has a lot of room for improvement.
Write at the end
In recent years, with the efforts of many domestic enterprises, the breakthrough of GPU hardware has been frequent, and the short-term blockade may be a dilemma, but from another point of view, it may not be an opportunity, and China's high-end enterprises represented by Haiguang Information will also be killed in layers of blockade!
At present, mainland GPU manufacturers are still far away from international leading manufacturers such as NVIDIA, but it is certain that with the advancement of science and technology, China's GPU computing power industry must and will get rid of the problem of being "stuck in the neck", so that China's AI companies can use China's GPUs to win this computing power war.
Code words are not easy, welcome to leave a message to discuss and communicate, like it to help me like!