The rise of Chinese enterprises is coming, or replace Nvidia as a substitute for Huawei chips
At present, many Chinese manufacturers are very concerned about finding new breakthroughs to replace NVIDIA chips. From this point of view, Huawei is also one of the alternatives for many enterprises.
That's true. According to the United States, although Nvidia has encountered a large number of shipping problems, in 2022, Huawei still received more than 5,000 orders from China for 910 B. Huawei's share will continue to rise in 2024, as the two largest U.S. technology companies have cut the number of chips they supply to Nvidia.
In addition, some time ago, there were rumors that Huawei will reduce the production scale of its Kirin processors in order to cope with Nvidia's 910 bai capabilities, so as to increase its 910 bai production capacity. If this is true, then Nvidia's stock price will be greatly impacted.
It's a pity that he's a little bit close. The difference between a CPU and a GPU is not a CPU and a GPU as compared to Nvidia's CPU, but a CPU and a GPU. On the other hand, Huawei still has room for improvement in this area, so we have to give it credit to Huawei, because we can't completely replace NVIDIA cores.
The development opportunity of Chinese companies has finally arrived, and this time, Huawei may be squeezed aside, that is, Haiguang Technology.
Hikari has two major families, one is a CPU and the other is a standard data unit (DCU) based on NVIDIA GPUs. According to the design of the PCOCH DCU, the storage rate and storage capacity of the storage unit are basically the same as those of the A100. However, the interaction rate is far from the same, at only 30% of the A100. According to the calculations, the system is only 50% of the A100.
Now, do you feel a little hope? Don't worry, the second-generation aircraft has greatly improved its performance, and it is expected to be comparable to 80% of the A100. Such a good performance is not a simple matter, if the new generation in the future, there will be more improvements, and it may even be comparable to the A100. But that's not the point, the point is CUDA.
As mentioned earlier, NVIDIA's real strength lies in CUDA, which has been building a software ecosystem of developers and industry leaders since 2006. However, Haiguang is a special case, they have the same effect as NVIDIA in the ecological environment of CUDA, and can be well compatible with mainstream models at home and abroad.
To put it simply, with such a large ecosystem, other enterprises can achieve "ecology" as long as they use Haiguang's software, and the cost will be reduced. This is a great opportunity to use in place of Nvidia. Although the domestic 910 b processor is very powerful, there are still many manufacturers who have chosen Nvidia's copycat products. From this point of view, this evolution is a huge breakthrough, and it can also play a certain auxiliary role.
It is expected that the new generation of Guangchi can be successfully developed as soon as possible to match the high-performance products of NVIDIA, which may be the rise of domestic graphics processors. What are your thoughts on this matter? Welcome to leave a message, like, share!