I don't know enough about this stuff but am I right in thinking that if you had the expertise to pull this off then doing the same for an eventual home (China) developed GPU would not be too much of an ask?
Interesting, why do you say that China can't design GPUs as powerful and NvIdia GPU's? Is that an expertise issue or a different issue? Design or manufacturing or both? Would you mind elaborating a little please?
manufacturing high-end GPU requires EUV from ASML. And the access to the EUV is restricted by US chip export control. I believe this is the main reason.
There are also many other factors like experience, talents and more.
I don't know enough about this stuff but am I right in thinking that if you had the expertise to pull this off then doing the same for an eventual home (China) developed GPU would not be too much of an ask?
This article mainly talks about software level stuff (CUDA and PTX), and CUDA is usually considered as the moat of Nvidia ecosystem.
But China cannot design and develop GPUs that are as powerful as Nvidia GPU like A100, H100...
But what DeppSeek achieved may indicate that they can do the same for AMD GPUs.
Interesting, why do you say that China can't design GPUs as powerful and NvIdia GPU's? Is that an expertise issue or a different issue? Design or manufacturing or both? Would you mind elaborating a little please?
Maybe both.
manufacturing high-end GPU requires EUV from ASML. And the access to the EUV is restricted by US chip export control. I believe this is the main reason.
There are also many other factors like experience, talents and more.
Appreciate you answering my questions, thank you very much.