Difficulties training the start-up’s latest model with Huawei’s Ascend chips highlight dependence on Nvidia
* encouraged to use Huawei chips by the government
* persistent technical issues during its R2 training process using Ascend chips
* The issues were the main reason the model’s launch was delayed from May,
* Huawei chips suffer from stability issues, slower inter-chip connectivity and inferior software compared with Nvidia’s products.
* Huawei sent a team of engineers to DeepSeek’s office, but despite that, DeepSeek could not conduct a successful training run on the Ascend chip
* Beijing has been demanding that Chinese tech companies justify their orders of Nvidia’s H20, to encourage Huawei sales
* DeepSeek was still working with Huawei to make the model compatible with Ascend for inference.
www.ft.com
* encouraged to use Huawei chips by the government
* persistent technical issues during its R2 training process using Ascend chips
* The issues were the main reason the model’s launch was delayed from May,
* Huawei chips suffer from stability issues, slower inter-chip connectivity and inferior software compared with Nvidia’s products.
* Huawei sent a team of engineers to DeepSeek’s office, but despite that, DeepSeek could not conduct a successful training run on the Ascend chip
* Beijing has been demanding that Chinese tech companies justify their orders of Nvidia’s H20, to encourage Huawei sales
* DeepSeek was still working with Huawei to make the model compatible with Ascend for inference.
DeepSeek’s next AI model delayed by attempt to use Chinese chips
Difficulties of training the start-up’s latest system with Huawei’s semiconductors highlight dependence on Nvidia
www.ft.com