China keeps buying hobbled Nvidia cards to train its AI models
arstechnica.com
Many Chinese tech companies are still at the stage of pre-training large language models, which burns a lot of performance from individual GPU chips and demands a high degree of data transfer capability. The H800 chips are five times faster than the A100 chips that had been Nvidia’s most powerful GP
1 Users
0 Comments
2 Highlights
2 Notes
Summary
Chinese tech companies are purchasing hobbled Nvidia graphics cards to train their artificial intelligence (AI) models. These companies are still in the pre-training stage of developing large language models, which requires significant performance from individual GPU chips and high data transfer capability. The H800 chips, which are being used by Chinese firms, are reportedly five times faster than Nvidia's most powerful GPUs, the A100 chips. This move by Chinese companies highlights the demand for powerful hardware to support AI development and the potential for China to become a leader in AI technology.
Top Highlights
- Many Chinese tech companies are still at the stage of pre-training large language models, which burns a lot of performance from individual GPU chips and demands a high degree of data transfer capability.
- The H800 chips are five times faster than the A100 chips that had been Nvidia’s most powerful GPUs, according to Patrick Moorhead, a US chip analyst at Moor Insights & Strategy.
Domain
Ready to highlight and find good content?
Glasp is a social web highlighter that people can highlight and organize quotes and thoughts from the web, and access other like-minded people’s learning.