Running training jobs across multiple nodes scales really well. A common assumption is that scale inevitably means slowdowns: more GPUs means more synchronization overhead, especially with multiple nodes communicating across a network. But we observed that the performance penalty isn’t as harsh as what you might think. Instead, we found near-linear strong scaling: fixing the global batch size and training on more GPUs led to proportional increases in training throughput. On a 1.3B parameter model, 4 nodes means a 3.9x gain over one node. On 16 nodes, it’s 14.4x. This is largely thanks to the super fast interconnects that major cloud providers have built in: @awscloud EC2 P4d instances provide 400 Gbps networking bandwidth, @Azure provides 1600 Gbps, and @OraclePaaS provides 800 Gbps.
Recent articles
- Using pip to install a Large Language Model that's under 100MB - 7th February 2025
- OpenAI o3-mini, now available in LLM - 31st January 2025
- A selfish personal argument for releasing code as Open Source - 24th January 2025