> Their current training cluster would be the 5th largest supercomputer if Tesla stopped all real workloads, ran Linpack, and submitted it to the Top500 list.

> Their current training cluster would be the 5th largest supercomputer if Tesla stopped all real workloads, ran Linpack, and submitted it to the Top500 list. That line is referring to their current Nvidia A100 powered supercomputer which they set up with 5,760 A100 GPUs recently this year [0]. Read the previous line of the line which you’ve shared from the post: > Tesla has been expanding the size of their GPU clusters for years. Their current training cluster would be the 5th largest supercomputer if Tesla stopped all real workloads, ran Linpack, and submitted it to the Top500 list. [0] https://blogs.nvidia.com/blog/2021/06/22/tesla-av-training-supercomputer-nvidia-a100-gpus/ https://blogs.nvidia.com/blog/2021/06/22/tesla-av-training-s...