Boosting Wan2.2 I2V Inference on 8xH100s, 56% Faster with Sequence Parallelism morphic.com 6 points by palakzat 17 hours ago
How does it scale for >8 H100s ?