I have some questions regarding training runtime and the number of GPUs.
In your paper on table 4 (page 22), it shows the actual training runtime of each experiment.
Are these reported runtimes are measure by running w/ multiple GPUs?
Especially for Navier-Stokes equation, you also experimented w/ 8 GPUs, and I want to know whether the 68.29 it/sec of runtime was achieved by multiple GPUs.
Hello, thank you for sharing your wonderful work!
I have some questions regarding training runtime and the number of GPUs.
In your paper on table 4 (page 22), it shows the actual training runtime of each experiment.
Are these reported runtimes are measure by running w/ multiple GPUs?
Especially for Navier-Stokes equation, you also experimented w/ 8 GPUs, and I want to know whether the 68.29 it/sec of runtime was achieved by multiple GPUs.
Thank you.