0726-1210

diffusion.py

@@ -71,7 +71,7 @@ from torch.distributed import init_process_group, destroy_process_group
 import torch.distributed as dist
 
 import argparse
-
+import socket
 
 # %%
 def ddp_setup(rank: int, world_size: int, master_addr, master_port):
@@ -738,7 +738,7 @@ if __name__ == "__main__":
     ############################ training ################################
     world_size = torch.cuda.device_count()
     if args.train:
-        print(f" training, master_addr = {master_addr}, local_world_size = {local_world_size}, world_size = {world_size} ".center(120,'-'))
+        print(f" training, ip_addr = {socket.gethostbyname(socket.gethostname())}, local_world_size = {local_world_size}, world_size = {world_size} ".center(120,'-'))
         mp.spawn(
                 train, 
                 args=(world_size, local_world_size, master_addr, master_port), 

phoenix_diffusion.sbatch

@@ -2,7 +2,7 @@
 #SBATCH -J diffusion # Job name
 #SBATCH -A gts-jw254-coda20 
 #SBATCH -qembers
-#SBATCH -N2 --gpus-per-node=RTX_6000:4 # -C A100-80GB              # Number of nodes and cores per node required
+#SBATCH -N4 --gpus-per-node=RTX_6000:2 # -C A100-80GB              # Number of nodes and cores per node required
 #SBATCH --mem-per-gpu=32G                        # Memory per core
 #SBATCH -t 10:00                                    # Duration of the job (Ex: 15 mins)
 #SBATCH -oReport-%j                         # Combined output and error messages file