fix check of multinode

Summary: rank is 0-based

Reviewed By: bowenc0221

Differential Revision: D30258674

fbshipit-source-id: 1e8331e303fdcce8c96c500e520ffadc000f06ba

Author: ppwwyyxx

projects/DeepLab/train_net.py

@@ -67,7 +67,7 @@ def build_evaluator(cls, cfg, dataset_name, output_folder=None):
             )
         if evaluator_type == "cityscapes_sem_seg":
             assert (
-                torch.cuda.device_count() >= comm.get_rank()
+                torch.cuda.device_count() > comm.get_rank()
             ), "CityscapesEvaluator currently do not work with multiple machines."
             return CityscapesSemSegEvaluator(dataset_name)
         if len(evaluator_list) == 0:

projects/Panoptic-DeepLab/train_net.py

@@ -69,7 +69,7 @@ def build_evaluator(cls, cfg, dataset_name, output_folder=None):
             evaluator_list.append(COCOPanopticEvaluator(dataset_name, output_folder))
         if evaluator_type == "cityscapes_panoptic_seg":
             assert (
-                torch.cuda.device_count() >= comm.get_rank()
+                torch.cuda.device_count() > comm.get_rank()
             ), "CityscapesEvaluator currently do not work with multiple machines."
             evaluator_list.append(CityscapesSemSegEvaluator(dataset_name))
             evaluator_list.append(CityscapesInstanceEvaluator(dataset_name))

projects/PointRend/train_net.py

@@ -81,12 +81,12 @@ def build_evaluator(cls, cfg, dataset_name, output_folder=None):
             )
         if evaluator_type == "cityscapes_instance":
             assert (
-                torch.cuda.device_count() >= comm.get_rank()
+                torch.cuda.device_count() > comm.get_rank()
             ), "CityscapesEvaluator currently do not work with multiple machines."
             return CityscapesInstanceEvaluator(dataset_name)
         if evaluator_type == "cityscapes_sem_seg":
             assert (
-                torch.cuda.device_count() >= comm.get_rank()
+                torch.cuda.device_count() > comm.get_rank()
             ), "CityscapesEvaluator currently do not work with multiple machines."
             return CityscapesSemSegEvaluator(dataset_name)
         if len(evaluator_list) == 0:

tools/plain_train_net.py

@@ -78,12 +78,12 @@ def get_evaluator(cfg, dataset_name, output_folder=None):
         evaluator_list.append(COCOPanopticEvaluator(dataset_name, output_folder))
     if evaluator_type == "cityscapes_instance":
         assert (
-            torch.cuda.device_count() >= comm.get_rank()
+            torch.cuda.device_count() > comm.get_rank()
         ), "CityscapesEvaluator currently do not work with multiple machines."
         return CityscapesInstanceEvaluator(dataset_name)
     if evaluator_type == "cityscapes_sem_seg":
         assert (
-            torch.cuda.device_count() >= comm.get_rank()
+            torch.cuda.device_count() > comm.get_rank()
         ), "CityscapesEvaluator currently do not work with multiple machines."
         return CityscapesSemSegEvaluator(dataset_name)
     if evaluator_type == "pascal_voc":

tools/train_net.py

@@ -65,12 +65,12 @@ def build_evaluator(cfg, dataset_name, output_folder=None):
         evaluator_list.append(COCOPanopticEvaluator(dataset_name, output_folder))
     if evaluator_type == "cityscapes_instance":
         assert (
-            torch.cuda.device_count() >= comm.get_rank()
+            torch.cuda.device_count() > comm.get_rank()
         ), "CityscapesEvaluator currently do not work with multiple machines."
         return CityscapesInstanceEvaluator(dataset_name)
     if evaluator_type == "cityscapes_sem_seg":
         assert (
-            torch.cuda.device_count() >= comm.get_rank()
+            torch.cuda.device_count() > comm.get_rank()
         ), "CityscapesEvaluator currently do not work with multiple machines."
         return CityscapesSemSegEvaluator(dataset_name)
     elif evaluator_type == "pascal_voc":