diff --git a/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_cosmoflow.json b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_cosmoflow.json
new file mode 100644
index 00000000..54949099
--- /dev/null
+++ b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_cosmoflow.json
@@ -0,0 +1,82 @@
+{
+    "cosmoflow_ref_32": {
+        "Benchmark": "cosmoflow",
+        "BS": 32,
+        "Epochs to converge": [ 8, 8, 9, 8, 6, 8, 6, 6, 7, 8, 8, 8, 9, 8, 7, 8, 6, 9, 8, 8, 8, 8, 9, 8, 9, 8, 8, 8, 8, 8, 8, 8, 8  ],
+        "Hyperparams": {
+            "global_batch_size": 32,
+            "opt_name": "SGD",
+            "opt_base_learning_rate": 0.001,
+            "opt_learning_rate_warmup_epochs": 2,
+            "opt_learning_rate_warmup_factor": 4.0,
+            "opt_learning_rate_decay_boundary_epochs": [ 4, 6 ],
+            "opt_learning_rate_decay_factor": [0.25, 0.0625],
+            "dropout": 0.5,
+            "opt_weight_decay": 0.0
+        }
+    },
+    "cosmoflow_ref_64": {
+        "Benchmark": "cosmoflow",
+        "BS": 64,
+        "Epochs to converge": [ 19, 18, 20, 17, 18, 19, 18, 18, 18, 18, 18, 18, 18, 19, 18, 18, 18, 18, 18, 20 ],
+        "Hyperparams": {
+            "global_batch_size": 64,
+            "opt_name": "SGD",
+            "opt_base_learning_rate": 0.001,
+            "opt_learning_rate_warmup_epochs": 4,
+            "opt_learning_rate_warmup_factor": 1.0,
+            "opt_learning_rate_decay_boundary_epochs": [ 16, 32 ],
+            "opt_learning_rate_decay_factor": 0.25,
+            "dropout": 0.0,
+            "opt_weight_decay": 0.01
+        }
+    },
+    "cosmoflow_ref_128": {
+        "Benchmark": "cosmoflow",
+        "BS": 128,
+        "Epochs to converge": [ 16, 18, 18, 18, 18, 17, 18, 18, 18, 18, 18, 18, 17, 18, 18, 18, 19, 18, 18, 18 ],
+        "Hyperparams": {
+            "global_batch_size": 128,
+            "opt_name": "SGD",
+            "opt_base_learning_rate": 0.004,
+            "opt_learning_rate_warmup_epochs": 4,
+            "opt_learning_rate_warmup_factor": 4.0,
+            "opt_learning_rate_decay_boundary_epochs": [ 16, 32 ],
+            "opt_learning_rate_decay_factor": 0.25,
+            "dropout": 0.5,
+            "opt_weight_decay": 0.0
+        }
+    },
+    "cosmoflow_ref_512": {
+        "Benchmark": "cosmoflow",
+        "BS": 512,
+        "Epochs to converge": [ 22, 23, 23, 24, 24, 21, 23, 23, 23, 23, 24, 23, 23, 21, 23, 23, 21, 23, 23, 23 ],
+        "Hyperparams": {
+            "global_batch_size": 512,
+            "opt_name": "SGD",
+            "opt_base_learning_rate": 0.006,
+            "opt_learning_rate_warmup_epochs": 6,
+            "opt_learning_rate_warmup_factor": 4.0,
+            "opt_learning_rate_decay_boundary_epochs": [ 19, 21, 22, 23 ],
+            "opt_learning_rate_decay_factor": [0.5, 0.25, 0.125, 0.0625],
+            "dropout": 0.0,
+            "opt_weight_decay": 0.01
+        }
+    },
+    "cosmoflow_ref_1024": {
+        "Benchmark": "cosmoflow",
+        "BS": 1024,
+        "Epochs to converge": [ 42, 38, 42, 40, 40, 39, 43, 37, 39, 37, 43, 39, 38, 42, 40, 42, 42, 38, 36, 43 ],
+        "Hyperparams": {
+            "global_batch_size": 1024,
+            "opt_name": "SGD",
+            "opt_base_learning_rate": 0.012,
+            "opt_learning_rate_warmup_epochs": 0,
+            "opt_learning_rate_warmup_factor": 1.0,
+            "opt_learning_rate_decay_boundary_epochs": [ 32, 64 ],
+            "opt_learning_rate_decay_factor": 0.25,
+            "dropout": 0.5,
+            "opt_weight_decay": 0.0
+        }
+    }
+}
diff --git a/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_deepcam.json b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_deepcam.json
new file mode 100644
index 00000000..bcc32625
--- /dev/null
+++ b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_deepcam.json
@@ -0,0 +1,119 @@
+{
+    "deepcam_ref_128": {
+        "Benchmark": "deepcam",
+        "BS": 128,
+        "Epochs to converge": [ 7, 7, 7, 7, 7, 7, 7, 8, 8, 6, 7, 8, 7, 6, 6, 7, 7, 6, 7, 7 ],
+        "Hyperparams": {
+            "batchnorm_groupsize": 1,
+            "global_batch_size": 128,
+            "gradient_accumulation_frequency": 1,
+            "num_workers": 32,
+            "opt_betas": [ 0.9, 0.999 ],
+            "opt_bias_correction": true,
+            "opt_eps": 1e-06,
+            "opt_grad_averaging": true,
+            "opt_lr": 0.00155,
+            "opt_max_grad_norm": 1.0,
+            "opt_weight_decay": 0.01,
+            "opt_name": "LAMB",
+            "scheduler_decay_rate": "0.1",
+            "scheduler_lr_warmup_factor": 1.0,
+            "scheduler_lr_warmup_steps": 0,
+            "scheduler_t_max": "9000",
+            "scheduler_eta_min": "0.0",
+            "scheduler_type": "cosine_annealing"
+        }
+    },
+    "deepcam_ref_256": {
+        "Benchmark": "deepcam",
+        "BS": 256,
+        "Epochs to converge": [ 9, 9, 9, 9, 9, 9, 9, 9, 9, 9 ],
+        "Hyperparams": {
+            "batchnorm_groupsize": 1,
+            "global_batch_size": 256,
+            "gradient_accumulation_frequency": 1,
+            "opt_betas": [ 0.9, 0.999 ],
+            "opt_bias_correction": true,
+            "opt_eps": 1e-06,
+            "opt_grad_averaging": true,
+            "opt_lr": 0.002,
+            "opt_max_grad_norm": 1.0,
+            "opt_weight_decay": 0.01,
+            "opt_name": "LAMB",
+            "scheduler_decay_rate": 0.1,
+            "scheduler_lr_warmup_factor": 1.0,
+            "scheduler_lr_warmup_steps": 0,
+            "scheduler_milestones": [ 4096, 8192 ],
+            "scheduler_type": "multistep"
+        }
+    },
+    "deepcam_ref_512": {
+        "Benchmark": "deepcam",
+        "BS": 512,
+        "Epochs to converge": [ 11, 11, 11, 11, 11, 10, 16, 12, 11, 11, 11, 10, 11, 11, 11, 10, 11, 11, 11, 11, 11, 11, 11, 12, 11, 11, 11, 11, 13, 10, 11, 13, 11, 11, 11, 10, 14, 10, 10, 10 ],
+        "Hyperparams": {
+            "batchnorm_groupsize": 1,
+            "global_batch_size": 512,
+            "gradient_accumulation_frequency": 1,
+            "opt_betas": [ 0.9, 0.999 ],
+            "opt_bias_correction": true,
+            "opt_eps": 1e-06,
+            "opt_grad_averaging": true,
+            "opt_lr": 0.004,
+            "opt_max_grad_norm": 1.0,
+            "opt_weight_decay": 0.01,
+            "opt_name": "LAMB",
+            "scheduler_decay_rate": 0.1,
+            "scheduler_lr_warmup_factor": 1.0,
+            "scheduler_lr_warmup_steps": 100,
+            "scheduler_milestones": [ 2048, 4096 ],
+            "scheduler_type": "multistep"
+        }
+    },
+    "deepcam_ref_1024": {
+        "Benchmark": "deepcam",
+        "BS": 1024,
+        "Epochs to converge": [ 13, 13, 12, 13, 13, 13, 14, 13, 13, 13 ],
+        "Hyperparams": {
+            "batchnorm_groupsize": 1,
+            "global_batch_size": 1024,
+            "gradient_accumulation_frequency": 1,
+            "opt_betas": [ 0.9, 0.999 ],
+            "opt_bias_correction": true,
+            "opt_eps": 1e-06,
+            "opt_grad_averaging": true,
+            "opt_lr": 0.004,
+            "opt_max_grad_norm": 1.0,
+            "opt_weight_decay": 0.01,
+            "opt_name": "LAMB",
+            "scheduler_decay_rate": 0.1,
+            "scheduler_lr_warmup_factor": 1.0,
+            "scheduler_lr_warmup_steps": 200,
+            "scheduler_milestones": [ 1100, 4096 ],
+            "scheduler_type": "multistep"
+        }
+    },
+    "deepcam_ref_2048": {
+        "Benchmark": "deepcam",
+        "BS": 2048,
+        "Epochs to converge": [ 23, 23, 22, 23, 22, 26, 22, 22, 24, 22 ],
+        "Hyperparams": {
+            "batchnorm_groupsize": 1,
+            "global_batch_size": 2048,
+            "gradient_accumulation_frequency": 1,
+            "opt_betas": [ 0.9, 0.999 ],
+            "opt_bias_correction": true,
+            "opt_eps": 1e-06,
+            "opt_grad_averaging": true,
+            "opt_lr": 0.0055,
+            "opt_max_grad_norm": 1.0,
+            "opt_weight_decay": 0.01,
+            "opt_name": "LAMB",
+            "scheduler_decay_rate": "0.1",
+            "scheduler_lr_warmup_factor": 1.0,
+            "scheduler_lr_warmup_steps": 400,
+            "scheduler_milestones": [ 800 ],
+            "scheduler_type": "multistep"
+        }
+    }
+}
diff --git a/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_oc20.json b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_oc20.json
new file mode 100644
index 00000000..d42987a7
--- /dev/null
+++ b/mlperf_logging/rcp_checker/hpc_3.0.0/rcps_oc20.json
@@ -0,0 +1,41 @@
+{
+    "oc20_ref_256": {
+        "Benchmark": "oc20",
+        "BS": 256,
+        "Epochs to converge": [ 20, 18, 20, 20, 22, 18, 19, 19, 21, 21 ],
+        "Hyperparams": {
+            "global_batch_size": 256,
+            "opt_base_learning_rate": 0.0004,
+            "opt_learning_rate_warmup_steps": 31252,
+            "opt_learning_rate_warmup_factor": 0.2,
+            "opt_learning_rate_decay_boundary_steps": [ 125008, 187512, 250016 ],
+            "opt_learning_rate_decay_factor": 0.1
+        }
+    },
+    "oc20_ref_1024": {
+        "Benchmark": "oc20",
+        "BS": 1024,
+        "Epochs to converge": [ 23, 25, 22, 25, 25, 25, 25, 24, 23, 25 ],
+        "Hyperparams": {
+            "global_batch_size": 1024,
+            "opt_base_learning_rate": 0.0012,
+            "opt_learning_rate_warmup_steps": 7816,
+            "opt_learning_rate_warmup_factor": 0.2,
+            "opt_learning_rate_decay_boundary_steps": [ 31264, 46896 ],
+            "opt_learning_rate_decay_factor": 0.1
+        }
+    },
+    "oc20_ref_2048": {
+        "Benchmark": "oc20",
+        "BS": 2048,
+        "Epochs to converge": [ 33, 32, 33, 33, 33, 34, 33, 33, 30, 33 ],
+        "Hyperparams": {
+            "global_batch_size": 2048,
+            "opt_base_learning_rate": 0.0016,
+            "opt_learning_rate_warmup_steps": 3908,
+            "opt_learning_rate_warmup_factor": 0.2,
+            "opt_learning_rate_decay_boundary_steps": [ 23448, 31264 ],
+            "opt_learning_rate_decay_factor": 0.1
+        }
+    }
+}
diff --git a/mlperf_logging/result_summarizer/result_summarizer.py b/mlperf_logging/result_summarizer/result_summarizer.py
index 1570e216..17e0e6e3 100644
--- a/mlperf_logging/result_summarizer/result_summarizer.py
+++ b/mlperf_logging/result_summarizer/result_summarizer.py
@@ -322,7 +322,7 @@ def _compute_strong_scaling_scores(desc, system_folder, usage, ruleset):
     benchmark_folder_parent = os.path.join(
         system_folder, 'strong') if usage == 'hpc' else system_folder
     if not os.path.isdir(benchmark_folder_parent):
-        return benchmark_scores
+        return benchmark_scores, benchmark_power_scores
     for benchmark_folder in _get_sub_folders(benchmark_folder_parent):
         folder_parts = benchmark_folder.split('/')
         # Check if this benchmark has power results
@@ -404,7 +404,7 @@ def _compute_weak_scaling_scores(desc, system_folder, usage, ruleset):
     benchmark_power_scores = {}
     has_power = None
     if not os.path.isdir(benchmark_folder_parent):
-        return benchmark_scores
+        return benchmark_scores, benchmark_power_scores
     for benchmark_folder in _get_sub_folders(benchmark_folder_parent):
         folder_parts = benchmark_folder.split('/')
         benchmark = _benchmark_alias(folder_parts[-1])
@@ -457,13 +457,24 @@ def _compute_weak_scaling_scores(desc, system_folder, usage, ruleset):
             olympic_avg = _compute_olympic_average(
                 power_scores, 1, 1)
             if olympic_avg is not None:
-                benchmark_power_scores[benchmark] = olympic_avg
+                benchmark_power_scores['{}:{}'.format(
+                    benchmark,
+                    'time_to_train_all',
+                )] = olympic_avg
+                benchmark_power_scores['{}:{}'.format(
+                    benchmark,
+                    'number_of_models',
+                )] = olympic_avg
+                benchmark_power_scores['{}:{}'.format(
+                    benchmark,
+                    'instance_scale',
+                )] = olympic_avg
 
     _fill_empty_benchmark_scores(benchmark_scores,
                                  usage,
                                  ruleset,
                                  weak_scaling=True)
-    _fill_empty_benchmark_scores(benchmark_power_scores, usage, ruleset)
+    _fill_empty_benchmark_scores(benchmark_power_scores, usage, ruleset, weak_scaling=True)
     return benchmark_scores, benchmark_power_scores
 
 
@@ -565,7 +576,7 @@ def summarize_results(folder, usage, ruleset, csv_file=None):
                                               ruleset,
                                               weak_scaling=True)
     power_summary = _get_empty_summary(usage, ruleset)
-    power_weak_scaling_summary = _get_empty_summary(usage, ruleset)
+    power_weak_scaling_summary = _get_empty_summary(usage, ruleset, weak_scaling=True)
 
     for system_folder in _get_sub_folders(results_folder):
         folder_parts = system_folder.split('/')
diff --git a/scripts/verify_for_v3.0_hpc.sh b/scripts/verify_for_v3.0_hpc.sh
new file mode 100755
index 00000000..9206ef56
--- /dev/null
+++ b/scripts/verify_for_v3.0_hpc.sh
@@ -0,0 +1,18 @@
+#!/bin/bash
+
+set -e
+
+# rcp_bypass and rcp_bert_train_samples packahe checker params
+# need to be retrieved at package_checker_params file at top-level submission dir.
+PACKAGE_CHECKER_PARAMS=""
+PACKAGE_CHECKER_PARAMS_FILE="$1/package_checker_params"
+if test -f "$PACKAGE_CHECKER_PARAMS_FILE"; then
+  while IFS= read -r line
+  do
+    PACKAGE_CHECKER_PARAMS="$PACKAGE_CHECKER_PARAMS --$line"
+  done < "$PACKAGE_CHECKER_PARAMS_FILE"
+fi
+
+python3 -m mlperf_logging.package_checker $1 hpc 3.0.0 $PACKAGE_CHECKER_PARAMS
+python3 -m mlperf_logging.result_summarizer $1 hpc 3.0.0
+python3 -m mlperf_logging.repo_checker $1 hpc 3.0.0