DEHNN/test.py at main · TILOS-AI-Institute/DEHNN · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
import os
import numpy as np
import pickle
import torch
import torch.nn
from torch_geometric.data import Dataset
from torch_geometric.data import Data, HeteroData
from torch_geometric.loader import NeighborLoader
from torch_geometric.nn.conv.gcn_conv import gcn_norm
from utils import *
import torch.nn as nn
import torch.optim as optim
from tqdm import tqdm

import matplotlib.pyplot as plt
from torch_geometric.nn import global_mean_pool, global_max_pool, global_add_pool

import sys
sys.path.insert(1, 'data/')
from pyg_dataset import NetlistDataset
from gen_h_dataset import process_vlsi_dataset
from data.utils import compute_degrees

sys.path.append("models/")
sys.path.append("models/layers/")
from models.model import GNN_node
from torch_geometric.utils import scatter

import logging
from typing import Dict, Tuple


def setup_logging():
    logging.basicConfig(
        filename='evaluation.txt',
        level=logging.INFO,
        format='%(asctime)s - %(message)s'
    )

def evaluate_model(model, h_dataset, device, config: Dict) -> Tuple[float, float, float, float]:
    model.eval()
    total_nrmse = total_ssim = total_mape = total_r2 = 0
    total_node_nrmse = total_node_ssim = total_node_mape = total_node_r2 = 0
    num_place = 0

    with torch.no_grad():
        for data_idx in tqdm(range(len(h_dataset))):
            data = h_dataset[data_idx]
            data['node', 'as_a_sink_of', 'net'].edge_index = data['node', 'as_a_sink_of', 'net'].edge_index.to(device)
            data['node', 'as_a_source_of', 'net'].edge_index = data['node', 'as_a_source_of', 'net'].edge_index.to(device)

            for inner_data_idx in range(len(data.variant_data_lst)):
                pos_lst, pos_lst_net, edge_attr, target_node, target_net = data.variant_data_lst[inner_data_idx]
                if (torch.sum(target_node) < 10) or (torch.sum(target_net) < 10):
                    continue

                data['node'].x = torch.concat([data['node'].node_features, pos_lst], dim=1)
                data['net'].x = torch.concat([data['net'].net_features, pos_lst_net], dim=1)
                data['node', 'as_a_sink_of', 'net'].edge_attr = edge_attr
                data.pos_lst = pos_lst
                data.num_sites_x = config['num_sites_x']
                data.num_sites_y = config['num_sites_y']
                target_node = target_node.to(device)

                node_representation, net_representation, vn_representation = model(data, device)
                target_map = convert_to_util_map(pos_lst, target_node, config['num_sites_x'],
                                               config['num_sites_y'], device, "amax").flatten()

                node_map = vn_representation.squeeze(1)
                total_nrmse += nrmse(target_map, node_map).item()
                total_ssim += ssim(target_map, node_map).item()
                total_mape += mape(target_map, node_map).item()
                total_r2 += r2_score(target_map, node_map).item()

                total_node_nrmse += nrmse(target_node, node_representation.flatten()).item()
                total_node_ssim += ssim(target_node, node_representation.flatten()).item()
                total_node_mape += mape(target_node, node_representation.flatten()).item()
                total_node_r2 += r2_score(target_node, node_representation.flatten()).item()

                num_place += 1

                if (total_nrmse > 100000) or (total_r2 < 0):
                    raise ValueError("Invalid metrics detected")

    metrics = {
        'map': (total_nrmse/num_place, total_ssim/num_place,
                total_mape/num_place, total_r2/num_place),
        'node': (total_node_nrmse/num_place, total_node_ssim/num_place,
                 total_node_mape/num_place, total_node_r2/num_place)
    }
    return metrics, num_place

def main():
    setup_logging()
    device = "cuda"
    num_sites_x, num_sites_y = 206, 300

    num_layer = 2
    num_dim = 32
    aggr = "add"
    device = "cuda"

    learning_rate = 0.0005

    configs = [
        #{'vn': False, 'cv': False, 'model_type': 'dehnn'},
        #{'vn': True, 'cv': False, 'model_type': 'dehnn'},
        #{'vn': True, 'cv': True, 'model_type': 'dehnn'},
        {'vn': True, 'cv': True, 'model_type': 'unet'}
    ]

    load_indices = np.array(['221', '181', '226', '206', '191', '190', '192', '182', '222', '197', '71', '81', '151', '161', '106', '160', '112', '75', '37', '82', '21', '45', '102', '140', '7'])
    all_load_indices = []
    for file in os.listdir("data/all_designs_netlist_data/"):
        index = file.split("_")[1]
        all_load_indices.append(index)

    test_load_indices = []
    for index in all_load_indices:
        if index not in load_indices:
            test_load_indices.append(index)

    test_load_indices = np.array(test_load_indices)
    load_indices = test_load_indices
    test = False # if only test but not train
    restart = True # if restart training
    reload_dataset = False # if reload already processed h_dataset
    target_data_dir = "data/target_data"

    dataset = NetlistDataset(data_dir="data/all_designs_netlist_data",
                            load_pe=True, pl=True, processed=True,
                            load_indices=load_indices)
    h_dataset = process_vlsi_dataset(dataset, target_data_dir)

    for config in configs:
        try:
            model_path = f"{config['model_type']}_{num_layer}_{num_dim}_{config['vn']}_{config['cv']}_model.pt"
            if not os.path.exists(model_path):
                logging.warning(f"Model not found: {model_path}")
                continue

            model = torch.load(model_path).to(device)
            logging.info(f"\nEvaluating configuration: {config}")

            metrics, num_place = evaluate_model(model, h_dataset, device,
                                             {'num_sites_x': num_sites_x,
                                              'num_sites_y': num_sites_y})

            logging.info(f"Number of placements evaluated: {num_place}")
            logging.info("Map Metrics:")
            logging.info(f"NRMSE: {metrics['map'][0]:.3f}")
            logging.info(f"SSIM: {metrics['map'][1]:.3f}")
            logging.info(f"MAPE: {metrics['map'][2]:.3f}")
            logging.info(f"R2: {metrics['map'][3]:.3f}")

            logging.info("Node Metrics:")
            logging.info(f"NRMSE: {metrics['node'][0]:.3f}")
            logging.info(f"SSIM: {metrics['node'][1]:.3f}")
            logging.info(f"MAPE: {metrics['node'][2]:.3f}")
            logging.info(f"R2: {metrics['node'][3]:.3f}")

        except Exception as e:
            logging.error(f"Error evaluating config {config}: {str(e)}")

if __name__ == "__main__":
    main()