first

draw.py

+import matplotlib.pyplot as plt
+import numpy as np
+
+# 数据
+labels = ['4', '8', '12', '16']
+methods = ['TGL', 'MSPipe',  'Ours']
+data_lastfm_tgn = [
+
+    [0.7820, 0.8277, 0.7683, 0.8998],  # 4
+    [None, 0.7846, 0.7419, 0.9329],  # 8
+    [None, 0.770734,0.7713,None ],  # 12
+    [None, 0.0, 0.0, 0.0, 0.0],  # 16
+]
+data_wikitalk_tgn = [
+
+        [0.9632, 0.9552 ,0.9355,0.9765],
+        [None,0.9361,0.9301 ,0.9800] ,
+        [None,0.930091,0.9234,None],
+        [None,None,None,None,None]
+]
+
+data_stackoverflow_tgn = [
+
+    [0.9547 ,0.9548,0.9273,0.9760 ],
+    [None,0.9217 ,0.9328 , 0.9792 ],
+    [None, 0.911069,None,None],
+    [None,None,None,None,None]
+]
+
+			
+
+data_gdelt_tgn = [
+
+    [0.977047,0.9832 ,0.9860 ,0.9866 ],
+    [None,0.9867 ,0.9860 ,0.9897 ],
+    [None, 986776,None,None],
+    [None,None,None,None,None]
+]
+
+data_lastfm_jodie = [
+[0.6116 ,	0.4998 	 , 0.6182 ],
+[0      ,   0.556566	,0.6131 ],
+[0	    ,    0.551095,	0],
+[0	    ,    0,	0],
+]
+
+data_lastfm_apan = [
+[0.5711 ,	0.5975 ,	0.6741 ],
+[0.0000 ,	0.5509 ,	0.5657 ],
+[0	,0.528071	,0.0000 ],
+[0	    ,    0,	0],
+
+]
+
+data_wikitalk_jodie = [
+[0.9506 ,	0.9123 ,	0.9541 ],
+[0.0000 ,	0.825151	,0.9495 ],
+[0	,0.798259,	0],
+[0	    ,    0,	0],
+
+]
+data_wikitalk_apan = [
+[0.9136 ,	0.9147 ,	0.9298 ],
+[0.0000 ,	0.7030 ,	0.9398 ],
+[0.0000 	,0.640181,	0.0000 ],
+[0	    ,    0,	0],
+
+]
+data_stackoverflow_jodie = [
+[0.9512 ,	0.8856 ,	0.9612 ],
+[0.0000 ,	0.8054985,	0.9580 ],
+[0	,0.78055,	0],
+[0,0,0]
+]
+data_stackoverflow_apan =[
+[0.8886 	,0.8615 ,	0.9400 ],
+[0.0000 ,	0.6824 	,0.9402 ],
+[0,	0.633996	,0.0000 ],
+[0	    ,  0,	0],
+
+]
+data_gdelt_jodie = [
+[0.9825 ,	0.983089	,0.9901 ],
+[0.0000 ,	0.9812035	,0.9914 ],
+[0	,0.9811305,	0],
+[0,0,0]
+		
+]
+data_gdelt_apan = [
+[0.9547 ,	0.9361 ,	0.9839 ],
+[0.0000 ,	0.7252, 	0.9876 ],
+[0	,0.601204	,0.0000 ],
+[0,0,0]
+
+]
+datasets = ['LASTFM','WikiTalk','StackOverflow','GDELT']
+models = ['apan','jodie']
+# 设置柱状图参数
+for dataset in datasets:
+    for model in models:
+        x = np.arange(len(labels))  # x轴位置
+        width = 0.15  # 每个柱的宽度
+        colormap = plt.get_cmap('Accent')
+        colors = colormap(np.linspace(0,1,5))
+        # 绘制柱状图
+
+        fig, ax = plt.subplots(figsize=(10, 6))
+
+        for i, method in enumerate(methods):
+             data = globals()['data_{}_{}'.format(dataset.lower(),model.lower())]
+             print(data)
+             method_data = [d[i] if d[i] is not None else 0 for d in data]
+             ax.bar(x + i * width, method_data, width, label=method,color=colors[3 if i==2 else i])
+             plt.ylim([0.45,1])
+             # 添加标签和标题
+             ax.set_xlabel('Number of GPUs')
+             ax.set_ylabel('Test AP')
+             ax.set_title('{}({})'.format(dataset,model))
+             ax.set_xticks(x + width * 2)
+             ax.set_xticklabels(labels)
+             ax.legend()
+
+        # 显示图形
+        plt.tight_layout()
+        plt.savefig('{}_{}_ap_number_of_gpus.png'.format(dataset,model))
+        plt.clf()
\ No newline at end of file

draw_comm.py

+import matplotlib.pyplot as plt
+import torch
+# 数据
+datasets = {
+   #"WIKI": [117195, 30114, 17527, 4142, 0],
+   "LASTFM": [768973, 350190, 230442, 66460, 0],
+   "WikiTalk": [6847641, 1148222, 643309, 144365, 0],
+   "StackOverflow": [94415023, 15062649, 8482896, 1948469, 0],
+   "GDELT": [98307910, 20454305, 11504991, 2722967, 0],}
+#
+#
+# X轴标签
+#x_labels = ["recent", r'0.1', r'0.05', r'0.01', r'0.00']
+#
+# datasets = {
+# #"WIKI":    [21603, 	11919 ,	4700 ,	0],
+# "LASTFM": [9471 	,8263 	,1889 ,	0 ],
+# "WikiTalk":[2561632 	,458949 ,	192920,0], 
+# "StackOverflow": [28092533 ,	5956559 ,	577277,0], 
+# "GDELT": [28911769 ,	1306950 ,	85580,0], 
+
+# }
+x_labels = ["recent", r'0.1', r'0.05', r'0.01', r'0.00']
+#x_labels = ["recent",r'$\theta$=0.1',r'$\theta$=0.05',r'$\theta=0.01','$\theta=0']
+#x_labels = ["all_update", r'0.3', r'0.7','local']
+# 为每个数据集绘制柱状图
+fig, axes = plt.subplots(2, 2, figsize=(10, 8), sharey=False)
+for nid,(dataset, values) in enumerate(datasets.items()):
+    print(int(nid/2),int(nid%2))
+    ax = axes[int(nid/2)][int(nid%2)]
+    #fig, ax = plt.subplots()
+    v = torch.tensor(values) * (100*4) * 4 /1024/1024
+    hights = v / v[0]
+    bar_width = 0.5
+    bars = ax.bar(x_labels, v,width=bar_width )
+    print(hights)
+    for i,bar in enumerate(bars):
+        height = bar.get_height()
+        print(hights[i])
+        ax.text(
+            bar.get_x() + bar.get_width() / 2,  # x位置
+            height,                             # y位置
+            f'{hights[i]:.2f} x',                        # 标注文本
+            ha='center',                        # 水平居中
+            va='bottom',                         # 垂直底部对齐,
+            fontsize=17
+        )
+
+    ax.set_title(f'{dataset} Dataset', fontsize=17)
+    ax.set_xlabel(r'boundary sample hyperparameter $\theta$', fontsize=17)
+    #ax.set_xlabel(r'synchronization hyperparameter $\alpha$', fontsize=17)
+    ax.set_ylabel('Communication Volume(MB)', fontsize=17)
+    
+    ax.tick_params(axis='y', labelsize=17)
+    ax.tick_params(axis='x', labelsize=17)
+    ax.set_yscale('log')  # 使用对数刻度
+    ax.set_ylim(0, max(v) * 2)
+    ax.grid(axis='y', linestyle='--', alpha=1)
+    #ax.set_show()
+plt.subplots_adjust(left=0.1, right=0.96, top=0.9, bottom=0.1, wspace=0.5, hspace=0.5)
+plt.tight_layout(rect=[0,0,0,0])
+plt.savefig('historical_exp/boundary_sample_comm.png')
+#plt.savefig('historical_exp/historical_comm.png')

draw_inference.py

+labels = ['4', '8', '12', '16']
+methods = ['TGL', 'MSPipe','DistTGL','Ours']
+table_label = ['TGL-1','TGL-4','MSPipe-4','DistTGL-4','Ours-4','MSPipe-8','DistTGL-8','Ours-8','MSPipe-12','DistTGL-12','Ours-12','MSPipe-16','DistTGL-16','Ours-16']
+table_label_no = ['TGL-1','TGL-4','MSPipe-4','Ours-4','MSPipe-8','Ours-8','MSPipe-12','Ours-12','MSPipe-16','Ours-16']
+dataset_label = ['WIKI','LASTFM','WikiTalk','StackOverflow','GDELT']
+table_ap_tgn = [
+    [0.9827, 0.8023, 0.9611, 0.9574, 0.9837],#tgl-1
+    [0.9808, 0.7820, 0.9632, 0.9547, 0.9770],#tgl-4
+    [0.9709, 0.8277, 0.9552, 0.9548, 0.9832],#mspipe
+    [0.9682, 0.7683, 0.9355, 0.9273, 0.9860],#
+    [0.9770, 0.8998, 0.9765, 0.9760, 0.9866],
+    [0.9703, 0.7846, 0.9361, 0.9216, 0.9867],
+    [0.9648, 0.7419, 0.9301, 0.9328, 0.9860],
+    [0.9768, 0.9329, 0.9800, 0.9792, 0.9897],
+    [0.9698, 0.7707, 0.9301, 0.9111, 0.9868],
+    [0.9624, 0.7713, 0.9234, 0.9074, 0.9871],
+    [0.0000, 0.0000, 0.0000, 0.0000, 0.0000],
+    [0.9699, 0.7868, 0.9257, 0.9090, 0.9870],
+    [0.9646, 0.6999, 0.8997, 0.9031, 0.9862],
+    [0.9659, 0.9498, 0.9816, 0.9830, 0.9916]
+]
+
+table_ap_jodie = [
+    [0.9080, 0.6202, 0.9578, 0.9620, 0.9831],#tgl-1
+    [0.8586, 0.6116, 0.9463 ,0.9494 , 0.9825],#tgl-4
+    [0.785868, 0.4998, 0.9123, 0.8856, 0.9831],
+    [0.9004, 0.6182, 0.9541, 0.9612, 0.9901],
+    [0.7735, 0.5566, 0.8252, 0.8055, 0.9812],
+    [0.8769, 0.6131, 0.9495, 0.9580, 0.9914],
+    [0.7747, 0.5511, 0.7983, 0.7806, 0.9811],
+    [0.0000, 0.0000, 0.0000, 0.0000, 0.0000],
+    [0.7691, 0.5427, 0.7858, 0.7684, 0.9810],
+    [0.8548, 0.6044, 0.9418, 0.9540, 0.9908]
+]
+
+table_ap_apan = [
+    [0.9415, 0.5659, 0.9043, 0.8752, 0.9612],
+    [0.8834, 0.5711, 0.9136, 0.8886, 0.9547],
+    [0.9124, 0.5975, 0.9147, 0.8615, 0.9361],
+    [0.9190, 0.6741, 0.9298, 0.9400, 0.9839],
+    [0.6752, 0.5509, 0.7030, 0.6824, 0.7252],
+    [0.8982, 0.5657, 0.9398, 0.9402, 0.9876],
+    [0.6011, 0.5281, 0.6402, 0.6340, 0.6012],
+    [0.0000, 0.0000, 0.0000, 0.0000, 0.0000],
+    [0.5449, 0.5156, 0.6004, 0.5898, 0.6163],
+    [0.8658, 0.6682, 0.9419, 0.9522, 0.9880]
+]
+
+table_train_tgn =[
+[1.7701 	,14.7205 	,88.1896 	,938.8964 ,	2001.318524],
+[0.6651 	,6.185821226	,35.9636 ,	328.1245 ,	854.80837],
+[0.8258 	,3.7705 	,15.3471 ,	141.1648 	,465.2264 ],
+[0.4001 	,3.5691 	,16.5933 ,	167.2491 ,	445.4328 ],
+[0.6327 	,4.4322 	,14.7757 ,	132.7098208,	537.1398258],
+[0.7085 	,2.2837 	,8.6870 	,76.9830 	,250.0945 ],
+[0.2992 	,2.0908 	,9.0872 	,97.8960 	,238.3315 ],
+[0.2772 	,2.3960 	,7.1694 	,72.0319 	,251.7710 ],
+[0.598028193	,1.708853126	,6.600371795	,59.73907948	,172.2328092],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+
+]
+table_train_jodie  =[
+[0.7056 ,	5.9280, 	16.0147 ,	134.6145 ,	376.2186 ],
+[0.2494 ,	2.0697, 	6.4145 	,52.0684 	,147.3938 ],
+[0.5604 ,	2.0900, 	4.5566 	,33.3442 	,84.38458259],
+[0.3246 ,	2.5305, 	5.7808 	,48.8147 	,148.8899 ],
+[0.518362013,	1.148036031	,2.871488156	,19.42606455,	48.72649877],
+[0.1708 ,	1.3234 	,3.1610 	,26.3006 ,	78.1362 ],
+[0.495687011	,0.929189181,	2.223804102	,15.14540188,	36.91594124],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+
+]
+table_train_apan =[
+[1.5406,	13.5400, 	118.8137, 	1032.4779 ,	1609.9425 ],
+[0.6653,	5.6160 ,	41.4232 ,	368.6106 ,	1045.6425 ],
+[0.7319,	2.9211 ,	25.9092 ,	238.7242 ,	643.3855 ],
+[0.4428,	3.3608 ,	11.3518 ,	96.5700 ,	244.8932 ],
+[0.6765,	1.9470 ,	14.0065 ,	123.8830 ,	331.8195 ],
+[0.2521,	2.0037 ,	8.5989 	,74.9677 ,	139.2487 ],
+[0.870823467,	2.671342006,	14.96090715,	99.97286532,	233.0780075],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+]
+table_eval_tgn = [
+[0.7499 	,4.8555 	,48.0304 	,505.1028 	,824.7955338],
+[0.2636 	,2.09782438	,15.9930 	,162.6785 	,412.4661068],
+[0.5704 	,1.2416 	,6.4172 	,68.7693 	,117.9146 ],
+[0.2692 	,2.0977 	,13.2586 	,163.4167 	,152.3379 ],
+[0.1166 	,0.8080 	,3.0650 	,28.00140551,	115.4568481],
+[0.5747 	,0.8989 	,4.0630 	,36.5432 	,61.79147074],
+[0.2714 	,2.0767 	,11.6519 	,143.2905 	,145.8947 ],
+[0.0627 	,0.6212 	,2.5015 	,23.1395 	,80.9373 ],
+[0.5550 	,0.7799 	,3.3021 	,27.7136 	,42.9780 ],
+[0.2738 	,2.0794 	,15.4595 	,152.1446 	,148.1033 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.7972 	,1.0099 	,2.7960 	,21.3148 	,35.1295 ],
+[0.2733 	,2.0562 	,13.1791 	,162.9651 	,151.0929 ],
+[0.0783 	,0.5772 	,1.8858 	,19.7090 	,64.9535 ],
+
+]
+table_eval_jodie =[
+[0.2091 	,1.6959 	,5.9168 ,48.6228 	,129.0989 ],
+[0.0719 	,0.6373 	,2.1611 ,18.0129 	,50.1390 ],
+[0.5078 	,0.4018 	,2.7975 ,15.9332 	,28.17862575],
+[0.0633 	,0.5023 	,1.1726 ,9.4948 	,42.8286 ],
+[0.502160359	,0.669311229	,2.076292474	,9.030235369,	16.12763978],
+[0.0377 	,0.3370 	,1.2482 ,	10.2044 	,34.6001 ],
+[0.525156487	,0.617744214,	1.698658919	,6.896312807	,12.20505439],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+]
+table_eval_apan =[
+[0.5907 ,	4.8906 	,37.4472 	,332.8560,	516.5578 ],
+[0.2405 ,	1.9909 	,11.9007 	,105.6109,	298.5957 ],
+[0.6249 ,	1.4006 	,8.5734 	,66.5308 ,	161.1591 ],
+[0.0819 ,	0.6588 	,2.2410 	,19.3600 ,	100.0944 ],
+[0.7001 ,	1.1295 	,4.9139 	,38.4971 ,	68.67845647],
+[0.0709 ,	0.8897 	,4.9156 	,41.6653 ,	114.7303 ],
+[0.860863774,	1.248189147	,4.57928021	,30.29166182,	69.67092261],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+]
+
+def get_data(model,data,method,part,mode):
+    table = globals()['table_{}_{}'.format(mode,model.lower())]
+    index_name = method+'-{}'.format(part)
+    if model =='TGN':
+        index = table_label.index(index_name) if index_name in table_label else -1
+        if index == -1:
+            return None
+        else:
+            #print(index,dataset_label.index(data),method)
+            return table[index][dataset_label.index(data)]
+    else:
+        index = table_label_no.index(index_name) if index_name in table_label_no else -1
+        if index == -1:
+            return None
+        else:
+            return table[index][dataset_label.index(data)]
+
+import matplotlib.pyplot as plt
+import numpy as np
+import torch
+# 读取文件内容
+import os
+datasets = ['LASTFM', 'WikiTalk', 'StackOverflow', 'GDELT']
+methods = ['DistTGL', 'Ours']
+cards = ['4', '8', '16']
+fig, axes = plt.subplots(1, len(datasets), figsize=(12, 3), sharey=False)
+
+bar_width = 0.2
+index = np.arange(len(cards))
+handles, labels = [], []
+for i, dataset in enumerate(datasets):
+    ax = axes[i]
+    
+    # 绘制柱状图
+    for j, method in enumerate(methods):
+        eval_times = [get_data('TGN',dataset,method,k,'eval') for k in cards]
+        #print(times)
+        ax.bar(index + j * bar_width, eval_times, bar_width, label=f'{method} Time')
+    for h, l in zip(*ax.get_legend_handles_labels()):
+        if l not in labels:
+            handles.append(h)
+            labels.append(l)
+    ax.set_title(dataset, fontsize=14)
+    ax.set_xticks(index + bar_width)
+    ax.set_xticklabels(cards, fontsize=12)
+    ax.set_xlabel('Number of GPUs', fontsize=12)
+    #ax.set_ylim(0, max(eval_times)*4)
+    
+    if i == 0:
+        ax.set_ylabel('Evaluation Time', fontsize=12)
+
+
+fig.legend(handles, labels, loc='upper center', ncol=6, fontsize=10)
+
+plt.tight_layout(rect=[0, 0, 1, 0.9])
+plt.savefig('scalability/TGN_inference.png')

draw_scalaility.py

+import matplotlib.pyplot as plt
+import numpy as np
+
+# 定义数据
+datasets = ['LASTFM', 'WikiTalk', 'StackOverflow', 'GDELT']
+methods = ['DistTGL', 'MSPipe', 'Ours']
+cards = ['4', '8', '16']
+
+# 每个数据集对应的训练时间和精度
+train_time = {
+    'LASTFM': {
+        'DistTGL': [3.5691, 2.0908, 1.1308],
+        'MSPipe': [3.7705, 2.2837, 1.58013],
+        'Ours': [4.4322, 2.3960, 1.99681]
+    },
+    'WikiTalk': {
+        'DistTGL': [16.5933, 9.0872, 5.8111],
+        'MSPipe': [15.3471, 8.6870, 5.23049],
+        'Ours': [14.7757, 7.1694, 5.199244]
+    },
+    'StackOverflow': {
+        'DistTGL': [167.2491, 97.8960, 56.4550],
+        'MSPipe': [141.1648, 76.9830, 47.676975],
+        'Ours': [132.7098, 72.0319, 62.127397]
+    },
+    'GDELT': {
+        'DistTGL': [445.4328, 238.333, 145.6853],
+        'MSPipe': [465.2264, 250.0945, 134.19114],
+        'Ours': [537.13983, 251.7710, 138.56199]
+    }
+}
+
+accuracy = {
+    'LASTFM': {
+        'DistTGL': [0.7683, 0.7419, 0.6999],
+        'MSPipe': [0.8277, 0.7846, 0.78682],
+        'Ours': [0.8998, 0.9329, 0.94982]
+    },
+    'WikiTalk': {
+        'DistTGL': [0.9355, 0.9301, 0.8997],
+        'MSPipe': [0.9552, 0.9376, 0.92567],
+        'Ours': [0.9765, 0.9800, 0.98155]
+    },
+    'StackOverflow': {
+        'DistTGL': [0.9273, 0.9328, 0.9031],
+        'MSPipe': [0.9548, 0.9548, 0.908905],
+        'Ours': [0.9760, 0.9792, 0.983042]
+    },
+    'GDELT': {
+        'DistTGL': [0.9860, 0.9860, 0.9862],
+        'MSPipe': [0.9832, 0.9867, 0.9870182],
+        'Ours': [0.9866, 0.9897, 0.991622]
+    }
+}
+
+# 使用默认样式
+fig, axes = plt.subplots(1, len(datasets), figsize=(12, 3), sharey=False)
+
+bar_width = 0.2
+index = np.arange(len(cards))
+handles, labels = [], []
+for i, dataset in enumerate(datasets):
+    ax = axes[i]
+    
+    # 绘制柱状图
+    for j, method in enumerate(methods):
+        times = [train_time[dataset][method][k] for k in range(len(cards))]
+        ax.bar(index + j * bar_width, times, bar_width, label=f'{method} Time')
+    for h, l in zip(*ax.get_legend_handles_labels()):
+        if l not in labels:
+            handles.append(h)
+            labels.append(l)
+    ax.set_title(dataset, fontsize=14)
+    ax.set_xticks(index + bar_width)
+    ax.set_xticklabels(cards, fontsize=12)
+    ax.set_xlabel('Number of GPUs', fontsize=12)
+    ax.set_ylim(0, max(max(times) for times in train_time[dataset].values()) * 2)
+    
+    if i == 0:
+        ax.set_ylabel('Training Time', fontsize=12)
+    
+    # 创建双轴并绘制折线图
+    ax2 = ax.twinx()
+    for j, method in enumerate(methods):
+        acc = [accuracy[dataset][method][k] for k in range(len(cards))]
+        ax2.plot(index +  bar_width, acc, marker='o', linestyle='--', label=f'{method} AP',markersize=5)
+    for h, l in zip(*ax2.get_legend_handles_labels()):
+        print(l)
+        if l not in labels:
+            handles.append(h)
+            labels.append(l)
+    if dataset == 'GDELT':
+        ax2.set_ylabel('Test AP', fontsize=12)
+    #axes[i].twinx().set_ylim(0.95, 1.0)
+
+
+fig.legend(handles, labels, loc='upper center', ncol=6, fontsize=14)
+
+plt.tight_layout(rect=[0, 0, 1, 0.9])
+plt.savefig('scalability/TGN.png')

draw_shared.py

+import matplotlib.pyplot as plt
+
+# 数据
+x_values = [0.02,0.04,0.06,0.08]
+y_values = [0.310754, 0.19102, 0.117148, 0.075549]
+
+# 创建折线图
+plt.figure(figsize=(10, 6))
+plt.plot(x_values,y_values, marker='o', linestyle='-', color='b', linewidth=2, markersize=8)
+
+# 添加标题和标签
+plt.title('Number of Cross Edge vs shared node ratio', fontsize=16)
+plt.xlabel('The ratio of Cross Edge', fontsize=14)
+plt.ylabel('shared nodes ratio', fontsize=14)
+
+# 显示网格
+plt.grid(True, linestyle='--', alpha=0.7)
+
+# 显示图表
+plt.show()
+plt.savefig('shared_ratio.png')

draw_time.py

+import matplotlib.pyplot as plt
+import numpy as np
+
+# 数据
+labels = ['4', '8', '12', '16']
+methods = ['TGL', 'MSPipe','DistTGL','Ours']
+table_label = ['TGL-1','TGL-4','MSPipe-4','DistTGL-4','Ours-4','MSPipe-8','DistTGL-8','Ours-8','MSPipe-12','DistTGL-12','Ours-12','MSPipe-16','DistTGL-16','Ours-16']
+table_label_no = ['TGL-1','TGL-4','MSPipe-4','Ours-4','MSPipe-8','Ours-8','MSPipe-12','Ours-12','MSPipe-16','Ours-16']
+dataset_label = ['WIKI','LASTFM','WikiTalk','StackOverflow','GDELT']
+table_train_tgn =[
+[1.7701 	,14.7205 	,88.1896 	,938.8964 ,	2001.318524],
+[0.6651 	,6.185821226	,35.9636 ,	328.1245 ,	854.80837],
+[0.8258 	,3.7705 	,15.3471 ,	141.1648 	,465.2264 ],
+[0.4001 	,3.5691 	,16.5933 ,	167.2491 ,	445.4328 ],
+[0.6327 	,4.4322 	,14.7757 ,	132.7098208,	537.1398258],
+[0.7085 	,2.2837 	,8.6870 	,76.9830 	,250.0945 ],
+[0.2992 	,2.0908 	,9.0872 	,97.8960 	,238.3315 ],
+[0.2772 	,2.3960 	,7.1694 	,72.0319 	,251.7710 ],
+[0.598028193	,1.708853126	,6.600371795	,59.73907948	,172.2328092],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+[0.0000 	,0.0000 	,0.0000 	,0.0000 	,0.0000 ],
+
+]
+table_train_jodie  =[
+[0.7056 ,	5.9280, 	16.0147 ,	134.6145 ,	376.2186 ],
+[0.2494 ,	2.0697, 	6.4145 	,52.0684 	,147.3938 ],
+[0.5604 ,	2.0900, 	4.5566 	,33.3442 	,84.38458259],
+[0.3246 ,	2.5305, 	5.7808 	,48.8147 	,148.8899 ],
+[0.518362013,	1.148036031	,2.871488156	,19.42606455,	48.72649877],
+[0.1708 ,	1.3234 	,3.1610 	,26.3006 ,	78.1362 ],
+[0.495687011	,0.929189181,	2.223804102	,15.14540188,	36.91594124],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+[0.0000,	0.0000 ,	0.0000 	,0.0000 ,	0.0000 ],
+
+]
+table_train_apan =[
+[1.5406,	13.5400, 	118.8137, 	1032.4779 ,	1609.9425 ],
+[0.6653,	5.6160 ,	41.4232 ,	368.6106 ,	1045.6425 ],
+[0.7319,	2.9211 ,	25.9092 ,	238.7242 ,	643.3855 ],
+[0.4428,	3.3608 ,	11.3518 ,	96.5700 ,	244.8932 ],
+[0.6765,	1.9470 ,	14.0065 ,	123.8830 ,	331.8195 ],
+[0.2521,	2.0037 ,	8.5989 	,74.9677 ,	139.2487 ],
+[0.870823467,	2.671342006,	14.96090715,	99.97286532,	233.0780075],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 	,0.0000 ,	0.0000 ,	0.0000 ],
+]
+table_eval_tgn = [
+[0.7499 	,4.8555 	,48.0304 	,505.1028 	,824.7955338],
+[0.2636 	,2.09782438	,15.9930 	,162.6785 	,412.4661068],
+[0.5704 	,1.2416 	,6.4172 	,68.7693 	,117.9146 ],
+[0.2692 	,2.0977 	,13.2586 	,163.4167 	,152.3379 ],
+[0.1166 	,0.8080 	,3.0650 	,28.00140551,	115.4568481],
+[0.5747 	,0.8989 	,4.0630 	,36.5432 	,61.79147074],
+[0.2714 	,2.0767 	,11.6519 	,143.2905 	,145.8947 ],
+[0.0627 	,0.6212 	,2.5015 	,23.1395 	,80.9373 ],
+[0.555030217	,0.779903646	,3.302122216	,27.71362416	,42.97800901],
+[0.0000 ,	0.0000 ,0.0000 	,0.0000 ,	0.0000 ],
+[0.0000 ,	0.0000 ,0.0000 	,0.0000 ,	0.0000 ],
+[0.0000 ,	0.0000 ,0.0000 	,0.0000 ,	0.0000 ],
+[0.0000 ,	0.0000 ,0.0000 	,0.0000 ,	0.0000 ],
+[0.0000 ,	0.0000 ,0.0000 	,0.0000 ,	0.0000 ],
+
+]
+table_eval_jodie =[
+[0.2091 	,1.6959 	,5.9168 ,48.6228 	,129.0989 ],
+[0.0719 	,0.6373 	,2.1611 ,18.0129 	,50.1390 ],
+[0.5078 	,0.4018 	,2.7975 ,15.9332 	,28.17862575],
+[0.0633 	,0.5023 	,1.1726 ,9.4948 	,42.8286 ],
+[0.502160359	,0.669311229	,2.076292474	,9.030235369,	16.12763978],
+[0.0377 	,0.3370 	,1.2482 ,	10.2044 	,34.6001 ],
+[0.525156487	,0.617744214,	1.698658919	,6.896312807	,12.20505439],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+[0.0000 	,0.0000 ,	0.0000,	0.0000 ,	0.0000 ],
+]
+table_eval_apan =[
+[0.5907 ,	4.8906 	,37.4472 	,332.8560,	516.5578 ],
+[0.2405 ,	1.9909 	,11.9007 	,105.6109,	298.5957 ],
+[0.6249 ,	1.4006 	,8.5734 	,66.5308 ,	161.1591 ],
+[0.0819 ,	0.6588 	,2.2410 	,19.3600 ,	100.0944 ],
+[0.7001 ,	1.1295 	,4.9139 	,38.4971 ,	68.67845647],
+[0.0709 ,	0.8897 	,4.9156 	,41.6653 ,	114.7303 ],
+[0.860863774,	1.248189147	,4.57928021	,30.29166182,	69.67092261],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+[0.0000 ,	0.0000 	,0.0000 	,0.0000 ,0.0000 ],
+]
+
+datasets = ['LASTFM','WikiTalk','StackOverflow','GDELT']
+models = ['TGN','APAN','JODIE']
+# 设置柱状图参数
+for dataset in datasets:
+    for model in models:
+        x = np.arange(len(labels))  # x轴位置
+        width = 0.15  # 每个柱的宽度
+        colormap = plt.get_cmap('Accent')
+        colors = colormap(np.linspace(0,1,5))
+        # 绘制柱状图
+
+        fig, ax = plt.subplots(figsize=(10, 6))
+
+        for i, method in enumerate(methods):
+             if(method == 'DistTGL' and model !='TGN'):
+                 continue
+             table = globals()['table_eval_{}'.format(model.lower())]
+             #data = globals()['data_{}_{}'.format(dataset.lower(),model.lower())]
+             #print(data)
+             method_data = []
+             for g in range(len(labels)):
+                num = int(labels[g])
+                index_name = method+'-{}'.format(num)
+                print(index_name)
+                if model =='TGN':
+                    index = table_label.index(index_name) if index_name in table_label else -1
+                    print(index)
+                    if index == -1:
+                        method_data.append(0)
+                    else:
+                        method_data.append(table[index][dataset_label.index(dataset)])
+                else:
+                    index = table_label_no.index(index_name) if index_name in table_label_no else -1
+                    print(index)
+                    if index == -1:
+                        method_data.append(0)
+                    else:
+                        method_data.append(table[index][dataset_label.index(dataset)])
+             if model == 'TGN':
+                ax.bar(x + i * width, method_data, width, label=method,color=colors[i] )
+             else:
+                ax.bar(x + min(i,2) * width, method_data, width, label=method,color=colors[i] )
+             # 添加标签和标题
+             ax.set_xlabel('Number of GPUs')
+             ax.set_ylabel('Time for test/s')
+             ax.set_title('{}({})'.format(dataset,model))
+             ax.set_xticks(x + width * 2)
+             ax.set_xticklabels(labels)
+             ax.legend()
+
+        # 显示图形
+        plt.tight_layout()
+        plt.savefig('{}_{}_eval_time_number_of_gpus.png'.format(dataset,model))
+        plt.clf()
\ No newline at end of file

historical_exp/draw_breakdown.py

+import matplotlib.pyplot as plt
+import numpy as np
+
+# 数据
+categories = ['recent', r'$\theta$=0.1', r'$\theta$=0.05', r'$\theta$=0.01', r'local']
+forward_computation = [232.8508632, 207.6101028, 207.3353659, 205.5815726, 206.2297827]
+memory_sync = [43.17275703, 15.06231188, 14.61119452, 13.68188678, 14.38758841]
+others = [36.34391371, 69.00549863, 62.89710956, 69.09224575, 47.82174432]
+memory_fetch = [542.8608579, 99.63405462, 69.2147035, 64.4748841, 48.06376967]
+
+# 堆叠柱状图
+fig, ax = plt.subplots(figsize=[4,4])
+
+ax.bar(categories, forward_computation, label='Forward Computation')
+ax.bar(categories, memory_sync, bottom=forward_computation, label='Memory Sync')
+ax.bar(categories, others, bottom=np.array(forward_computation) + np.array(memory_sync), label='Others')
+ax.bar(categories, memory_fetch, bottom=np.array(forward_computation) + np.array(memory_sync) + np.array(others), label='Memory Fetch')
+
+# 添加标签和标题
+ax.set_ylabel('Time/s',fontsize=14)
+ax.legend()
+
+plt.xticks(rotation=45,fontsize = 14)
+plt.tight_layout()
+plt.savefig('boundary_time.png')

partition/accuracy/partition_acc/LASTFM/APAN/8-dis_tgl-0-all_update-boundery_recent_decay-0.1.out

+probability is 0.1
+probability is 0.1
+probability is 0.1
+probability is 0.1
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 3
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 1
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 2
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 0
+LOCAL RANK 1, RANK5
+LOCAL RANK 2, RANK6
+LOCAL RANK 3, RANK7
+LOCAL RANK 0, RANK4
+in
+in
+in
+in
+local rank is 2 world_size is 8 memory group is 0 memory rank is 6 memory group size is 8
+local rank is 3 world_size is 8 memory group is 0 memory rank is 7 memory group size is 8
+local rank is 1 world_size is 8 memory group is 0 memory rank is 5 memory group size is 8
+local rank is 0 world_size is 8 memory group is 0 memory rank is 4 memory group size is 8
+
+
+
+
+[0, 1, 2, 3, 4, 5, 6, 7][0, 1, 2, 3, 4, 5, 6, 7]
+[0, 1, 2, 3, 4, 5, 6, 7]
+[0, 1, 2, 3, 4, 5, 6, 7]
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+memory used is torch.Size([1980, 172]) torch.float32 0.0012686848640441895
+
+dist rank is 4 after node feats defination:
+dist rank is 6 after node feats defination:
+
+dist rank is 5 after node feats defination:
+
+dist rank is 7 after node feats defination:
+
+dist rank is 4 after node feats defination:
+

partition/accuracy/partition_acc/LASTFM/APAN/8-metis-0-all_update-boundery_recent_decay-0.1.out

+probability is 0.1
+probability is 0.1
+probability is 0.1
+probability is 0.1
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 2
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+
+rank 1
+rank 3
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 0
+LOCAL RANK 2, RANK6
+LOCAL RANK 1, RANK5
+LOCAL RANK 3, RANK7
+LOCAL RANK 0, RANK4
+in
+in
+in
+in
+local rank is 2 world_size is 8 memory group is 0 memory rank is 6 memory group size is 8
+local rank is 3 world_size is 8 memory group is 0 memory rank is 7 memory group size is 8
+local rank is 1 world_size is 8 memory group is 0 memory rank is 5 memory group size is 8
+
+
+
+[0, 1, 2, 3, 4, 5, 6, 7][0, 1, 2, 3, 4, 5, 6, 7]
+[0, 1, 2, 3, 4, 5, 6, 7]
+
+local rank is 0 world_size is 8 memory group is 0 memory rank is 4 memory group size is 8
+
+[0, 1, 2, 3, 4, 5, 6, 7]
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+memory used is torch.Size([1980, 172]) torch.float32 0.0012686848640441895
+
+dist rank is 4 after node feats defination:
+dist rank is 6 after node feats defination:
+
+dist rank is 4 after node feats defination:
+
+dist rank is 5 after node feats defination:
+
+dist rank is 7 after node feats defination:
+

partition/accuracy/partition_acc/LASTFM/TGN/12-ours_shared-0.01-local-recent.out

+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 3
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 0
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 2
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 1
+LOCAL RANK 3, RANK3
+LOCAL RANK 0, RANK0
+LOCAL RANK 2, RANK2
+LOCAL RANK 1, RANK1
+in
+in
+in
+in
+local rank is 2 world_size is 12 memory group is 0 memory rank is 2 memory group size is 12
+local rank is 3 world_size is 12 memory group is 0 memory rank is 3 memory group size is 12
+local rank is 0 world_size is 12 memory group is 0 memory rank is 0 memory group size is 12
+
+
+
+[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11][0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
+[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
+
+local rank is 1 world_size is 12 memory group is 0 memory rank is 1 memory group size is 12
+
+[0, 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11]
+../../SPEED/partition/divided_nodes_seed_starrygl/LASTFM/123457/LASTFM_12parts_top0.1/output2.txt
+../../SPEED/partition/divided_nodes_seed_starrygl/LASTFM/123457/LASTFM_12parts_top0.1/output1.txt
+../../SPEED/partition/divided_nodes_seed_starrygl/LASTFM/123457/LASTFM_12parts_top0.1/output3.txt
+../../SPEED/partition/divided_nodes_seed_starrygl/LASTFM/123457/LASTFM_12parts_top0.1/output0.txt
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+tot edge 1293103 circ edge 154993 same edge 0
+
+memory used is torch.Size([1980, 172]) torch.float32 0.0012686848640441895
+
+dist rank is 0 after node feats defination:
+dist rank is 0 after node feats defination:
+
+dist rank is 1 after node feats defination:
+
+dist rank is 2 after node feats defination:
+
+dist rank is 3 after node feats defination:
+

partition/accuracy/partition_acc/LASTFM/TGN/8-dis-tgl-0-all_update-boundery_recent_decay-0.1.out

+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 2
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 3
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 1
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 0
+LOCAL RANK 3, RANK7
+LOCAL RANK 1, RANK5
+LOCAL RANK 0, RANK4
+LOCAL RANK 2, RANK6
+in
+in
+in
+in
+local rank is 2 world_size is 8 memory group is 0 memory rank is 6 memory group size is 8
+local rank is 3 world_size is 8 memory group is 0 memory rank is 7 memory group size is 8
+
+local rank is 1 world_size is 8 memory group is 0 memory rank is 5 memory group size is 8
+
+[0, 1, 2, 3, 4, 5, 6, 7]
+
+[0, 1, 2, 3, 4, 5, 6, 7]
+[0, 1, 2, 3, 4, 5, 6, 7]
+local rank is 0 world_size is 8 memory group is 0 memory rank is 4 memory group size is 8
+
+[0, 1, 2, 3, 4, 5, 6, 7]

partition/accuracy/partition_acc/LASTFM/TGN/8-dis_tgl-0-all_update-boundery_recent_decay-0.1.out

+probability is 0.66
+probability is 0.66
+probability is 0.66
+probability is 0.66
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 2
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 0
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 1
+Network interfaces with prefix '('ens4f1np1', 'ens6f0np0')': ['ens4f1np1']
+rank 3