我正在尝试使用 GPU 加速我的 XGBClassifier。
我已经尝试过这段代码，但 GPU 和 CPU 需要相同的时间：
import xgboost as xgb
from sklearn.datasets import fetch_covtype
from sklearn.model_selection import train_test_split
import time
# Fetch dataset using sklearn
cov = fetch_covtype()
X = cov.data
y = cov.target
# Create 0.75/0.25 train/test split
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.25, train_size=0.75,
random_state=42)
# Specify sufficient boosting iterations to reach a minimum
num_round = 100
# Leave most parameters as default
param = {'objective': 'multi:softmax', # Specify multiclass classification
'num_class': 8, # Number of possible output classes
'tree_method': 'gpu_hist' # Use GPU accelerated algorithm
}
# Convert input data from numpy to XGBoost format
dtrain = xgb.DMatrix(X_train, label=y_train)
dtest = xgb.DMatrix(X_test, label=y_test)
gpu_res = {} # Store accuracy result
tmp = time.time()
# Train model
xgb.train(param, dtrain, num_round, evals=[(dtest, 'test')], evals_result=gpu_res)
print("GPU Training Time: %s seconds" % (str(time.time() - tmp)))
# Repeat for CPU algorithm
tmp = time.time()
param['tree_method'] = 'hist'
cpu_res = {}
xgb.train(param, dtrain, num_round, evals=[(dtest, 'test')], evals_result=cpu_res)
print("CPU Training Time: %s seconds" % (str(time.time() - tmp)))
[0] test-mlogloss:1.47278
[1] test-mlogloss:1.20606
[2] test-mlogloss:1.03133
[3] test-mlogloss:0.91109
[4] test-mlogloss:0.82384
[5] test-mlogloss:0.75852
[6] test-mlogloss:0.70788
[7] test-mlogloss:0.66899
[8] test-mlogloss:0.63753
[9] test-mlogloss:0.61094
[10] test-mlogloss:0.59005
[11] test-mlogloss:0.57204
[12] test-mlogloss:0.55773
.....
[97] test-mlogloss:0.32834
[98] test-mlogloss:0.32719
[99] test-mlogloss:0.32589
GPU Training Time: 17.633554697036743 seconds
[0] test-mlogloss:1.47291
[1] test-mlogloss:1.20547
[2] test-mlogloss:1.03109
[3] test-mlogloss:0.91091
[4] test-mlogloss:0.82473
.....
[98] test-mlogloss:0.32501
[99] test-mlogloss:0.32282
CPU Training Time: 17.619328022003174 seconds
使用 nvidia-smi
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 470.57.02 Driver Version: 470.57.02 CUDA Version: 11.4 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
| | | MIG M. |
|===============================+======================+======================|
| 0 NVIDIA GeForce ... On | 00000000:01:00.0 Off | N/A |
| N/A 48C P8 5W / N/A | 2666MiB / 5934MiB | 6% Default |
| | | N/A |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: |
| GPU GI CI PID Type Process name GPU Memory |
| ID ID Usage |
|=============================================================================|
| 0 N/A N/A 1051 G /usr/lib/xorg/Xorg 227MiB |
| 0 N/A N/A 1364 G /usr/bin/gnome-shell 49MiB |
| 0 N/A N/A 3003 G /usr/lib/firefox/firefox 191MiB |
| 0 N/A N/A 78204 G ...AAAAAAAAA= --shared-files 32MiB |
| 0 N/A N/A 100461 G /usr/lib/firefox/firefox 1MiB |
| 0 N/A N/A 102349 G .../debug.log --shared-files 7MiB |
| 0 N/A N/A 102856 C ...s/rapids-21.06/bin/python 1103MiB |
| 0 N/A N/A 103345 G /usr/lib/firefox/firefox 1MiB |
| 0 N/A N/A 103615 C ...s/rapids-21.06/bin/python 1043MiB |
到底是怎么回事？我忘了安装一些东西吗？Actuakky，CPU 比 GPU 快...
我使用了本教程，一切正常（过程中没有错误）
https://www.youtube.com/watch?v=hfef_ZF7kEo&t=572s
关于为什么不起作用的任何想法？