Hendrik/xgboost
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

[GPU Plugin] Fast histogram speed improvements. Updated benchmarks. (#2258)

Browse Source
This commit is contained in:
Rory Mitchell
2017-05-09 04:21:38 +12:00
committed by Tianqi Chen
parent 98ea461532
commit 6bf968efe6
12 changed files with 433 additions and 241 deletions
Download Patch File Download Diff File Expand all files Collapse all files

62
plugin/updater_gpu/benchmark/benchmark.py
View File

@@ -1,32 +1,50 @@
#pylint: skip-file
# pylint: skip-file
import sys, argparse
import xgboost as xgb
import numpy as np
from sklearn.datasets import make_classification
from sklearn.model_selection import train_test_split
import time
n = 1000000
num_rounds = 100
num_rounds = 500
X,y = make_classification(n, n_features=50, random_state=7)
X_train, X_test, y_train, y_test = train_test_split(X, y, random_state=0)
dtrain = xgb.DMatrix(X_train, y_train)
dtest = xgb.DMatrix(X_test, y_test)
def run_benchmark(args, gpu_algorithm, cpu_algorithm):
print("Generating dataset: {} rows * {} columns".format(args.rows,args.columns))
X, y = make_classification(args.rows, n_features=args.columns, random_state=7)
dtrain = xgb.DMatrix(X, y)
param = {'objective': 'binary:logistic',
'tree_method': 'exact',
'updater': 'grow_gpu_hist',
'max_depth': 8,
'silent': 1,
'eval_metric': 'auc'}
res = {}
tmp = time.time()
xgb.train(param, dtrain, num_rounds, [(dtrain, 'train'), (dtest, 'test')],
evals_result=res)
print ("GPU: %s seconds" % (str(time.time() - tmp)))
param = {'objective': 'binary:logistic',
'tree_method': 'exact',
'max_depth': 6,
'silent': 1,
'eval_metric': 'auc'}
tmp = time.time()
param['updater'] = 'grow_fast_histmaker'
xgb.train(param, dtrain, num_rounds, [(dtrain, 'train'), (dtest, 'test')], evals_result=res)
print ("CPU: %s seconds" % (str(time.time() - tmp)))
param['updater'] = gpu_algorithm
print("Training with '%s'" % param['updater'])
tmp = time.time()
xgb.train(param, dtrain, args.iterations)
print ("Time: %s seconds" % (str(time.time() - tmp)))
param['updater'] = cpu_algorithm
print("Training with '%s'" % param['updater'])
tmp = time.time()
xgb.train(param, dtrain, args.iterations)
print ("Time: %s seconds" % (str(time.time() - tmp)))
parser = argparse.ArgumentParser()
parser.add_argument('--algorithm', choices=['all', 'grow_gpu', 'grow_gpu_hist'], required=True)
parser.add_argument('--rows',type=int,default=1000000)
parser.add_argument('--columns',type=int,default=50)
parser.add_argument('--iterations',type=int,default=500)
args = parser.parse_args()
if 'grow_gpu_hist' in args.algorithm:
run_benchmark(args, args.algorithm, 'grow_fast_histmaker')
if 'grow_gpu ' in args.algorithm:
run_benchmark(args, args.algorithm, 'grow_colmaker')
if 'all' in args.algorithm:
run_benchmark(args, 'grow_gpu', 'grow_colmaker')
run_benchmark(args, 'grow_gpu_hist', 'grow_fast_histmaker')