-
Notifications
You must be signed in to change notification settings - Fork 3.8k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
[gpu] /root/repo/LightGBM/compute/include/boost/compute/utility/wait_list.hpp:166: void boost::compute::wait_list::wait() const: Assertion `clWaitForEvents(size(), get_event_ptr()) == 0' failed. Aborted (core dumped) #2648
Comments
ping @huanzhang12 |
@pseudotensor I cannot reproduce the issue in the following environment with the following versions. Can you please try the latest version of LightGBM? Operating System: Windows Server 2016 Datacenter 1607 (add verbosity and @@ -4,6 +4,7 @@ import numpy as np
import lightgbm as lgb
print(lgb.__version__)
model_orig, X, y = pickle.load(open("lgbm_waitforfailure.pkl", "rb"))
-p = {'boosting_type': 'gbdt', 'class_weight': None, 'colsample_bytree': 0.8, 'importance_type': 'gain', 'learning_rate': 0.25, 'max_depth': 6, 'min_child_samples': 1, 'min_child_weight': 1, 'min_split_gain': 0.0, 'n_estimators': 1800, 'n_jobs': 4, 'num_leaves': 64, 'objective': 'multiclass', 'random_state': None, 'reg_alpha': 0.0, 'reg_lambda': 1.0, 'silent': True, 'subsample': 0.7, 'subsample_for_bin': 200000, 'subsample_freq': 1, 'num_class': 4, 'max_bin': 255, 'scale_pos_weight': 1, 'max_delta_step': 0, 'min_data_in_bin': 1, 'seed': 12345, 'device_type': 'gpu', 'gpu_device_id': 0, 'gpu_platform_id': 0, 'gpu_use_dp': True, 'feature_fraction_seed': 12346, 'bagging_seed': 12347, 'verbose': -1}
+p = {'boosting_type': 'gbdt', 'class_weight': None, 'colsample_bytree': 0.8, 'importance_type': 'gain', 'learning_rate': 0.25, 'max_depth': 6, 'min_child_samples': 1, 'min_child_weight': 1, 'min_split_gain': 0.0, 'n_estimators': 1800, 'n_jobs': 4, 'num_leaves': 64, 'objective': 'multiclass', 'random_state': None, 'reg_alpha': 0.0, 'reg_lambda': 1.0, 'silent': False, 'subsample': 0.7, 'subsample_for_bin': 200000, 'subsample_freq': 1, 'num_class': 4, 'max_bin': 255, 'scale_pos_weight': 1, 'max_delta_step': 0, 'min_data_in_bin': 1, 'seed': 12345, 'device_type': 'gpu', 'gpu_device_id': 0, 'gpu_platform_id': 0, 'gpu_use_dp': True, 'feature_fraction_seed': 12346, 'bagging_seed': 12347, 'verbose': 5}
model = lgb.LGBMClassifier(**p)
model.fit(X, y)
+print(model.predict(X))
|
@StrikerRUS It's reproducible on edb9149 with boost updated to 1.72 on Linux
You can download h2o4gpu the version of lgbm from https://h2o-release.s3.amazonaws.com/h2o4gpu/snapshots/ai/h2o/h2o4gpu/0.3-cuda10/h2o4gpu-0.3.2%2Bpr.818.e8ed15e-cp36-cp36m-linux_x86_64.whl use
to import lgbm |
@sh1ng Is hardware the same as in #2648 (comment)? |
BTW, have you tried to play around with |
Operating System: Ubuntu 18.04.3 LTS CPU/GPU model: Intel(R) Core(TM) i7-8550U CPU @ 1.80GHz/GeForce MX150 C++/Python/R version: Python 3.6.8 the whl was built on centos-7 The same if I install lightgbm from pypi with opencl support
tested on cuda-10.0 and cuda-10.2 |
If I'm not mistaken there's only one OpenCL device
|
removing |
@sh1ng Thank you very much for the cc @huanzhang12 |
Recompiling with defined It means |
Please do NOT remove the gpu_use_dp flag. We are using it in the CUDA version. |
ping @huanzhang12 for the
|
We got the same error using: lightgbm: 3.3.3.99 |
Operating System: Ubuntu 16.04 LTS
CPU/GPU model: Xeon / 1080ti
C++/Python/R version: 3.6.6
LightGBM version or commit hash: 2.2.4
lgbm_waitforfailure.zip
The text was updated successfully, but these errors were encountered: