Andrew V. Adinetz 24c2e41287 Fixed the bug with illegal memory access in test_large_sizes.py with 4 GPUs. (#3068) ...

- thrust::copy() called from dvec::copy() for gpairs invoked a GPU kernel instead of
  cudaMemcpy()
- this resulted in illegal memory access if the GPU running the kernel could not access
  the data being copied
- new version of dvec::copy() for thrust::device_ptr iterators calls cudaMemcpy(),
  avoiding the problem.

2018-02-01 16:54:46 +13:00

..

c_api

SHAP values for feature contributions (#2438)

2017-10-12 12:35:51 -07:00

common

Fixed the bug with illegal memory access in test_large_sizes.py with 4 GPUs. (#3068)

2018-02-01 16:54:46 +13:00

data

Sync number of features after loaded matrix in different workers. (#2722)

2017-11-29 11:19:12 -08:00

gbm

Objective function evaluation on GPU with minimal PCIe transfers (#2935)

2018-01-12 21:33:39 +13:00

metric

Improve multi-threaded performance (#2104)

2017-03-25 10:35:01 -07:00

objective

Fix GPU bugs (#3051)

2018-01-23 13:14:15 +13:00

predictor

Objective function evaluation on GPU with minimal PCIe transfers (#2935)

2018-01-12 21:33:39 +13:00

tree

Fix GPU bugs (#3051)

2018-01-23 13:14:15 +13:00

cli_main.cc

fix bug in loading config for pred task (#2795)

2017-10-20 00:10:14 -05:00

learner.cc

Objective function evaluation on GPU with minimal PCIe transfers (#2935)

2018-01-12 21:33:39 +13:00

logging.cc

Have ConsoleLogger log to stderr instead of stdout (#1714)

2016-11-10 12:39:52 -08:00