apache / mxnet UNCLAIMED

Lightweight, Portable, Flexible Distributed/Mobile Deep Learning with Dynamic, Mutation-aware Dataflow Dep Scheduler; for Python, R, Julia, Scala, Go, Javascript and more

20813 0 0 C++

Introduce the ENABLE_CUDA_RTC build option (#9428)

When OFF the ENABLE_CUDA_RTC option skips building the CUDA runtime
compilation feature (CudaModule) that requires access to the CUDA
driver API. The MXNet shared library build with this option OFF has no
direct dependency on the CUDA driver library (e.g. libcuda.so) and can
be used on both GPU and CPU hosts. CPU-only hosts require no stub
CUDA driver library.

Can Balioglu committed 8y ago

262c74c2c8228bfc60e0673bdf5ead0bf2e1973d

Parent: e36a945

Committed by Chris Olivier <cjolivier01@gmail.com> on 1/15/2018, 8:21:23 PM