MCPcopy Create free account
hub / github.com/MhLiao/TextBoxes_plusplus / CaffeMallocHost

Function CaffeMallocHost

include/caffe/syncedmem.hpp:15–26  ·  view source on GitHub ↗

If CUDA is available and in GPU mode, host memory will be allocated pinned, using cudaMallocHost. It avoids dynamic pinning for transfers (DMA). The improvement in performance seems negligible in the single GPU case, but might be more significant for parallel training. Most importantly, it improved stability for large models on many GPUs.

Source from the content-addressed store, hash-verified

source not stored for this graph (policy: none)

Callers 1

to_cpuMethod · 0.85

Calls

no outgoing calls

Tested by

no test coverage detected