Revision 40af7bb68fa3ca245676b81938568aedefcb5ece authored by Bowen Bao on 18 September 2018, 01:10:34 UTC, committed by Bowen Bao on 18 September 2018, 01:10:34 UTC
* CPU.
- Short-circuit the call to ComputeConvolveGeometryExplicit() when
MKL is enabled.
- Replace div/mod inside ComputeConvolveGeometryExplicit with
fast_divmod.
* GPU.
- Instead of creating a new CuDnnConvolutionEngine for every batch,
just update the geometry related info, and try to reuse the
workspacememory from previous run.
1 parent 1f28c7d
| File | Mode | Size |
|---|---|---|
| Manual_How_to_create_user_minibatch_sources.ipynb | -rw-r--r-- | 24.8 KB |
| Manual_How_to_debug.ipynb | -rw-r--r-- | 159.7 KB |
| Manual_How_to_feed_data.ipynb | -rw-r--r-- | 51.9 KB |
| Manual_How_to_train_using_declarative_and_imperative_API.ipynb | -rw-r--r-- | 28.6 KB |
| Manual_How_to_use_learners.ipynb | -rw-r--r-- | 41.7 KB |
| Manual_How_to_use_network_optimizations.ipynb | -rw-r--r-- | 16.6 KB |
| Manual_How_to_write_a_custom_deserializer.ipynb | -rw-r--r-- | 19.2 KB |
Computing file changes ...
