projects / platform / core / ml / nnfw.git / commit
? search:
summary | shortlog | log | commit | commitdiff | tree
(parent: e6caf83) | patch
[onert] Use CachedData instead of MMapedData (#4007)
authorJiHwan Yeo <jihwan.yeo@samsung.com>
Fri, 28 Aug 2020 04:19:24 +0000 (13:19 +0900)
committerGitHub <noreply@github.com>
Fri, 28 Aug 2020 04:19:24 +0000 (13:19 +0900)
commitba00ad0df87cb67cec79ea2f4fc27ce9081c6629
tree12f73a24559dc8d85fae2e6745011a026ef1952atree | snapshot
parente6caf839d4406973c7ecc4a862d79328872bde65commit | diff
[onert] Use CachedData instead of MMapedData (#4007)

- MMapedData reduces load/prepare time, but increases inference time
- Use CachedData for weight tensor for inference performance
  - Includes Revert "[onert] Remove unused function for munmap (#3710)"

ONE-DCO-1.0-Signed-off-by: JiHwan Yeo <jihwan.yeo@samsung.com>
runtime/onert/frontend/base_loader/include/base_loader.h diff | blob | history
Domain: Machine Learning / ML Framework;