[FP16][Tensor] Remove unnecessary copy on save

nnstreamer / nntrainer

NNtrainer is Software Framework for Training Neural Network Models on Devices.

Apache License 2.0

134 stars 71 forks source link

[FP16][Tensor] Remove unnecessary copy on save #2657

Closed DonghakPark closed 10 hours ago

DonghakPark commented 2 days ago

There are unnecessary copy of tensor, in case of fp16

It seems that when developing previously, the tensor structure was not accurately established, so it attempted to save by forcibly converting to FP16.

Now, when performing getData<_FP16>(), it is automatically converted, so the process of putting every tensor one by one in the temp array is unnecessary and only slows down the speed.

Self evaluation:

Build test: [X]Passed [ ]Failed [ ]Skipped
Run test: [X]Passed [ ]Failed [ ]Skipped

taos-ci commented 2 days ago

:memo: TAOS-CI Version: 1.5.20200925. Thank you for submitting PR #2657. Please a submit 1commit/1PR (one commit per one PR) policy to get comments quickly from reviewers. Your PR must pass all verificiation processes of cibot before starting a review process from reviewers. If you are new member to join this project, please read manuals in documentation folder and wiki page. In order to monitor a progress status of your PR in more detail, visit http://ci.nnstreamer.ai/.