[GPU/OpenCL] Handling pre-compiled and existing kernels

nnstreamer / nntrainer

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

Apache License 2.0

135 stars 71 forks source link

[GPU/OpenCL] Handling pre-compiled and existing kernels #2535

Closed s-debadri closed 2 months ago

s-debadri commented 3 months ago

Handling existing kernels:

Modified clCreateKernel to read pre-compiled kernel binaries.
Handling already initialized kernels using global bitmask.
Added enumerator for kernel names and a method to resolve as string.

Signed-off-by: Debadri Samaddar s.debadri@samsung.com

taos-ci commented 3 months ago

:memo: TAOS-CI Version: 1.5.20200925. Thank you for submitting PR #2535. 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/.

myungjoo commented 3 months ago

LGTM.

Later, kernel path policy should be aware of security and privileges of Android and Tizen devices. In such devices, it should be able to accept different paths per app or the caller.

However, as long as the kernel path is configurable, it can wait until opencl support becomes mature.