2 - Support simultaneous request handling from different user apps
3 - Support timeout and priority handling for multiple requests
4 - Add and verify prebuilt tvn models (TRIV v1) in FastModel
5 - Sync with the NPU driver version in NPU_Linux-kernel
8 - Verify test cases on FastModel
12 - Support new NPU binary format (i.e., version 2)
13 - Support non-contiguous multiple input/output tensors
14 - Support data manipulation and quantization (only for npubinfmt v2)
17 - Add a prototype for NPU-ASR
21 - Change names of sub-packages from libnpuengine to npu-engine
24 - Tarballs containing test data are removed
25 - Test data are generated at the build time
28 - NPU Vision functionality verified in the FastModel envionment
29 - Support Ubuntu 16.04 packaging & emulation of NPU executions
30 - More test cases with single/multi-layer NPU models