You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Qualcomm AI Engine Direct - Support Platform Abstraction Layer (PAL) (pytorch#20050)
### Summary
This PR introduces **Platform Abstraction Layer (PAL)** support to
enable consistent, platform-independent behavior, including Linux and
Windows.
### Scope
The goal of this change is to encapsulate platform-specific logic behind
a unified interface, improving portability, maintainability, and
extensibility of the codebase.
### Supported Platforms
- Linux
- Windows
### Currently Supported Functionality
The initial PAL implementation focuses on the following capabilities:
- **Dynamic loading**
Provides a unified interface for loading shared libraries at runtime
across supported platforms.
- **Shared library name construction**
Abstracts platform-specific shared library naming conventions (e.g.,
.so, .dll) to ensure correct library resolution on each platform.
### Future Work
- Additional PAL features if required.
---------
Co-authored-by: chenweng <chenweng@qti.qualcomm.com>
0 commit comments