Implementations of https://github.com/onnx/onnx/blob/main/docs/Operators.md#DynamicQuantizeLinear using C++, OpenMP, and CUDA
To compile simply make in the directory.
| Filename | Description | Executable name |
|---|---|---|
| dynamic_quantize_linear_basic.cpp | base unoptimized C++ implementation, uses std::minmax_elements | dqlbase |
| dynamic_quantize_linear_mod.cpp | manual implementation of minmax_element. Implemented as a void function by passing in output vector as a reference | dqlmod |
| dynamic_quantize_linear_openmp.cpp | Optimized solution with OpenMP | dqlomp |
| dynamic_quantize_linear_omp_mod.cpp | void function implementation with OpenMP | dqlompmod |
| dynamic_quantize_linear_cuda.cu | CUDA implementation using thrust library for minmax_element | dqlcuda |