Migrating the Quasirandom Generator from CUDA* to SYCL*

Overview  

The QuasirandomGenerator sample implements Niederreiter Quasirandom Sequence Generator and Inverse Cumulative Normal Distribution functions for the generation of Standard Normal Distributions. The original CUDA* source code is migrated to SYCL for portability across GPUs from multiple vendors.

Key Implementation Details 

The code defines a kernel function quasirandomGeneratorKernel() which generates random numbers for a given seed and number of samples N. The kernel is launched using a parallel-for loop and is executed on a GPU device. The initTableGPU() function initializes a constant memory table used by the quasirandomGeneratorKernel() function. Finally, the code includes an implementation of Moro's inverse cumulative normal distribution function approximation, which is used to convert the random numbers generated by the quasirandomGeneratorKernel() function into normally distributed random numbers.

Original CUDA source files: quasirandomGenerator.

Migrated SYCL source files including step-by-step instructions: guided_quasirandomGenerator_SYCLMigration.

References 

• Data Parallel C++, by James Reinders et al 
• oneAPI GPU Optimization Guide 
• CUDA Toolkit documentation
• Install oneAPI for NVIDIA GPUs