mrg32k3a

oneMKL - Data Parallel C++ Developer Reference

Download PDF

ID 772045

Date 3/22/2024

Version

Public

A newer version of this document is available. Customers should click here to go to the newest version.

Visible to Intel only — GUID: GUID-CABB31EC-A086-4E02-B3FE-C97E0D364D4E

View Details

mrg32k3a

Description
API

Description

The combined multiple recursive pseudorandom number generator MRG32k3a [L’Ecuyer99a].

API

Syntax

namespace oneapi::mkl::rng {
  class mrg32k3a {
  public:
    static constexpr std::uint32_t default_seed = 1;

    mrg32k3a(sycl::queue queue, std::uint32_t seed = default_seed);
    mrg32k3a(sycl::queue queue, std::uint32_t seed, mrg32k3a_mode::optimal mode);
    mrg32k3a(sycl::queue queue, std::uint32_t seed, mrg32k3a_mode::custom mode);
    mrg32k3a(sycl::queue queue, std::initializer_list<std::uint32_t> seed);
    mrg32k3a(sycl::queue queue, std::initializer_list<std::uint32_t> seed,
             mrg32k3a_mode::optimal mode);
    mrg32k3a(sycl::queue queue, std::initializer_list<std::uint32_t> seed,
             mrg32k3a_mode::custom mode);
    mrg32k3a(const mrg32k3a& other);
    mrg32k3a(mrg32k3a&& other);
    mrg32k3a& operator=(const mrg32k3a& other);
    mrg32k3a& operator=(mrg32k3a&& other);
    ~mrg32k3a();
  };
}

Devices supported: CPU and GPU.

Include Files

oneapi/mkl/rng.hpp

Input Parameters

Name	Type	Description
queue	`sycl::queue`	Valid `sycl::queue`, calls of the `oneapi::mkl::rng::generate()` routine submits kernels in this queue.
seed	`std::uint32_tstd::initializer_list<std::uint32_t>`	Initial conditions of the generator state or engine state.
mode	`mrg32k3a_mode::optimalmrg32k3a_mode::custom`	Initial conditions of the generator state or engine state.

NOTE:

Sub-sequence parallelization [L’Ecuyer02] is supported via the mode parameter. mrg32k3a_mode::optimal uses the optimal number of sub-streams, and mrg32k3a_mode::custom enables the passing a user-defined number of sub-streams (e.g. mrg32k3a_mode::custom my_mode{123}). There is also a helper structure template, mrg32k3a_mode::optimal_v, for optimal mode. Refer to the corresponding example for usage details.
Currently sub-sequence parallelization is supported only for uniform and gaussian distributions on GPU device.

See VS Notes for the detailed descriptions.

Select Your Language

Using Intel.com Search

Quick Links

Recent Searches

Advanced Search

Only search in

oneMKL - Data Parallel C++ Developer Reference

mrg32k3a

Description

API