Cherk3mEx

cuBLAS Extension APIs - cublasCherk3mEx

Description

This code demonstrates a usage of cuBLAS Cherk3mEx function to perform a Hermitian rank-k update

A = | 1.1 + 1.2j | 2.3 + 2.4j |
    | 3.5 + 3.6j | 4.7 + 4.8j |

C = | 1.1 + 1.2j | 2.3 + 2.4j |
    | 3.5 + 3.6j | 4.7 + 4.8j |

This function is an extension of cublasCherk where the input matrix and output matrix can have a lower precision but the computation is still done in the type cuComplex. This routine is implemented using the Gauss complexity reduction algorithm which can lead to an increase in performance up to 25%

See documentation for further details.

Supported SM Architectures

All GPUs supported by CUDA Toolkit (https://developer.nvidia.com/cuda-gpus)

Supported OSes

Linux
Windows

Supported CPU Architecture

x86_64
ppc64le
arm64-sbsa

CUDA APIs involved

• cublasCherk3mEx API

Building (make)

Prerequisites

• A Linux/Windows system with recent NVIDIA drivers.
• CMake version 3.18 minimum

Build command on Linux

$ mkdir build
$ cd build
$ cmake ..
$ make

Make sure that CMake finds expected CUDA Toolkit. If that is not the case you can add argument -DCMAKE_CUDA_COMPILER=/path/to/cuda/bin/nvcc to cmake command.

Build command on Windows

$ mkdir build
$ cd build
$ cmake -DCMAKE_GENERATOR_PLATFORM=x64 ..
$ Open cublas_examples.sln project in Visual Studio and build

Usage

$  ./cublas_Cherk3mEx_example

Sample example output:

A
1.10 + 1.20j 2.30 + 2.40j 
3.50 + 3.60j 4.70 + 4.80j 
=====
C
13.70 + 0.00j 30.50 + 0.48j 
0.00 + 0.00j 70.34 + 0.00j 
=====