This code demonstrates a usage of cuBLAS GemmBatchedEx function to compute batches of matrix-matrix products
A = | 1.0 | 2.0 | 5.0 | 6.0 |
| 3.0 | 4.0 | 7.0 | 8.0 |
B = | 5.0 | 6.0 | 9.0 | 10.0 |
| 7.0 | 8.0 | 11.0 | 12.0 |
This function is an extension of cublas\<t>gemmBatched that performs the matrix-matrix multiplication of a batch of matrices and allows the user to individually specify the data types for each of the A, B and C matrix arrays, the precision of computation and the GEMM algorithm to be run. Like cublas\<t>gemmBatched, the batch is considered to be "uniform", i.e. all instances have the same dimensions (m, n, k), leading dimensions (lda, ldb, ldc) and transpositions (transa, transb) for their respective A, B and C matrices. The address of the input matrices and the output matrix of each instance of the batch are read from arrays of pointers passed to the function by the caller. Supported combinations of arguments are listed further down in this section.
See documentation for further details.
All GPUs supported by CUDA Toolkit (https://developer.nvidia.com/cuda-gpus)
Linux
Windows
x86_64
ppc64le
arm64-sbsa
- A Linux/Windows system with recent NVIDIA drivers.
- CMake version 3.18 minimum
$ 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.
$ mkdir build
$ cd build
$ cmake -DCMAKE_GENERATOR_PLATFORM=x64 ..
$ Open cublas_examples.sln project in Visual Studio and build
$ ./cublas_GemmBatchedEx_example
Sample example output:
A[0]
1 2
3 4
=====
A[1]
5 6
7 8
=====
B[0]
5 6
7 8
=====
B[1]
9 10
11 12
=====
C[0]
19 43
22 50
=====
C[1]
111 151
122 166
=====