The accelerator interface (ACC) consists of ISO_C_BINDING based Fortran code of DBCSR's ACC-backend interface and LIBSMM/ACC-interface. The interface is implemented by CUDA (for Nvidia GPUs), the HIP (for AMD GPUs), and the OpenCL accelerator backends.
The code for both the CUDA and the HIP backend is unified, and can be found in the cuda directory. At compile-time either one or the other backend is chosen per macro (__CUDA or __HIP). Similarly, the code for the OpenCL backend is activated by a build-time macro (__OPENCL).
There are two stand-alone sample codes or drivers exercising the ACC-interface. The driver code (only depending on above mentioned interfaces) can be built locally and in a rather self-contained fashion, i.e., no DBCSR library is needed (except runtime libraries such as CUDA, HIP, OpenCL). For OpenCL, the LIBXSMM library is mandatory and preferred as baseline and for validation in any case. To build LIBXSMM, a folder libxsmm in parallel to DBCSR's root directory (dbcsr) is expected to be present and prebuilt.
git clone -b main https://github.com/libxsmm/libxsmm.git
cd libxsmm
make GNU=1 -j
To build the driver code (opencl in below example), change into the respective backend folder (cuda or opencl), and invoke make (DBG=0|1|2 is supported among other optional key-value pairs).
git clone https://github.com/cp2k/dbcsr.git
cd dbcsr/src/acc/opencl
make
NOTE: To activate a certain device, the drivers consider an environment variable called DEVICE. For example, DEVICE=1 ./acc_bench_trans activates the second device (at least two devices must be discovered). This environment variable is implemented by the driver code and meant to work across backends, i.e., the OpenCL backend also supports ACC_OPENCL_DEVICE=1 (see Developer Guide for the OpenCL backend).
The drivers support command line options (nrepeat, stack_size, m, n, ...). Command line arguments are positional but allow 0 as placeholder to refer to the default value (acc_bench_smm 0 0 5 13 5 performs the default number of repetitions with the default stacksize when running the 5x13x5-kernel). For example, running the tranpose benchmark may look like:
$ OMP_PROC_BIND=TRUE ./acc_bench_trans 5 30000 23 23
./acc_bench_trans 5 30000 23 23
typename (id=3): double
copy-in: 17.2 ms 7.2 GB/s
device: 8.7 ms 14.2 GB/s
host: 8.4 ms 14.6 GB/s
errors: 0
For timing, comparison (host code), and validation, LIBXSMM is required. The drivers exercise the respective backend. For example with the CUDA backend:
cd src/acc/cuda
make WITH_GPU=P100
../acc_bench_smm
For the OpenCL backend:
cd src/acc/opencl
make
../acc_bench_smm
In above cases, acc_bench_trans and acc_bench_smm are built using the respective backend. Both driver codes can be built for double-precision (default) or single-precision using a build-time macro (make ELEM_TYPE=float or -DELEM_TYPE=float in general).