Parallel 2D Acoustic Finite Difference Seismic Modelling using the Staggered Grid. The original CUDA source code is from here. For SYCL version, the CUDA code was migrated using Intel DPCT, and then the resulting code was modified to remove the dpct headers.

In addition to the input files in input and geometry folders, there are 4 data files that need to be copied manually (if not already done):

The data files can be downloaded from https://github.com/richaras/SeisAcoMod2D/tree/master/data

• Make sure that no folder named 'data' already exists inside the 'SeisAcoMod2D' folder.
• Download the four *.bin files from https://github.com/richaras/SeisAcoMod2D/tree/master/data in a separate location.
• Create a 'data' folder inside 'SeisAcoMod2D' folder (in the same level as 'input' and 'geometry').
• Copy the downloaded *.bin files into the newly created 'data' folder.

ls data

For Intel GPU -
First, source icpx compiler. Then,

cd SeisAcoMod2D/SYCL
mkdir build
cd build
CXX=mpiicpc cmake -DGPU_AOT=pvc ..
make -sj

Note:

• To enable AOT compilation, please use the flag -DGPU_AOT=pvc for PVC.

For AMD GPU -
First source clang++ compiler. Then,

cd SeisAcoMod2D/SYCL
mkdir build
cd build
CXX=mpiicpc cmake -DUSE_AMDHIP_BACKEND=gfx90a ..
make -sj

Note:

• We use the flag -DUSE_AMDHIP_BACKEND=gfx90a for MI250. Use the correct value for your GPU.

For NVIDIA GPU -
First source clang++ compiler. Then,

cd SeisAcoMod2D/SYCL
mkdir build
cd build
CXX=mpiicpc cmake -DUSE_NVIDIA_BACKEND=YES -DUSE_SM=80 ..
make -sj

Note:

• We use the flag -DUSE_SM=80 for A100 or -DUSE_SM=90 for H100.

cd SeisAcoMod2D/CUDA
mkdir build
cd build
CXX=mpiicpc cmake -DUSE_SM=80 ..
make -sj

Note:

• We use the flag -DUSE_SM=80 for A100 or -DUSE_SM=90 for H100.

cd SeisAcoMod2D/HIP
mkdir build
cd build
CXX=mpiicpc cmake -DROCM_PATH=/opt/rocm ..
make -sj

After building, to run the workload, cd into the build folder. Then

For running SYCL on PVC 1T

ONEAPI_DEVICE_SELECTOR=level_zero:gpu ZE_AFFINITY_MASK='0.0' mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

For running SYCL on PVC 2T

ONEAPI_DEVICE_SELECTOR=level_zero:gpu ZE_AFFINITY_MASK='0' EnableImplicitScaling=1 mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

For running SYCL on NVIDIA GPU

ONEAPI_DEVICE_SELECTOR=cuda:gpu mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

For running CUDA on NVIDIA GPU

mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

For running SYCL on AMD GPU

ONEAPI_DEVICE_SELECTOR=hip:gpu mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

For running HIP on AMD GPU

mpiexec -bootstrap ssh -n 2 ./SeisAcoMod2D ../../input/twoLayer_model_5000x5000z_small.json

Output gives the total time (in sec) for running the whole workload.