diff --git a/02-MSA-hello-world-gpu/Makefile b/02-MSA-hello-world-gpu/Makefile
new file mode 100644
index 0000000000000000000000000000000000000000..c51d61d957d0decf1cb391bed05c7a34ddaac7e3
--- /dev/null
+++ b/02-MSA-hello-world-gpu/Makefile
@@ -0,0 +1,28 @@
+CUCOMP = nvcc
+CUFLAGS = -arch=sm_80
+
+ifdef EBROOTOPENMPI
+MPI_HOME+=$(EBROOTOPENMPI)
+endif
+ifdef EBROOTPSMPI
+MPI_HOME+=$(EBROOTPSMPI)
+endif
+
+INCLUDES = -I$(MPI_HOME)/include
+LIBRARIES = -L$(MPI_HOME)/lib -lmpi
+
+all: hello.cpu.out hello.gpu.out
+
+hello.cpu.out: hello-world.c
+ mpicc $< -o $@
+
+hello.gpu.out: hello.gpu.o
+ $(CUCOMP) $(CUFLAGS) $(LIBRARIES) $< -o $@
+
+hello.gpu.o: hello-world.cu
+ $(CUCOMP) $(CUFLAGS) $(INCLUDES) -c $< -o $@
+
+.PHONY: clean
+
+clean:
+ rm -f hello.cpu.out hello.gpu.out *.o
\ No newline at end of file
diff --git a/02-MSA-hello-world-gpu/hello-world.c b/02-MSA-hello-world-gpu/hello-world.c
new file mode 100644
index 0000000000000000000000000000000000000000..9d501bc9ff1e73e01ef888629f768d6d778af464
--- /dev/null
+++ b/02-MSA-hello-world-gpu/hello-world.c
@@ -0,0 +1,34 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <mpi.h>
+
+int main(int argc, char** argv){
+ MPI_Init(&argc, &argv);
+
+ int size;
+ MPI_Comm_size(MPI_COMM_WORLD, &size);
+
+ int rank;
+ MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+
+ MPI_Status stat;
+
+ if(size != 2){
+ if(rank == 0){
+ printf("This program requires exactly 2 MPI ranks, but you are attempting to use %d! Exiting...\n", size);
+ }
+ MPI_Finalize();
+ exit(0);
+ }
+ int tag = 10;
+
+ const char *payload = "world!";
+
+ if (rank == 0) {
+ MPI_Send(payload, 6, MPI_CHAR, 1, tag, MPI_COMM_WORLD);
+ }
+
+ printf("\n");
+
+ return 0;
+}
diff --git a/02-MSA-hello-world-gpu/hello-world.cu b/02-MSA-hello-world-gpu/hello-world.cu
new file mode 100644
index 0000000000000000000000000000000000000000..da1b592e0c32c8165dc99c8d102bc4b3767aa902
--- /dev/null
+++ b/02-MSA-hello-world-gpu/hello-world.cu
@@ -0,0 +1,66 @@
+#include <stdio.h>
+#include <stdlib.h>
+#include <mpi.h>
+
+#define CUDA_RT_CALL(call) \
+ { \
+ cudaError_t cudaStatus = call; \
+ if (cudaSuccess != cudaStatus) \
+ fprintf(stderr, \
+ "ERROR: CUDA RT call \"%s\" in line %d of file %s failed " \
+ "with " \
+ "%s (%d).\n", \
+ #call, __LINE__, __FILE__, cudaGetErrorString(cudaStatus), cudaStatus); \
+ }
+__global__ void hello(const char * payload){
+ int i = blockDim.x * blockIdx.x + threadIdx.x;
+
+ if (i == 0) {
+ printf("%s", payload);
+ }
+}
+
+int main(int argc, char** argv){
+ MPI_Init(&argc, &argv);
+
+ int size;
+ MPI_Comm_size(MPI_COMM_WORLD, &size);
+
+ int rank;
+ MPI_Comm_rank(MPI_COMM_WORLD, &rank);
+
+ MPI_Status stat;
+
+ if(size != 2){
+ if(rank == 0){
+ printf("This program requires exactly 2 MPI ranks, but you are attempting to use %d! Exiting...\n", size);
+ }
+ MPI_Finalize();
+ exit(0);
+ }
+ int tag = 10;
+
+ const char *payload = "hello ";
+
+ char * d_payload;
+ CUDA_RT_CALL( cudaMalloc((void**)&d_payload, 6) );
+ CUDA_RT_CALL( cudaMemcpy(d_payload, payload, 6, cudaMemcpyHostToDevice) );
+
+ hello<<<1, 1>>>(d_payload);
+
+ CUDA_RT_CALL( cudaPeekAtLastError() );
+ CUDA_RT_CALL( cudaDeviceSynchronize() );
+
+ if (rank == 1) {
+ MPI_Recv(d_payload, 6, MPI_CHAR, 0, tag, MPI_COMM_WORLD, &stat);
+ }
+
+ hello<<<1, 1>>>(d_payload);
+
+ CUDA_RT_CALL( cudaPeekAtLastError() );
+ CUDA_RT_CALL( cudaDeviceSynchronize() );
+
+ printf("\n");
+
+ return 0;
+}
diff --git a/02-MSA-hello-world-gpu/job_msa_juwels.sh b/02-MSA-hello-world-gpu/job_msa_juwels.sh
new file mode 100644
index 0000000000000000000000000000000000000000..32bd3696ee2e173aaa8207ee2d0c3466e2ffafc6
--- /dev/null
+++ b/02-MSA-hello-world-gpu/job_msa_juwels.sh
@@ -0,0 +1,14 @@
+#!/bin/bash -x
+#SBATCH --account=training2317
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1
+#SBATCH --output=slurm-out.%j
+#SBATCH --error=slurm-err.%j
+#SBATCH --time=00:15:00
+#SBATCH --partition=devel
+#SBATCH hetjob
+#SBATCH --nodes=1
+#SBATCH --ntasks-per-node=1
+#SBATCH --partition=develbooster
+
+srun xenv -P -L GCC -L ParaStationMPI ./hello.cpu.out : xenv -P -L GCC -L ParaStationMPI -L MPI-settings/CUDA ./hello.gpu.out
diff --git a/02-MSA-ping-pong/Makefile b/03-MSA-ping-pong/Makefile
similarity index 100%
rename from 02-MSA-ping-pong/Makefile
rename to 03-MSA-ping-pong/Makefile
diff --git a/02-MSA-ping-pong/README.md b/03-MSA-ping-pong/README.md
similarity index 100%
rename from 02-MSA-ping-pong/README.md
rename to 03-MSA-ping-pong/README.md
diff --git a/02-MSA-ping-pong/compile.sh b/03-MSA-ping-pong/compile.sh
similarity index 100%
rename from 02-MSA-ping-pong/compile.sh
rename to 03-MSA-ping-pong/compile.sh
diff --git a/02-MSA-ping-pong/job_msa_jureca.sh b/03-MSA-ping-pong/job_msa_jureca.sh
similarity index 100%
rename from 02-MSA-ping-pong/job_msa_jureca.sh
rename to 03-MSA-ping-pong/job_msa_jureca.sh
diff --git a/02-MSA-ping-pong/job_msa_juwels.sh b/03-MSA-ping-pong/job_msa_juwels.sh
similarity index 100%
rename from 02-MSA-ping-pong/job_msa_juwels.sh
rename to 03-MSA-ping-pong/job_msa_juwels.sh
diff --git a/02-MSA-ping-pong/ping-pong.c b/03-MSA-ping-pong/ping-pong.c
similarity index 100%
rename from 02-MSA-ping-pong/ping-pong.c
rename to 03-MSA-ping-pong/ping-pong.c
diff --git a/02-MSA-ping-pong/ping-pong.cu b/03-MSA-ping-pong/ping-pong.cu
similarity index 100%
rename from 02-MSA-ping-pong/ping-pong.cu
rename to 03-MSA-ping-pong/ping-pong.cu