Soon, I will add test programs to check the mappings of processes and…

… threads to the hardware. Here's the README for now.

hello/README.md

@@ -0,0 +1,113 @@
+
+### Report the mapping of workers to the hardware
+
+These programs report the mapping of CPUs and GPUs for each process
+and thread. There are three variants:
+
+* MPI: `mpi`
+* OpenMP: `omp`
+* MPI+OpenMP: `mpi+omp`
+
+#### Running
+
+Usage is straightforward. Use the `-v` option for verbose GPU output and
+the `-h` option for help.
+
+```
+$ srun -n4 ./mpi
+node173  Task   0/  4 runing on 4 CPUs: 0,3,6,9
+         Task   0/  4 has 2 GPUs: 0x63 0x43 
+node173  Task   1/  4 runing on 4 CPUs: 12,15,18,21
+         Task   1/  4 has 2 GPUs: 0x3 0x27 
+node173  Task   2/  4 runing on 4 CPUs: 24,27,30,33
+         Task   2/  4 has 2 GPUs: 0xe3 0xc3 
+node173  Task   3/  4 runing on 4 CPUs: 36,39,42,45
+         Task   3/  4 has 2 GPUs: 0x83 0xa3 
+```
+
+```
+$ srun -n4 ./mpi -v
+node173  Task   0/  4 runing on 4 CPUs: 0,3,6,9
+         Task   0/  4 has 2 GPUs: 0x63 0x43 
+	Default device: 0x63
+	0x63: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x43: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+node173  Task   1/  4 runing on 4 CPUs: 12,15,18,21
+         Task   1/  4 has 2 GPUs: 0x3 0x27 
+	Default device: 0x3
+	0x03: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x27: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+node173  Task   2/  4 runing on 4 CPUs: 24,27,30,33
+         Task   2/  4 has 2 GPUs: 0xe3 0xc3 
+	Default device: 0xe3
+	0xe3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0xc3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+node173  Task   3/  4 runing on 4 CPUs: 36,39,42,45
+         Task   3/  4 has 2 GPUs: 0x83 0xa3 
+	Default device: 0x83
+	0x83: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0xa3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+```
+
+```
+$ OMP_NUM_THREADS=4 srun -n2 ./omp
+Process runing on 1 CPUs: 0
+Process has 4 GPUs: 0x63 0x43 0x3 0x27 
+	Default device: 0x63
+	0x63: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x43: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x03: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x27: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+Thread   0/  4 running on 1 CPUs: 0
+Thread   0/  4 assigned to GPU: 0x63
+Thread   1/  4 running on 1 CPUs: 6
+Thread   1/  4 assigned to GPU: 0x43
+Thread   2/  4 running on 1 CPUs: 12
+Thread   2/  4 assigned to GPU: 0x3
+Thread   3/  4 running on 1 CPUs: 18
+Thread   3/  4 assigned to GPU: 0x27
+
+Process runing on 1 CPUs: 24
+Process has 4 GPUs: 0xe3 0xc3 0x83 0xa3 
+	Default device: 0xe3
+	0xe3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0xc3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0x83: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+	0xa3: Vega 20, 31 GB Mem, 60 Multiprocessors, 1.725 GHZ, 9.6 CC
+Thread   0/  4 running on 1 CPUs: 24
+Thread   0/  4 assigned to GPU: 0xe3
+Thread   1/  4 running on 1 CPUs: 30
+Thread   1/  4 assigned to GPU: 0xc3
+Thread   2/  4 running on 1 CPUs: 36
+Thread   2/  4 assigned to GPU: 0x83
+Thread   3/  4 running on 1 CPUs: 42
+Thread   3/  4 assigned to GPU: 0xa3
+```
+
+#### Building
+
+These program are built with a single Makefile. By default typing
+`make` will only build the CPU-related programs. To enable GPU
+information, the user needs to set an environment variable.
+
+```
+# Build with CPU support
+$ make
+
+# Build with support for AMD GPUs
+$ HAVE_AMD_GPUS=1 make 
+
+# Build with support for NVIDIA GPUs
+$ HAVE_NVIDIA_GPUS=1 make 
+```
+
+To build with AMD GPU support, the ROCm environment must be
+present. Similarly, for NVIDIA support, the CUDA environment must be
+present. 
+
+
+
+
+
+
+