[test] Add MCH alltoallv check

config/cscs.py

@@ -129,7 +129,7 @@ class ReframeSettings:
                                      'PrgEnv-gnu', 'PrgEnv-gnu-nompi'],
                         'descr': 'Kesch compute nodes',
                         'resources': {
-                            '_rfm_gpu': ['--gres=gpu:{num_gpus_per_node}']
+                            '_rfm_gpu': ['--gres=gpu:{num_gpus_per_node}'],
                         }
                     }
                 }

cscs-checks/mch/alltoallv.py

@@ -0,0 +1,86 @@
+import reframe as rfm
+import reframe.utility.sanity as sn
+
+
+@rfm.parameterized_test(['default'], ['nocomm'], ['nocomp'])
+class Alltoallv(rfm.RegressionTest):
+    def __init__(self, variant):
+        super().__init__()
+        self.valid_systems = ['dom:gpu', 'daint:gpu', 'kesch:cn']
+        self.valid_prog_environs = ['PrgEnv-gnu']
+        self.variables = {'G2G': '1'}
+        self.executable = 'build/src/comm_overlap_benchmark'
+        if variant != 'default':
+            self.executable_opts = ['--' + variant]
+
+        self.sourcesdir = 'https://github.com/cosunae/comm_overlap_bench'
+        # Checkout to the branch alltoallv
+        self.prebuild_cmd = ['git checkout alltoallv']
+        self.build_system = 'CMake'
+        self.build_system.builddir = 'build'
+        self.build_system.config_opts = ['-DCMAKE_BUILD_TYPE=Release',
+                                         '-DENABLE_MPI_TIMER=ON']
+
+        if self.current_system.name == 'kesch':
+            self.num_tasks = 144
+            self.num_gpus_per_node = 16
+            self.num_tasks_per_node = 16
+            self.num_tasks_per_socket = 8
+            self.modules = ['craype-accel-nvidia35', 'cmake']
+            self.variables['MV2_USE_CUDA'] = '1'
+            self.build_system.config_opts += [
+                '-DMPI_VENDOR=mvapich2',
+                '-DCUDA_COMPUTE_CAPABILITY="sm_37"'
+            ]
+            self.build_system.max_concurrency = 1
+        else:
+            self.num_tasks = 4
+            self.num_gpus_per_node = 1
+            self.num_tasks_per_node = 1
+            self.modules = ['craype-accel-nvidia60', 'CMake']
+            self.variables['MPICH_RDMA_ENABLED_CUDA'] = '1'
+            self.build_system.config_opts += [
+                '-DCUDA_COMPUTE_CAPABILITY="sm_60"'
+            ]
+            self.build_system.max_concurrency = 8
+
+        self.sanity_patterns = sn.assert_found(r'ELAPSED TIME:', self.stdout)
+        self.perf_patterns = {
+            'elapsed_time': sn.extractsingle(r'ELAPSED TIME:\s+(\S+)',
+                                             self.stdout, 1, float, 1)
+        }
+        ref_values = {
+            'kesch': {
+                'nocomm':  5.7878,
+                'nocomp':  5.62155,
+                'default': 5.53777
+            },
+            'daint': {
+                'nocomm':  0.0171947,
+                'nocomp':  0.0137893,
+                'default': 0.0138493
+            }
+        }
+        if self.current_system.name == 'dom':
+            sysname = 'daint'
+        else:
+            sysname = self.current_system.name
+
+        ref = ref_values[sysname][variant]
+        self.reference = {
+            'kesch:cn': {
+                'elapsed_time': (ref, None, 0.15)
+            },
+            '*': {
+                'elapsed_time': (ref, None, 0.15)
+            }
+        }
+
+        self.maintainers = ['AJ', 'VK']
+        self.tags = {'production'}
+
+    def setup(self, *args, **kwargs):
+        super().setup(*args, **kwargs)
+        if self.current_system.name == 'kesch':
+            self.job.launcher.options = ['--distribution=block:block',
+                                         '--cpu_bind=q']