Merge pull request #795 from rfbgo/hp2p

Add hp2p app definition

var/ramble/repos/builtin/applications/hp2p/application.py

@@ -0,0 +1,86 @@
+# Copyright 2022-2024 The Ramble Authors
+#
+# Licensed under the Apache License, Version 2.0 <LICENSE-APACHE or
+# https://www.apache.org/licenses/LICENSE-2.0> or the MIT license
+# <LICENSE-MIT or https://opensource.org/licenses/MIT>, at your
+# option. This file may not be copied, modified, or distributed
+# except according to those terms.
+
+from ramble.appkit import *
+
+
+class Hp2p(ExecutableApplication):
+    """
+    HP2P (Heavy Peer To Peer) benchmark is a test which performs MPI Point-to-Point non-blocking communications between all MPI processes. Its goal is to measure the bandwidths and the latencies in a situation where the network is busy. This benchmark can help to detect network problems like congestions or problems with switches or links.
+    """
+
+    name = "hp2p"
+
+    tags("benchmark", "mpi")
+    maintainers("rfbgo")
+
+    software_spec("hp2p", pkg_spec="[email protected]", package_manager="spack*")
+    software_spec(
+        "openmpi412", pkg_spec="[email protected]", package_manager="spack*"
+    )
+
+    executable(
+        "execute",
+        "hp2p.exe -n {iter} -s {msg_size} -m {nb_msg} -b 1",
+        use_mpi=True,
+    )
+
+    workload("standard", executables=["execute"])
+
+    workload_variable(
+        "iter",
+        default="1000",
+        description="Number of iterations",
+        workloads=["standard"],
+    )
+
+    workload_variable(
+        "msg_size",
+        default="1024",
+        description="Message size",
+        workloads=["standard"],
+    )
+
+    workload_variable(
+        "nb_msg",
+        default="10",
+        description="Number of msg per comm",
+        workloads=["standard"],
+    )
+
+    success_criteria(
+        "prints_done", mode="string", match=r".*=== SUMMARY ===.*"
+    )
+
+    foms = [
+        "Number of iteration",
+        "Min bandwidth",
+        "Max bandwidth",
+        "Avg bandwidth",
+        "Std bandwidth",
+        "Min latency",
+        "Max latency",
+        "Avg latency",
+        "Std latency",
+        "Min bisection bandwidth",
+        "Max bisection bandwidth",
+        "Avg bisection bandwidth",
+        "Std bisection bandwidth",
+        "Min bisection efficiency",
+        "Max bisection efficiency",
+        "Avg bisection efficiency",
+    ]
+
+    for fom in foms:
+        fom_regex = rf"\s*{fom}\s+\:\s(?P<val>\d+\.*\d*)\s*(?P<unit>.*)"
+        figure_of_merit(
+            fom,
+            fom_regex=fom_regex,
+            group_name="val",
+            units="{unit}",
+        )