add graph500 benchmark

var/ramble/repos/builtin/applications/graph500/application.py

@@ -0,0 +1,268 @@
+# Copyright 2022-2026 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 Graph500(ExecutableApplication):
+    """The Graph500 benchmark is a data-intensive graph algorithms benchmark
+    designed for high-performance computing, memory subsystem analysis, and
+    large-scale graph traversal evaluation.
+
+    It stresses memory bandwidth, latency, and irregular all-to-all communication
+    patterns using Kronecker/R-MAT power-law graph generation and Breadth-First
+    Search (BFS).
+
+    https://graph500.org
+    https://github.com/graph500/graph500
+    """
+
+    name = "graph500"
+
+    maintainers("juntangc")
+
+    tags(
+        "graph",
+        "benchmark",
+        "mpi-benchmark",
+        "micro-benchmark",
+        "eda-proxy",
+    )
+
+    version("3.0.0", "Version 3.0.0 of Graph500", preferred=True)
+
+    with when("package_manager_family=spack"):
+        define_compiler("gcc14", pkg_spec="gcc@14.2.0")
+
+        software_spec(
+            "intel-mpi",
+            pkg_spec="intel-oneapi-mpi@2021.17.2",
+        )
+
+        software_spec(
+            "graph500",
+            pkg_spec="graph500@{application::graph500::version} +procs_not_power_of_two",
+            compiler="gcc14",
+        )
+
+        required_package("graph500")
+
+    executable(
+        "execute-bfs",
+        "graph500_reference_bfs {scale} {edge_factor} {additional_args}",
+        use_mpi=True,
+    )
+
+    executable(
+        "execute-bfs-sssp",
+        "graph500_reference_bfs_sssp {scale} {edge_factor} {additional_args}",
+        use_mpi=True,
+    )
+
+    workload("bfs", executable="execute-bfs")
+    workload("bfs_sssp", executable="execute-bfs-sssp")
+
+    # Common pre-defined scale workloads (2^N vertices)
+    scales = [16, 18, 20, 21, 22, 24, 26, 28, 30]
+    for s in scales:
+        workload(f"scale_{s}", executable="execute-bfs")
+
+    workload_group(
+        "all_workloads",
+        workloads=["bfs", "bfs_sssp"] + [f"scale_{s}" for s in scales],
+    )
+
+    workload_variable(
+        "scale",
+        default="21",
+        description="Scale of the graph (2^scale vertices; e.g. 21 = 2M vertices)",
+        workloads=["bfs", "bfs_sssp"],
+    )
+
+    # Set default scale for each scale-specific workload
+    for s in scales:
+        workload_variable(
+            "scale",
+            default=str(s),
+            description=f"Scale {s} graph (2^{s} vertices)",
+            workloads=[f"scale_{s}"],
+        )
+
+    workload_variable(
+        "edge_factor",
+        default="16",
+        description="Half the average degree of a vertex (default 16)",
+        workload_group="all_workloads",
+    )
+
+    workload_variable(
+        "additional_args",
+        default="",
+        description="Additional arguments for graph500 execution",
+        workload_group="all_workloads",
+    )
+
+    # =========================================================================
+    # Figures of Merit (FOMs)
+    # =========================================================================
+    fom_regex_float = r"[0-9]+(?:\.[0-9]+)?(?:[eE][\+\-]?[0-9]+)?"
+
+    # Primary Benchmark Metric: Harmonic Mean TEPS
+    figure_of_merit(
+        "Harmonic Mean TEPS",
+        fom_regex=r"bfs\s+harmonic_mean_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "Harmonic Stddev TEPS",
+        fom_regex=r"bfs\s+harmonic_stddev_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "Min TEPS",
+        fom_regex=r"bfs\s+min_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "First Quartile TEPS",
+        fom_regex=r"bfs\s+firstquartile_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "Median TEPS",
+        fom_regex=r"bfs\s+median_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "Third Quartile TEPS",
+        fom_regex=r"bfs\s+thirdquartile_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    figure_of_merit(
+        "Max TEPS",
+        fom_regex=r"bfs\s+max_TEPS:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="TEPS",
+    )
+
+    # Timing Figures of Merit
+    figure_of_merit(
+        "Graph Generation Time",
+        fom_regex=r"graph_generation:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Construction Time",
+        fom_regex=r"construction_time:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Min BFS Time",
+        fom_regex=r"bfs\s+min_time:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Median BFS Time",
+        fom_regex=r"bfs\s+median_time:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Mean BFS Time",
+        fom_regex=r"bfs\s+mean_time:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Max BFS Time",
+        fom_regex=r"bfs\s+max_time:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    figure_of_merit(
+        "Mean Validate Time",
+        fom_regex=r"bfs\s+mean_validate:\s+(?P<fom>"
+        + fom_regex_float
+        + r")",
+        group_name="fom",
+        units="s",
+    )
+
+    # Configuration Metadata FOMs
+    figure_of_merit(
+        "Graph Scale",
+        fom_regex=r"SCALE:\s+(?P<scale>[0-9]+)",
+        group_name="scale",
+        units="",
+    )
+
+    figure_of_merit(
+        "Edge Factor",
+        fom_regex=r"edgefactor:\s+(?P<ef>[0-9]+)",
+        group_name="ef",
+        units="",
+    )
+
+    figure_of_merit(
+        "MPI Processes",
+        fom_regex=r"num_mpi_processes:\s+(?P<nprocs>[0-9]+)",
+        group_name="nprocs",
+        units="",
+    )
+
+    success_criteria(
+        "passed",
+        mode="string",
+        match=r"bfs\s+harmonic_mean_TEPS:\s+" + fom_regex_float,
+    )