Artifact of SSBSE '25 Challenge: GreenMalloc: Allocator Optimisation for Industrial Workloads
1. Abstract This is the artifact of the paper GreenMalloc: Allocator Optimisation for Industrial Workloads. It contains the code of GreenMalloc and the results of the evaluation on gem5 with glibc malloc and tcmalloc. The GitHub of the project is available online as open source: https://github.com/dakaidan/SSBSE25-GreenBenchmark. 2. Optimised Values glibc: - mmap_max - mmap_threshold - perturb - top_pad - trim_threshold - arena_test - arena_max tcmalloc: - TCMALLOC_SAMPLE_PARAMETER - TCMALLOC_RELEASE_RATE - TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES - TCMALLOC_AGGRESSIVE_DECOMMIT - TCMALLOC_OVERRIDE_PAGESIZE - TCMALLOC_HEAP_LIMIT_MB - TCMALLOC_SKIP_MMAP - TCMALLOC_SKIP_SBRK - TCMALLOC_MEMFS_LIMIT_MB - TCMALLOC_MEMFS_ABORT_ON_FAIL - TCMALLOC_MEMFS_IGNORE_MMAP_FAIL - TCMALLOC_MEMFS_MAP_PRIVATE - TCMALLOC_MEMFS_DISABLE_FALLBACK Specifically, we tune: TUNABLE_SPECS: ClassVar[Dict[str, Dict[str, Any]]] = { "mmap_max": {"type": "int", "low": 0, "high": 16 * 1024 * 1024}, "mmap_threshold": {"type": "int", "low": 0, "high": 1 << 20}, "perturb": {"type": "int", "low": 0, "high": 255}, "top_pad": {"type": "int", "low": 0, "high": 1 << 20}, "trim_threshold": {"type": "int", "low": 0, "high": 1 << 20}, "arena_test": {"type": "int", "low": 0, "high": 1 << 20}, "arena_max": {"type": "int", "low": 0, "high": 128},} TUNABLE_SPECS: ClassVar[Dict[str, Dict[str, Any]]] = { "TCMALLOC_SAMPLE_PARAMETER": {"type": "int", "low": 0, "high": 1 << 20}, "TCMALLOC_RELEASE_RATE": {"type": "int", "low": 0, "high": 1 << 20}, "TCMALLOC_MAX_TOTAL_THREAD_CACHE_BYTES": {"type": "int", "low": 0, "high": 1 << 30}, "TCMALLOC_AGGRESSIVE_DECOMMIT": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_OVERRIDE_PAGESIZE": {"type": "int", "low": 4096, "high": 1 << 20}, "TCMALLOC_HEAP_LIMIT_MB": {"type": "int", "low": 0, "high": 1 << 20}, "TCMALLOC_SKIP_MMAP": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_SKIP_SBRK": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_MEMFS_LIMIT_MB": {"type": "int", "low": 0, "high": 1 << 20}, "TCMALLOC_MEMFS_ABORT_ON_FAIL": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_MEMFS_IGNORE_MMAP_FAIL": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_MEMFS_MAP_PRIVATE": {"type": "bool", "low": 0, "high": 1}, "TCMALLOC_MEMFS_DISABLE_FALLBACK": {"type": "bool", "low": 0, "high": 1}, }
Institutions
- King's College London (GB)
- University College London (GB)
Publication Details
- Journal
- Zenodo (CERN European Organization for Nuclear Research)
- Published
- 2025-09-21
- DOI
- https://doi.org/10.5281/zenodo.17182847
- Primary Topic
- Library Science and Administration
- Type
- article
- Field-Weighted Citation Impact
- 0.00