{ "__type": "IngestedDoc", "__tag": 4010, "_content": {}, "_ordered_sections": [], "item_file": null, "item_line": null, "item_type": null, "aliases": [], "example_section_data": { "__type": "Section", "__tag": 4015, "children": [], "title": [], "level": 0, "target": null }, "see_also": [], "signature": null, "references": null, "qa": "reference:global_state", "arbitrary": [ { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "NumPy has a few import-time, compile-time, or runtime configuration options which change the global behaviour. Most of these are related to performance or for debugging purposes and will not be interesting to the vast majority of users." } ] } ], "title": [ { "__type": "Text", "__tag": 4046, "value": "Global Configuration Options" } ], "level": 0, "target": "global_state" }, { "__type": "Section", "__tag": 4015, "children": [], "title": [ { "__type": "Text", "__tag": 4046, "value": "Performance-related options" } ], "level": 1, "target": null }, { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "NumPy itself is normally intentionally limited to a single thread during function calls, however it does support multiple Python threads running at the same time. Note that for performant linear algebra NumPy uses a BLAS backend such as OpenBLAS or MKL, which may use multiple threads that may be controlled by environment variables such as " }, { "__type": "InlineCode", "__tag": 4051, "value": "OMP_NUM_THREADS" }, { "__type": "Text", "__tag": 4046, "value": " depending on what is used. One way to control the number of threads is the package " }, { "__type": "Link", "__tag": 4049, "children": [ { "__type": "Text", "__tag": 4046, "value": "threadpoolctl" } ], "url": "https://pypi.org/project/threadpoolctl/", "title": "" } ] } ], "title": [ { "__type": "Text", "__tag": 4046, "value": "Number of threads used for linear algebra" } ], "level": 2, "target": null }, { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "When working with very large arrays on modern Linux kernels, you can experience a significant speedup when " }, { "__type": "Link", "__tag": 4049, "children": [ { "__type": "Text", "__tag": 4046, "value": "transparent hugepage" } ], "url": "https://www.kernel.org/doc/html/latest/admin-guide/mm/transhuge.html", "title": "" }, { "__type": "Text", "__tag": 4046, "value": " is used. The current system policy for transparent hugepages can be seen by " } ] }, { "__type": "Code", "__tag": 4050, "value": "cat /sys/kernel/mm/transparent_hugepage/enabled", "execution_status": null }, { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "When set to " }, { "__type": "InlineCode", "__tag": 4051, "value": "madvise" }, { "__type": "Text", "__tag": 4046, "value": " NumPy will typically use hugepages for a performance boost. This behaviour can be modified by setting the environment variable " } ] }, { "__type": "Code", "__tag": 4050, "value": "NUMPY_MADVISE_HUGEPAGE=0", "execution_status": null }, { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "or setting it to " }, { "__type": "InlineCode", "__tag": 4051, "value": "1" }, { "__type": "Text", "__tag": 4046, "value": " to always enable it. When not set, the default is to use madvise on Kernels 4.6 and newer. These kernels presumably experience a large speedup with hugepage support. This flag is checked at import time." } ] } ], "title": [ { "__type": "Text", "__tag": 4046, "value": "madvise hugepage on Linux" } ], "level": 2, "target": null }, { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Setting " }, { "__type": "InlineCode", "__tag": 4051, "value": "NPY_DISABLE_CPU_FEATURES" }, { "__type": "Text", "__tag": 4046, "value": " will exclude simd features at runtime. See " }, { "__type": "CrossRef", "__tag": 4002, "value": "runtime-simd-dispatch", "reference": { "__type": "LocalRef", "__tag": 4022, "kind": "docs", "path": "reference:simd:build-options" }, "kind": "exists" }, { "__type": "Text", "__tag": 4046, "value": "." } ] } ], "title": [ { "__type": "Text", "__tag": 4046, "value": "SIMD feature selection" } ], "level": 2, "target": null }, { "__type": "Section", "__tag": 4015, "children": [], "title": [ { "__type": "Text", "__tag": 4046, "value": "Debugging-related options" } ], "level": 1, "target": null }, { "__type": "Section", "__tag": 4015, "children": [ { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "Some users might pass ownership of the data pointer to the " }, { "__type": "InlineCode", "__tag": 4051, "value": "ndarray" }, { "__type": "Text", "__tag": 4046, "value": " by setting the " }, { "__type": "InlineCode", "__tag": 4051, "value": "OWNDATA" }, { "__type": "Text", "__tag": 4046, "value": " flag. If they do this without setting (manually) a memory allocation policy, the default will be to call " }, { "__type": "InlineCode", "__tag": 4051, "value": "free" }, { "__type": "Text", "__tag": 4046, "value": ". If " }, { "__type": "InlineCode", "__tag": 4051, "value": "NUMPY_WARN_IF_NO_MEM_POLICY" }, { "__type": "Text", "__tag": 4046, "value": " is set to " }, { "__type": "InlineCode", "__tag": 4051, "value": "\"1\"" }, { "__type": "Text", "__tag": 4046, "value": ", a " }, { "__type": "InlineCode", "__tag": 4051, "value": "RuntimeWarning" }, { "__type": "Text", "__tag": 4046, "value": " will be emitted. A better alternative is to use a " }, { "__type": "InlineCode", "__tag": 4051, "value": "PyCapsule" }, { "__type": "Text", "__tag": 4046, "value": " with a deallocator and set the " }, { "__type": "InlineCode", "__tag": 4051, "value": "ndarray.base" }, { "__type": "Text", "__tag": 4046, "value": "." } ] }, { "__type": "Admonition", "__tag": 4056, "kind": "versionchanged", "base_type": "neutral", "children": [ { "__type": "AdmonitionTitle", "__tag": 4055, "children": [ { "__type": "Text", "__tag": 4046, "value": "versionchanged 1.25.2" } ] }, { "__type": "Paragraph", "__tag": 4045, "children": [ { "__type": "Text", "__tag": 4046, "value": "This variable is only checked on the first import." } ] } ] } ], "title": [ { "__type": "Text", "__tag": 4046, "value": "Warn if no memory allocation policy when deallocating data" } ], "level": 2, "target": null } ], "local_refs": [] }