Add benchmarks for larger catalogs. (#292)

astronomy-commons · Apr 25, 2024 · 5653387 · 5653387
1 parent 4ba21b1
commit 5653387
Show file tree

Hide file tree

Showing 9 changed files with 226,651 additions and 0 deletions.
diff --git a/benchmarks/benchmarks.py b/benchmarks/benchmarks.py
@@ -18,6 +18,7 @@
 SMALL_SKY_DIR_NAME = "small_sky"
 SMALL_SKY_ORDER1 = "small_sky_order1"
 SMALL_SKY_XMATCH_NAME = "small_sky_xmatch"
+BENCH_DATA_DIR = os.path.join(os.path.dirname(__file__), "data")
 
 
 def load_small_sky():
@@ -62,3 +63,13 @@ def time_box_filter_on_partition():
         np.linspace(-1000, 1000, 100_000), columns=[metadata.catalog_info.ra_column]
     )
     box_filter(mock_partition_df, ra=(-20, 40), dec=None, metadata=metadata).compute()
+
+
+def time_create_midsize_catalog():
+    path = os.path.join(BENCH_DATA_DIR, "midsize_catalog")
+    return lsdb.read_hipscat(path)
+
+
+def time_create_large_catalog():
+    path = os.path.join(BENCH_DATA_DIR, "large_catalog")
+    return lsdb.read_hipscat(path)
diff --git a/benchmarks/data/large_catalog/_common_metadata b/benchmarks/data/large_catalog/_common_metadata
diff --git a/benchmarks/data/large_catalog/_metadata b/benchmarks/data/large_catalog/_metadata
diff --git a/benchmarks/data/large_catalog/catalog_info.json b/benchmarks/data/large_catalog/catalog_info.json
@@ -0,0 +1,15 @@
+{
+    "catalog_name": {
+        "catalog_name": "midsize_catalog",
+        "catalog_type": "object",
+        "total_rows": 3000000000,
+        "epoch": "J2000",
+        "ra_column": "ra",
+        "dec_column": "dec"
+    },
+    "catalog_type": "object",
+    "total_rows": null,
+    "epoch": "J2000",
+    "ra_column": "ra",
+    "dec_column": "dec"
+}