Early changes

.github/workflows/profiler.yml

@@ -0,0 +1,25 @@
+name: TileDB-SOMA Profiler run
+
+on:
+  push:
+    branches:
+        - main
+
+    pull_request:
+
+    permissions:
+      contents: read
+      pull-requests: read
+
+jobs:
+
+  check-links:
+
+  name: Setup profiler
+
+  run: pip install profiler
+
+  name: Run profiler test
+
+  run: |
+    ./profiler/perf_checker.sh

profiler/ann_data.py

@@ -0,0 +1,21 @@
+import cellxgene_census
+import tiledbsoma as soma
+from time import perf_counter
+
+census_S3_latest = dict(census_version="latest")
+census_local_copy = dict(uri="/Users/brobatmili/projects/census_data/")
+
+
+def main():
+    t1 = perf_counter()
+    with cellxgene_census.open_soma(**census_local_copy) as census:
+        with census["census_data"]["homo_sapiens"].axis_query(
+                measurement_name="RNA", obs_query=soma.AxisQuery(
+                    value_filter="""tissue_general == 'eye'""")
+
+        ) as query:
+            query.to_anndata(X_name="raw")
+    t2 = perf_counter()
+    print(f"End to end time {t2 - t1}")
+
+main()

profiler/perf_checker.sh

@@ -0,0 +1,5 @@
+#!/bin/bash
+
+#python -m profiler "python ann_data.py" -t gtime
+
+python ./top_profiler.py

profiler/top_profiler.py

@@ -0,0 +1,34 @@
+from subprocess import PIPE
+import subprocess
+import data
+
+threshold = 1.10  # Percent difference
+
+db = data.FileBasedProfileDB()
+
+dt = db.find('python ann_data.py')
+last_two = dt[-2:]
+
+for s in dt:
+    second_max_ts: float = 0;
+    max_ts: float = 0;
+    l = 0
+    for ii in dt:
+        if ii.timestamp > max_ts:
+            second_max_ts = max_ts
+            max_ts = l;
+            l = l + 1;
+    print(f"max_ts {max_ts} second_max_ts {second_max_ts}")
+    print(s.user_time_sec)
+    last_index = -1
+
+    print(f"Found indexes are {max_ts} {second_max_ts}")
+    L = [1, 2]
+    L[0] = dt[max_ts].user_time_sec
+    L[1] = dt[second_max_ts].user_time_sec
+    print(f"{L[0]} {L[1]}")
+    for i in range(0, 8):
+        print(f"{i} dt[i].user_time_sec = {dt[i].user_time_sec}")
+
+    if threshold * float(L[0]) > float(L[1])  or threshold * float(L[1]) > float(L[0]):
+        raise SystemExit(f"Potential performance degradation detected {L[0]} va {L[1]}")