chore: update benchmark sorting logic, run environment, and add join benchmarks. (#768)

Author: Genesis929

noxfile.py

@@ -818,7 +818,7 @@ def notebook(session: nox.Session):
     _print_performance_report("notebooks/")
 
 
-@nox.session(python=SYSTEM_TEST_PYTHON_VERSIONS)
+@nox.session(python=DEFAULT_PYTHON_VERSION)
 def benchmark(session: nox.Session):
     session.install("-e", ".[all]")
     base_path = os.path.join("scripts", "benchmark")
@@ -855,7 +855,7 @@ def _print_performance_report(path: str):
     """
     print("---BIGQUERY USAGE REPORT---")
     results_dict = {}
-    bytes_reports = sorted(Path(path).rglob("*.bytesprocessed"), key=lambda x: x.name)
+    bytes_reports = sorted(Path(path).rglob("*.bytesprocessed"))
     for bytes_report in bytes_reports:
         with open(bytes_report, "r") as bytes_file:
             filename = bytes_report.relative_to(path).with_suffix("")
@@ -865,7 +865,7 @@ def _print_performance_report(path: str):
             results_dict[filename] = [query_count, total_bytes]
         os.remove(bytes_report)
 
-    millis_reports = sorted(Path(path).rglob("*.slotmillis"), key=lambda x: x.name)
+    millis_reports = sorted(Path(path).rglob("*.slotmillis"))
     for millis_report in millis_reports:
         with open(millis_report, "r") as millis_file:
             filename = millis_report.relative_to(path).with_suffix("")

scripts/benchmark/db-benchmark/join/J1_1e9_NA_0_0/q1.py

@@ -0,0 +1,16 @@
+# Contains code from https://github.com/duckdblabs/db-benchmark/blob/master/pandas/join-pandas.py
+
+import bigframes.pandas as bpd
+
+print("Join benchmark 1: small inner on int")
+
+x = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_NA_0_0")
+small = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_1e3_0_0")
+
+ans = x.merge(small, on="id1")
+print(ans.shape)
+
+chk = [ans["v1"].sum(), ans["v2"].sum()]
+print(chk)
+
+bpd.reset_session()

scripts/benchmark/db-benchmark/join/J1_1e9_NA_0_0/q2.py

@@ -0,0 +1,16 @@
+# Contains code from https://github.com/duckdblabs/db-benchmark/blob/master/pandas/join-pandas.py
+
+import bigframes.pandas as bpd
+
+print("Join benchmark 2: medium inner on int")
+
+x = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_NA_0_0")
+medium = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_1e6_0_0")
+
+ans = x.merge(medium, on="id2")
+print(ans.shape)
+
+chk = [ans["v1"].sum(), ans["v2"].sum()]
+print(chk)
+
+bpd.reset_session()

scripts/benchmark/db-benchmark/join/J1_1e9_NA_0_0/q3.py

@@ -0,0 +1,16 @@
+# Contains code from https://github.com/duckdblabs/db-benchmark/blob/master/pandas/join-pandas.py
+
+import bigframes.pandas as bpd
+
+print("Join benchmark 3: medium outer on int")
+
+x = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_NA_0_0")
+medium = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_1e6_0_0")
+
+ans = x.merge(medium, how="left", on="id2")
+print(ans.shape)
+
+chk = [ans["v1"].sum(), ans["v2"].sum()]
+print(chk)
+
+bpd.reset_session()

scripts/benchmark/db-benchmark/join/J1_1e9_NA_0_0/q4.py

@@ -0,0 +1,16 @@
+# Contains code from https://github.com/duckdblabs/db-benchmark/blob/master/pandas/join-pandas.py
+
+import bigframes.pandas as bpd
+
+print("Join benchmark 4: medium inner on factor")
+
+x = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_NA_0_0")
+medium = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_1e6_0_0")
+
+ans = x.merge(medium, on="id5")
+print(ans.shape)
+
+chk = [ans["v1"].sum(), ans["v2"].sum()]
+print(chk)
+
+bpd.reset_session()

scripts/benchmark/db-benchmark/join/J1_1e9_NA_0_0/q5.py

@@ -0,0 +1,16 @@
+# Contains code from https://github.com/duckdblabs/db-benchmark/blob/master/pandas/join-pandas.py
+
+import bigframes.pandas as bpd
+
+print("Join benchmark 5: big inner on int")
+
+x = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_NA_0_0")
+big = bpd.read_gbq("bigframes-dev-perf.dbbenchmark.J1_1e9_1e9_0_0")
+
+ans = x.merge(big, on="id3")
+print(ans.shape)
+
+chk = [ans["v1"].sum(), ans["v2"].sum()]
+print(chk)
+
+bpd.reset_session()