[skip-ci] Add cftime groupby, resample benchmarks

asv_bench/asv.conf.json

@@ -30,6 +30,7 @@
     // determined by looking for tools on the PATH environment
     // variable.
     "environment_type": "conda",
+    "conda_channels": ["conda-forge"],
 
     // timeout in seconds for installing any dependencies in environment
     // defaults to 10 min

asv_bench/benchmarks/groupby.py

@@ -18,23 +18,29 @@ def setup(self, *args, **kwargs):
                 "c": xr.DataArray(np.arange(2 * self.n)),
             }
         )
-        self.ds2d = self.ds1d.expand_dims(z=10)
+        self.ds2d = self.ds1d.expand_dims(z=10).copy()
         self.ds1d_mean = self.ds1d.groupby("b").mean()
         self.ds2d_mean = self.ds2d.groupby("b").mean()
 
     @parameterized(["ndim"], [(1, 2)])
     def time_init(self, ndim):
         getattr(self, f"ds{ndim}d").groupby("b")
 
-    @parameterized(["method", "ndim"], [("sum", "mean"), (1, 2)])
-    def time_agg_small_num_groups(self, method, ndim):
+    @parameterized(
+        ["method", "ndim", "use_flox"], [("sum", "mean"), (1, 2), (True, False)]
+    )
+    def time_agg_small_num_groups(self, method, ndim, use_flox):
         ds = getattr(self, f"ds{ndim}d")
-        getattr(ds.groupby("a"), method)().compute()
+        with xr.set_options(use_flox=use_flox):
+            getattr(ds.groupby("a"), method)().compute()
 
-    @parameterized(["method", "ndim"], [("sum", "mean"), (1, 2)])
-    def time_agg_large_num_groups(self, method, ndim):
+    @parameterized(
+        ["method", "ndim", "use_flox"], [("sum", "mean"), (1, 2), (True, False)]
+    )
+    def time_agg_large_num_groups(self, method, ndim, use_flox):
         ds = getattr(self, f"ds{ndim}d")
-        getattr(ds.groupby("b"), method)().compute()
+        with xr.set_options(use_flox=use_flox):
+            getattr(ds.groupby("b"), method)().compute()
 
     def time_binary_op_1d(self):
         (self.ds1d.groupby("b") - self.ds1d_mean).compute()
@@ -115,10 +121,13 @@ def setup(self, *args, **kwargs):
     def time_init(self, ndim):
         getattr(self, f"ds{ndim}d").resample(time="D")
 
-    @parameterized(["method", "ndim"], [("sum", "mean"), (1, 2)])
-    def time_agg_small_num_groups(self, method, ndim):
+    @parameterized(
+        ["method", "ndim", "use_flox"], [("sum", "mean"), (1, 2), (True, False)]
+    )
+    def time_agg_small_num_groups(self, method, ndim, use_flox):
         ds = getattr(self, f"ds{ndim}d")
-        getattr(ds.resample(time="3M"), method)().compute()
+        with xr.set_options(use_flox=use_flox):
+            getattr(ds.resample(time="3M"), method)().compute()
 
     @parameterized(["method", "ndim"], [("sum", "mean"), (1, 2)])
     def time_agg_large_num_groups(self, method, ndim):
@@ -132,3 +141,34 @@ def setup(self, *args, **kwargs):
         super().setup(**kwargs)
         self.ds1d = self.ds1d.chunk({"time": 50})
         self.ds2d = self.ds2d.chunk({"time": 50, "z": 4})
+
+
+class ResampleCFTime:
+    def setup(self, *args, **kwargs):
+        self.ds1d = xr.Dataset(
+            {
+                "b": ("time", np.arange(365.0 * 24)),
+            },
+            coords={
+                "time": xr.date_range(
+                    "2001-01-01", freq="H", periods=365 * 24, calendar="noleap"
+                )
+            },
+        )
+        self.ds2d = self.ds1d.expand_dims(z=10)
+        self.ds1d_mean = self.ds1d.resample(time="48H").mean()
+        self.ds2d_mean = self.ds2d.resample(time="48H").mean()
+
+
+class GroupByLongTime:
+    params = [[True, False], [True, False]]
+    param_names = ["use_cftime", "use_flox"]
+
+    def setup(self, use_cftime, use_flox):
+        arr = np.random.randn(10, 10, 365 * 30)
+        time = xr.date_range("2000", periods=30 * 365, use_cftime=use_cftime)
+        self.da = xr.DataArray(arr, dims=("y", "x", "time"), coords={"time": time})
+
+    def time_mean(self, use_cftime, use_flox):
+        with xr.set_options(use_flox=use_flox):
+            self.da.groupby("time.year").mean()