Fix ZeroDivisionError

nsight/transformation.py

@@ -117,7 +117,9 @@ def convert_non_sortable_columns(dframe: pd.DataFrame) -> pd.DataFrame:
     )
 
     # Compute relative standard deviation as a percentage
-    agg_df["RelativeStdDevPct"] = (agg_df["StdDev"] / agg_df["AvgValue"]) * 100
+    agg_df["RelativeStdDevPct"] = (
+        agg_df["StdDev"] / agg_df["AvgValue"].replace(0, np.nan)
+    ) * 100
 
     # Flag measurements as stable if relative stddev is less than 2%
     agg_df["StableMeasurement"] = agg_df["RelativeStdDevPct"] < 2.0

tests/test_profiler.py

@@ -1068,6 +1068,13 @@ def profiled_func(x: int, y: int) -> None:
             "invalid_multiple",
             id="invalid_multiple",
         ),
+        pytest.param(
+            [
+                "sm__idc_divergent_instructions.avg",
+            ],
+            "valid_single_zero",
+            id="valid_single_zero",
+        ),
     ],
 )
 def test_parameter_metrics(metrics: Sequence[str], expected_result: str) -> None:
@@ -1099,6 +1106,16 @@ def profiled_func(x: int, y: int) -> None:
         assert (
             df["AvgValue"].notna() & df["AvgValue"] > 0
         ).all(), f"Invalid AvgValue for metric {metrics}"
+    elif expected_result == "valid_single_zero":
+        profile_output = profiled_func()
+        df = profile_output.to_dataframe()
+
+        # Verify that zero values are handled correctly
+        # For sm__idc_divergent_instructions, the value should be zero
+        # since simple add operations don't have divergent instructions
+        assert (
+            df["AvgValue"] == 0
+        ).all(), f"Expected zero value for metric {metrics[0]} in simple kernels, got {df['AvgValue'].tolist()}"
     elif expected_result == "invalid_multiple":
         with pytest.raises(
             ValueError,