Add synthetic string repro graph benchmarks

Author: lmeyerov

benchmarks/gfql/filter_pagerank/benchmark_to_cugraph_gpu.py

@@ -29,6 +29,8 @@
     "gplus",
     "synthetic_contiguous",
     "synthetic_offset",
+    "synthetic_string_offset",
+    "synthetic_string_gplus_shape",
 ]
 
 STRATEGY_CHOICES = [
@@ -41,6 +43,9 @@
 ]
 
 
+GPLUS_SHAPE_VERTICES = 107_614
+
+
 def ensure_cached(dataset: str, data_dir: Path) -> Path:
     spec = dataset_spec(dataset)
     path = data_dir / str(spec["filename"])
@@ -103,16 +108,29 @@ def strategy_skip_reason(strategy: str, profile: Dict[str, object]) -> str | Non
 
 def build_synthetic_edges(graph_kind: str, edge_count: int):
     import cudf  # type: ignore
+    import cupy as cp  # type: ignore
+
+    if graph_kind == "synthetic_string_gplus_shape":
+        start = 1_000_000
+        base = cudf.Series(cp.arange(edge_count, dtype=cp.int32) % cp.int32(GPLUS_SHAPE_VERTICES))
+        src = (base + start).astype("int32").astype("str")
+        dst = ((base + 1) % GPLUS_SHAPE_VERTICES + start).astype("int32").astype("str")
+        return cudf.DataFrame({"src": src, "dst": dst})
 
     if graph_kind == "synthetic_contiguous":
         start = 0
     elif graph_kind == "synthetic_offset":
         start = 1_000_000
+    elif graph_kind == "synthetic_string_offset":
+        start = 1_000_000
     else:
         raise ValueError(f"Unsupported synthetic graph kind: {graph_kind}")
 
     src = cudf.Series(range(start, start + edge_count), dtype="int32")
     dst = cudf.Series(range(start + 1, start + edge_count + 1), dtype="int32")
+    if graph_kind == "synthetic_string_offset":
+        src = src.astype("str")
+        dst = dst.astype("str")
     return cudf.DataFrame({"src": src, "dst": dst})