add country-colored perfect-score wall-vs-total graph

README.md

@@ -195,6 +195,11 @@ finish in 1.75 h while a 4 B model takes 14.5 h.
 
 ![Wall clock vs total parameters](plots/wall_vs_total.png)
 
+**Wall clock vs total parameters — perfect-score runs, by model origin.** Same filter and
+color key as the active-parameter version; wall time still includes voided tasks.
+
+![Wall clock vs total parameters, perfect-score runs by country](plots/wall_vs_total_perfect_country.png)
+
 **Wall clock vs output tokens.** The clearest single driver: the runs that spend more output
 tokens take longer. The Spark outliers are the high-token runs, not the large models.
 

scripts/make_plots.py

@@ -130,6 +130,10 @@ def main() -> None:
             "Wall clock vs active parameters — perfect-score runs (GB10)",
             "Active parameters (B, log)", "Wall clock (hours)",
             os.path.join(args.out, "wall_vs_active_perfect_country.png"))
+    scatter_country(perfect, "total_b", "wall_h",
+            "Wall clock vs total parameters — perfect-score runs (GB10)",
+            "Total parameters (B, log)", "Wall clock (hours)",
+            os.path.join(args.out, "wall_vs_total_perfect_country.png"))
     scatter(gb, "total_b", "wall_h",
             "Wall clock vs total parameters (GB10, 42 tasks)",
             "Total parameters (B, log)", "Wall clock (hours)",
@@ -218,7 +222,7 @@ def main() -> None:
     fig.savefig(os.path.join(args.out, "perf_per_byte.png"), bbox_inches="tight")
     plt.close(fig)
 
-    print(f"wrote 6 plots to {args.out} ({len(gb)} GB10 / {len(data) - len(gb)} Thor rows)")
+    print(f"wrote 7 plots to {args.out} ({len(gb)} GB10 / {len(data) - len(gb)} Thor rows)")
 
 
 if __name__ == "__main__":