Improve benchmark output.

example/bench.cpp

@@ -2,34 +2,37 @@
 #include "tc/groups.hpp"
 
 #include <ctime>
-#include <vector>
 #include <iostream>
+#include <iomanip>
+
+template<class G>
+void test(const G &group) {
+    auto s = std::clock();
+    auto cosets = group.solve();
+    auto e = std::clock();
+
+    double diff = (double) (e - s) / CLOCKS_PER_SEC;
+    int order = cosets.size();
+
+    std::cout
+        << std::setw(7) << group.name << ", "
+        << std::setw(7) << order << ", "
+        << std::fixed << std::setprecision(6) << diff << "s"
+        << std::endl;
+}
 
 int main() {
-    std::vector<tc::Group> groups = {
-        tc::group::H(2),
-        tc::group::H(3),
-        tc::group::H(4),
-        tc::group::T(100),
-        tc::group::T(500),
-        tc::group::T(1000),
-        tc::group::E(6),
-        tc::group::E(7),
-        tc::group::B(6),
-        tc::group::B(7),
-        tc::group::B(8)
-    };
-
-    for (const auto &group : groups) {
-        auto s = std::clock(); // to measure CPU time
-        auto cosets = group.solve();
-        auto e = std::clock();
-
-        double diff = (double) (e - s) / CLOCKS_PER_SEC;
-        int order = cosets.size();
-
-        std::cout << group.name << "," << order << "," << diff << std::endl;
-    }
+    test(tc::group::H(2));
+    test(tc::group::H(3));
+    test(tc::group::H(4));
+    test(tc::group::T(100));
+    test(tc::group::T(500));
+    test(tc::group::T(1000));
+    test(tc::group::E(6));
+    test(tc::group::E(7));
+    test(tc::group::B(6));
+    test(tc::group::B(7));
+    test(tc::group::B(8));
 
     return 0;
 }