diff --git a/src/klotski_core/benchmark/benchmark.cc b/src/klotski_core/benchmark/benchmark.cc
index cecd9a1..570d042 100644
--- a/src/klotski_core/benchmark/benchmark.cc
+++ b/src/klotski_core/benchmark/benchmark.cc
@@ -1,9 +1,9 @@
 #include "benchmark.h"
 #include "all_cases.h"
 
-using namespace klotski;
+using klotski::Benchmark;
 
-float Benchmark::basic_ranges(Benchmark::TIME format) {
+float Benchmark::basic_ranges(TIME format) {
     if (BasicRanges::status() != BasicRanges::NO_INIT) {
         return -1; // data already built -> skip
     }
@@ -12,7 +12,7 @@ float Benchmark::basic_ranges(Benchmark::TIME format) {
     return time_format(start, format);
 }
 
-float Benchmark::all_cases(Benchmark::TIME format) {
+float Benchmark::all_cases(TIME format) {
     if (AllCases::status() != AllCases::NO_INIT) {
         return -1; // data already built -> skip
     }
diff --git a/src/klotski_core/benchmark/benchmark.h b/src/klotski_core/benchmark/benchmark.h
index 79315e2..c48bfb0 100644
--- a/src/klotski_core/benchmark/benchmark.h
+++ b/src/klotski_core/benchmark/benchmark.h
@@ -1,6 +1,22 @@
 #pragma once
 
-#include <ostream>
+/// This is a built-in performance testing module for klotski core, which will
+/// directly call the underlying functions to avoid the additional overhead
+/// of the glue ABI.
+
+/// Multiple test items are provided here, and the corresponding float value is
+/// returned according to the specified time format (seconds, milliseconds,
+/// microseconds, nanoseconds).
+
+/// The test items are all multi-thread safe, but you should not run multiple
+/// test items at the same time, which will lead to unstable tests in many ways,
+/// such as changes in CPU turbo frequency.
+
+/// Pay attention to the two test items `basic_ranges` and `all_cases`, they can
+/// only be run once (the reason for the construction of static data), and cannot
+/// be run after short code related items.
+
+#include <ctime>
 
 namespace klotski {
     class Benchmark {
@@ -8,14 +24,10 @@ namespace klotski {
         enum TIME {
             S, MS, US, NS
         };
-        static float all_cases(enum TIME format);
-        static float basic_ranges(enum TIME format);
+        static float all_cases(TIME format = MS);
+        static float basic_ranges(TIME format = MS);
 
     private:
-//        static std::string time_s(clock_t start);
-//        static std::string time_ms(clock_t start);
-//        static std::string time_us(clock_t start);
-//        static std::string time_ns(clock_t start);
         static float time_format(clock_t start, enum TIME format);
     };
 }
diff --git a/src/klotski_core/benchmark/chore.cc b/src/klotski_core/benchmark/chore.cc
index 43cf7ee..0d2bd1c 100644
--- a/src/klotski_core/benchmark/chore.cc
+++ b/src/klotski_core/benchmark/chore.cc
@@ -1,52 +1,8 @@
 #include "benchmark.h"
 
-using namespace klotski;
+using klotski::Benchmark;
 
-// TODO: remove std::cout output -> only return benchmark status
-
-/// colorful string
-//std::string Benchmark::color_red(const std::string &str) {
-//    return std::string("\033[31m") + str + "\033[0m";
-//}
-//
-//std::string Benchmark::color_blue(const std::string &str) {
-//    return std::string("\033[36m") + str + "\033[0m";
-//}
-//
-//std::string Benchmark::color_green(const std::string &str) {
-//    return std::string("\033[32m") + str + "\033[0m";
-//}
-//
-//std::string Benchmark::color_yellow(const std::string &str) {
-//    return std::string("\033[33m") + str + "\033[0m";
-//}
-
-/// used-time to green string
-//std::string Benchmark::time_s(clock_t start) {
-//    return color_green(
-//        std::to_string((clock() - start) / CLOCKS_PER_SEC) + "s"
-//    );
-//}
-//
-//std::string Benchmark::time_ms(clock_t start) {
-//    return color_green(
-//        std::to_string((clock() - start) * 1000 / CLOCKS_PER_SEC) + "ms"
-//    );
-//}
-//
-//std::string Benchmark::time_us(clock_t start) {
-//    return color_green(
-//        std::to_string((clock() - start) * 1000 * 1000 / CLOCKS_PER_SEC) + "us"
-//    );
-//}
-//
-//std::string Benchmark::time_ns(clock_t start) {
-//    return color_green(
-//        std::to_string((clock() - start) * 1000 * 1000 * 1000 / CLOCKS_PER_SEC) + "us"
-//    );
-//}
-
-float Benchmark::time_format(clock_t start, Benchmark::TIME format) {
+float Benchmark::time_format(clock_t start, TIME format) {
     auto time = float(clock() - start);
     switch (format) {
         case S:
diff --git a/src/klotski_core/core/core.h b/src/klotski_core/core/core.h
index 4ebab3a..91a084b 100644
--- a/src/klotski_core/core/core.h
+++ b/src/klotski_core/core/core.h
@@ -24,7 +24,7 @@
 /// the help of a sub-BFS search, the target can be obtained. It can be shown that this step
 /// produces at most 15 derived layouts, so it can be stored and computed in a queue of length
 /// 16. By performing such a search on each block in the layout, we can get all the next-step
-/// layouts, which have a minimum of 0 and a maximum of 68.
+/// layouts, which have a minimum of `0` and a maximum of `68`.
 
 /// For a single block, search for the situation after it has moved one grid, `addr` is its current
 /// position information, which is (0 ~ 19) * 3. When moving up and down, judge the value of `addr`
diff --git a/src/klotski_core/ffi/tmain.cc b/src/klotski_core/ffi/tmain.cc
index 004c9bb..e471b0d 100644
--- a/src/klotski_core/ffi/tmain.cc
+++ b/src/klotski_core/ffi/tmain.cc
@@ -5,45 +5,22 @@
 #include "klotski.h"
 #include "core.h"
 
-//#include "common.h"
+#include "common.h"
 #include "benchmark.h"
 
 #include "all_cases.h"
 #include "common_code.h"
 
-void tmain() {
-    printf("tmain start\n");
-
-//    klotski::RawCode::from_common_code(0x1A9BF0C00).is_horizontal_mirror();
-//    auto ret = klotski::RawCode::from_common_code(0x1A9BF0C00).to_horizontal_mirror();
-//    auto ret = klotski::RawCode::from_common_code(0x4FEA13400).to_horizontal_mirror();
+using klotski::Benchmark;
 
-//    std::cout << ret.to_common_code() << std::endl << ret << std::endl;
-
-//    klotski::RawCode::from_common_code(0x1A9BF0C00).is_vertical_mirror();
-//    auto ret = klotski::RawCode::from_common_code(0x1A9BF0C00).to_vertical_mirror();
-//
-//    std::cout << ret.to_common_code() << std::endl << ret << std::endl;
+void tmain() {
+//    printf("tmain start\n");
 
-//    std::cout << klotski::RawCode::from_common_code(0x4FEA13400) << std::endl;
-//    std::cout << klotski::RawCode::from_common_code(0x8346AFC00) << std::endl;
-//    std::cout << klotski::RawCode::from_common_code(0x1A9BF0C00) << std::endl;
-    std::cout << klotski::RawCode::from_common_code(0x6BFA47000) << std::endl;
+    std::cout << "basic ranges: " << Benchmark::basic_ranges() << "ms" << std::endl;
+    std::cout << "all cases: " << Benchmark::all_cases() << "ms" << std::endl;
 
     return;
 
-//    printf("%d\n", ALL_CASES_SIZE_SUM);
-//    std::cout << ALL_CASES_SIZE_SUM << std::endl;
-
-//    uint64_t common_code = 0x1A9BC0C00;
-//    klotski::Common::range_reverse(common_code);
-
-//    printf("%f\n", Benchmark::basic_ranges(Benchmark::MS));
-//    printf("%f\n", Benchmark::all_cases(Benchmark::MS));
-//    std::cout << Benchmark::basic_ranges(Benchmark::MS) << std::endl;
-//    std::cout << Benchmark::all_cases(Benchmark::MS) << std::endl;
-
-
     std::vector<uint64_t> next;
 
     auto core = klotski::Core([&next](uint64_t code, uint64_t) {