Fixing the problems with the pull request.

Author: TomerEven

benchmarks/bulk-insert-and-query.cc

@@ -737,12 +737,12 @@ int main(int argc, char * argv[]) {
       cout << setw(NAME_WIDTH) << names[a] << cf << endl;
   }
   a = 31;
-//  if (algorithmId == a || algorithmId < 0 || (algos.find(a) != algos.end())) {
-//     auto cf = FilterBenchmark<
-//          VQFilter<uint64_t, SimpleMixSplit>>(
-//          add_count, to_add, intersectionsize, mixed_sets,  true, false);
-//      cout << setw(NAME_WIDTH) << names[a] << cf << endl;
-//  }
+  if (algorithmId == a || algorithmId < 0 || (algos.find(a) != algos.end())) {
+     auto cf = FilterBenchmark<
+          VQFilter<uint64_t, SimpleMixSplit>>(
+          add_count, to_add, intersectionsize, mixed_sets,  true, false);
+      cout << setw(NAME_WIDTH) << names[a] << cf << endl;
+  }
 #endif
 #ifdef __PF_AVX512__
   a = 32;
@@ -774,23 +774,6 @@ int main(int argc, char * argv[]) {
                 add_count, to_add, intersectionsize, mixed_sets,  false, false);
         cout << setw(NAME_WIDTH) << names[a] << cf << endl;
     }
-    /* 
-    a = 38;
-    if (algorithmId == a || algorithmId < 0 || (algos.find(a) != algos.end())) {
-        using prefix_l2 = Prefix_Filter<TrivialFilter>;
-        auto cf = FilterBenchmark<Prefix_Filter<prefix_l2>>(add_count, to_add, intersectionsize, mixed_sets,  false, false);
-        cout << setw(NAME_WIDTH) << names[a] << cf << endl;
-    }
-    a = 39;
-    if (algorithmId == a || algorithmId < 0 || (algos.find(a) != algos.end())) {
-        using prefix_l2 = Prefix_Filter<TrivialFilter>;
-        using prefix_l3 = Prefix_Filter<prefix_l2>;
-        using prefix_l4 = Prefix_Filter<prefix_l3>;
-
-        auto cf = FilterBenchmark<prefix_l4>(add_count, to_add, intersectionsize, mixed_sets,  false, false);
-        cout << setw(NAME_WIDTH) << names[a] << cf << endl;
-    } 
-    */
 #endif
 
   // Bloom ----------------------------------------------------------

benchmarks/filterapi.h

@@ -27,10 +27,10 @@
 #include "vqf_cpp.h"
 #include "simd-block.h"
 #endif
-#define __PF_AVX512__  (__AVX512CD__ & __AVX512DQ__ &__AVX512DQ__ &__AVX512DQ__)
+#define __PF_AVX512__  (__AVX512BW__ & __AVX512VL__ & __AVX512CD__ & __AVX512DQ__)
 #ifdef __PF_AVX512__
 #include "prefix/min_pd256.hpp"
-#include "tcShortcut/TC-shortcut.hpp"
+#include "tc-shortcut/tc-shortcut.hpp"
 #endif
 #include "simd-block-fixed-fpp.h"
 #include "ribbon_impl.h"
@@ -225,7 +225,6 @@ struct FilterAPI<TC_shortcut<HashFamily>> {
     static void Add(uint64_t key, Table *table) {
         if (!table->insert(key)) {
             std::cout << table->info() << std::endl;
-            //            std::cout << "max_load: \t" << 0.945 << std::endl;
             throw std::logic_error(table->get_name() + " is too small to hold all of the elements");
         }
     }
@@ -299,7 +298,7 @@ inline size_t get_l2_slots<SimdBlockFilterFixed<>>(size_t l1_items, const double
 }
 
 
-template<typename Table, typename HashFamily = TwoIndependentMultiplyShift>
+template<typename Table, typename HashFamily = hashing::TwoIndependentMultiplyShift>
 class Prefix_Filter {
     const size_t filter_max_capacity;
     const size_t number_of_pd;
@@ -326,7 +325,6 @@ class Prefix_Filter {
         }
 
         constexpr uint64_t pd256_plus_init_header = (((INT64_C(1) << min_pd::QUOTS) - 1) << 6) | 32;
-//         std_fill<__m256i *, __m256i>(pd_array, pd_array + number_of_pd, __m256i{0, 0, 0, 0});
         for (size_t i = 0; i < number_of_pd; i++){
           pd_array[i] = __m256i{pd256_plus_init_header, 0, 0, 0};
         }

src/cuckoo/printutil.cc

@@ -6,21 +6,21 @@
 
 namespace cuckoofilter {
 
-// std::string PrintUtil::bytes_to_hex(const char *data, size_t len) {
-//   std::string hexstr = "";
-//   static const char hexes[] = "0123456789ABCDEF ";
+std::string PrintUtil::bytes_to_hex(const char *data, size_t len) {
+  std::string hexstr = "";
+  static const char hexes[] = "0123456789ABCDEF ";
 
-//   for (size_t i = 0; i < len; i++) {
-//     unsigned char c = data[i];
-//     hexstr.push_back(hexes[c >> 4]);
-//     hexstr.push_back(hexes[c & 0xf]);
-//     hexstr.push_back(hexes[16]);
-//   }
-//   return hexstr;
-// };
+  for (size_t i = 0; i < len; i++) {
+    unsigned char c = data[i];
+    hexstr.push_back(hexes[c >> 4]);
+    hexstr.push_back(hexes[c & 0xf]);
+    hexstr.push_back(hexes[16]);
+  }
+  return hexstr;
+};
 
-// std::string PrintUtil::bytes_to_hex(const std::string &s) {
-//   return bytes_to_hex((const char *)s.data(), s.size());
-// };
+std::string PrintUtil::bytes_to_hex(const std::string &s) {
+  return bytes_to_hex((const char *)s.data(), s.size());
+};
 
 }  // namespace cuckoofilter

src/hashutil.h

@@ -6,45 +6,31 @@
 #include <sys/types.h>
 
 #include <string>
-#include <fstream>
 
 #include <random>
 
 namespace hashing {
+// See Martin Dietzfelbinger, "Universal hashing and k-wise independent random
+// variables via integer arithmetic without primes".
 
-inline size_t sysrandom(void *dst, size_t dstlen) {
-        char *buffer = reinterpret_cast<char *>(dst);
-        std::ifstream stream("/dev/urandom", std::ios_base::binary | std::ios_base::in);
-        stream.read(buffer, dstlen);
-
-        return dstlen;
+class TwoIndependentMultiplyShift {
+  unsigned __int128 multiply_, add_;
+ public:
+  TwoIndependentMultiplyShift() {
+    ::std::random_device random;
+    for (auto v : {&multiply_, &add_}) {
+      *v = random();
+      for (int i = 1; i <= 4; ++i) {
+        *v = *v << 32;
+        *v |= random();
+      }
     }
-
-    // See Martin Dietzfelbinger, "Universal hashing and k-wise independent random
-    // variables via integer arithmetic without primes".
-    class TwoIndependentMultiplyShift {
-        unsigned __int128 multiply_, add_;
-
-    public:
-        TwoIndependentMultiplyShift() {
-            std::uint_least64_t seed;
-            sysrandom(&seed, sizeof(seed));
-            std::mt19937_64 rng(seed);
-            std::uniform_int_distribution<uint64_t> dist(0, UINT64_MAX);
-
-            for (auto v : {&multiply_, &add_}) {
-                for (size_t i = 0; i < 8; i++) {
-                    unsigned __int128 hi = dist(rng);
-                    unsigned __int128 lo = dist(rng);
-                    *v ^= (hi << 64) | lo;
-                }
-            }
-        }
-        
-        inline uint64_t operator()(uint64_t key) const {
-            return (add_ + multiply_ * static_cast<decltype(multiply_)>(key)) >> 64;
-        }
-    };
+  }
+  inline uint64_t operator()(uint64_t key) const {
+    return (add_ + multiply_ * static_cast<decltype(multiply_)>(key)) >> 64;
+  }
+  
+};
 
 
 class SimpleMixSplit {
@@ -74,4 +60,4 @@ class SimpleMixSplit {
 
 }
 
-#endif  // CUCKOO_FILTER_HASHUTIL_H_
+#endif  // CUCKOO_FILTER_HASHUTIL_H_

src/tcShortcut/TC-shortcut.hpp src/tc-shortcut/tc-shortcut.hppsrc/tcShortcut/TC-shortcut.hpp renamed to src/tc-shortcut/tc-shortcut.hpp

@@ -6,7 +6,7 @@
 #include "hashutil.h"
 #include "./tc-sym.hpp"
 
-template <typename HashFamily = TwoIndependentMultiplyShift>
+template <typename HashFamily = hashing::TwoIndependentMultiplyShift>
 class TC_shortcut {
     HashFamily h0;
     size_t capacity{0};
@@ -280,7 +280,7 @@ class TC_shortcut {
     }
 
     auto get_name() const -> std::string {
-        return "TC-shortcut";
+        return "tc-shortcut";
     }
 
     auto get_cap() const -> size_t {