| 1 | #include <iostream> |
| 2 | #include <iomanip> |
| 3 | #include <city.h> |
| 4 | #include <Common/Stopwatch.h> |
| 5 | #include <Common/SipHash.h> |
| 6 | #include <IO/ReadBufferFromFileDescriptor.h> |
| 7 | #include <IO/ReadHelpers.h> |
| 8 | #include "config_core.h" |
| 9 | #if USE_SSL |
| 10 | # include <openssl/md5.h> |
| 11 | #endif |
| 12 | |
| 13 | |
| 14 | int main(int, char **) |
| 15 | { |
| 16 | using Strings = std::vector<std::string>; |
| 17 | using Hashes = std::vector<char>; |
| 18 | Strings strings; |
| 19 | size_t rows = 0; |
| 20 | size_t bytes = 0; |
| 21 | |
| 22 | { |
| 23 | Stopwatch watch; |
| 24 | |
| 25 | DB::ReadBufferFromFileDescriptor in(STDIN_FILENO); |
| 26 | |
| 27 | while (!in.eof()) |
| 28 | { |
| 29 | strings.push_back(std::string()); |
| 30 | DB::readEscapedString(strings.back(), in); |
| 31 | DB::assertChar('\n', in); |
| 32 | bytes += strings.back().size() + 1; |
| 33 | } |
| 34 | |
| 35 | watch.stop(); |
| 36 | rows = strings.size(); |
| 37 | std::cerr << std::fixed << std::setprecision(2) |
| 38 | << "Read " << rows << " rows, " << bytes / 1000000.0 << " MB" |
| 39 | << ", elapsed: " << watch.elapsedSeconds() |
| 40 | << " (" << rows / watch.elapsedSeconds() << " rows/sec., " << bytes / 1000000.0 / watch.elapsedSeconds() << " MB/sec.)" |
| 41 | << std::endl; |
| 42 | } |
| 43 | |
| 44 | Hashes hashes(16 * rows); |
| 45 | |
| 46 | { |
| 47 | Stopwatch watch; |
| 48 | |
| 49 | for (size_t i = 0; i < rows; ++i) |
| 50 | { |
| 51 | *reinterpret_cast<UInt64*>(&hashes[i * 16]) = CityHash_v1_0_2::CityHash64(strings[i].data(), strings[i].size()); |
| 52 | } |
| 53 | |
| 54 | watch.stop(); |
| 55 | |
| 56 | UInt64 check = CityHash_v1_0_2::CityHash64(hashes.data(), hashes.size()); |
| 57 | |
| 58 | std::cerr << std::fixed << std::setprecision(2) |
| 59 | << "CityHash64 (check = " << check << ")" |
| 60 | << ", elapsed: " << watch.elapsedSeconds() |
| 61 | << " (" << rows / watch.elapsedSeconds() << " rows/sec., " << bytes / 1000000.0 / watch.elapsedSeconds() << " MB/sec.)" |
| 62 | << std::endl; |
| 63 | } |
| 64 | |
| 65 | /* { |
| 66 | Stopwatch watch; |
| 67 | |
| 68 | std::vector<char> seed(16); |
| 69 | |
| 70 | for (size_t i = 0; i < rows; ++i) |
| 71 | { |
| 72 | sipHash( |
| 73 | reinterpret_cast<unsigned char *>(&hashes[i * 16]), |
| 74 | reinterpret_cast<const unsigned char *>(strings[i].data()), |
| 75 | strings[i].size(), |
| 76 | reinterpret_cast<const unsigned char *>(seed.data())); |
| 77 | } |
| 78 | |
| 79 | watch.stop(); |
| 80 | |
| 81 | UInt64 check = CityHash_v1_0_2::CityHash64(hashes.data(), hashes.size()); |
| 82 | |
| 83 | std::cerr << std::fixed << std::setprecision(2) |
| 84 | << "SipHash (check = " << check << ")" |
| 85 | << ", elapsed: " << watch.elapsedSeconds() |
| 86 | << " (" << rows / watch.elapsedSeconds() << " rows/sec., " << bytes / 1000000.0 / watch.elapsedSeconds() << " MB/sec.)" |
| 87 | << std::endl; |
| 88 | }*/ |
| 89 | |
| 90 | { |
| 91 | Stopwatch watch; |
| 92 | |
| 93 | for (size_t i = 0; i < rows; ++i) |
| 94 | { |
| 95 | SipHash hash; |
| 96 | hash.update(strings[i].data(), strings[i].size()); |
| 97 | hash.get128(&hashes[i * 16]); |
| 98 | } |
| 99 | |
| 100 | watch.stop(); |
| 101 | |
| 102 | UInt64 check = CityHash_v1_0_2::CityHash64(hashes.data(), hashes.size()); |
| 103 | |
| 104 | std::cerr << std::fixed << std::setprecision(2) |
| 105 | << "SipHash, stream (check = " << check << ")" |
| 106 | << ", elapsed: " << watch.elapsedSeconds() |
| 107 | << " (" << rows / watch.elapsedSeconds() << " rows/sec., " << bytes / 1000000.0 / watch.elapsedSeconds() << " MB/sec.)" |
| 108 | << std::endl; |
| 109 | } |
| 110 | |
| 111 | #if USE_SSL |
| 112 | { |
| 113 | Stopwatch watch; |
| 114 | |
| 115 | for (size_t i = 0; i < rows; ++i) |
| 116 | { |
| 117 | MD5_CTX state; |
| 118 | MD5_Init(&state); |
| 119 | MD5_Update(&state, reinterpret_cast<const unsigned char *>(strings[i].data()), strings[i].size()); |
| 120 | MD5_Final(reinterpret_cast<unsigned char *>(&hashes[i * 16]), &state); |
| 121 | } |
| 122 | |
| 123 | watch.stop(); |
| 124 | |
| 125 | UInt64 check = CityHash_v1_0_2::CityHash64(hashes.data(), hashes.size()); |
| 126 | |
| 127 | std::cerr << std::fixed << std::setprecision(2) |
| 128 | << "MD5 (check = " << check << ")" |
| 129 | << ", elapsed: " << watch.elapsedSeconds() |
| 130 | << " (" << rows / watch.elapsedSeconds() << " rows/sec., " << bytes / 1000000.0 / watch.elapsedSeconds() << " MB/sec.)" |
| 131 | << std::endl; |
| 132 | } |
| 133 | #endif |
| 134 | |
| 135 | return 0; |
| 136 | } |
| 137 | |