Updated benchmarks after some small optimizations to reduce branch misprediction

Author: p-ranav

README.md

@@ -47,17 +47,17 @@ Operating System: macOS Catalina version 10.15.3
 
 | Dataset | File Size | Rows | Cols | Time |
 |:---     |       ---:|  ---:|  ---:|  ---:|
-| [Denver Crime Data](https://www.kaggle.com/paultimothymooney/denver-crime-data) | 111 MB | 479,100 | 19 | 0.198s |
-| [AirBnb Paris Listings](https://www.kaggle.com/juliatb/airbnb-paris) | 196 MB | 141,730 | 96 | 0.345s |
-| [2015 Flight Delays and Cancellations](https://www.kaggle.com/usdot/flight-delays) | 574 MB | 5,819,079 | 31 | 0.994s |
-| [StackLite: Stack Overflow questions](https://www.kaggle.com/stackoverflow/stacklite) | 870 MB | 17,203,824 | 7 | 1.547s |
-| [Used Cars Dataset](https://www.kaggle.com/austinreese/craigslist-carstrucks-data) | 1.4 GB | 539,768 | 25 | 2.381s |
-| [Title-Based Semantic Subject Indexing](https://www.kaggle.com/hsrobo/titlebased-semantic-subject-indexing) | 3.7 GB | 12,834,026 | 4 | 6.965s|
-| [Bitcoin tweets - 16M tweets](https://www.kaggle.com/alaix14/bitcoin-tweets-20160101-to-20190329) | 4 GB | 47,478,748 | 9 | 7.945s |
-| [DDoS Balanced Dataset](https://www.kaggle.com/devendra416/ddos-datasets) | 6.3 GB | 12,794,627 | 85 | 13.578s |
-| [Seattle Checkouts by Title](https://www.kaggle.com/city-of-seattle/seattle-checkouts-by-title) | 7.1 GB | 34,892,623 | 11 | 15.350s |
-| [SHA-1 password hash dump](https://www.kaggle.com/urvishramaiya/have-i-been-pwnd) | 11 GB | 2,62,974,241 | 2 | 22.069s |
-| [DOHUI NOH scaled_data](https://www.kaggle.com/seaa0612/scaled-data) | 16 GB | 496,782 | 3213 | 34.923s |
+| [Denver Crime Data](https://www.kaggle.com/paultimothymooney/denver-crime-data) | 111 MB | 479,100 | 19 | 0.174s |
+| [AirBnb Paris Listings](https://www.kaggle.com/juliatb/airbnb-paris) | 196 MB | 141,730 | 96 | 0.289s |
+| [2015 Flight Delays and Cancellations](https://www.kaggle.com/usdot/flight-delays) | 574 MB | 5,819,079 | 31 | 1.047s |
+| [StackLite: Stack Overflow questions](https://www.kaggle.com/stackoverflow/stacklite) | 870 MB | 17,203,824 | 7 | 1.505s |
+| [Used Cars Dataset](https://www.kaggle.com/austinreese/craigslist-carstrucks-data) | 1.4 GB | 539,768 | 25 | 1.979s |
+| [Title-Based Semantic Subject Indexing](https://www.kaggle.com/hsrobo/titlebased-semantic-subject-indexing) | 3.7 GB | 12,834,026 | 4 | 5.929s|
+| [Bitcoin tweets - 16M tweets](https://www.kaggle.com/alaix14/bitcoin-tweets-20160101-to-20190329) | 4 GB | 47,478,748 | 9 | 7.040s |
+| [DDoS Balanced Dataset](https://www.kaggle.com/devendra416/ddos-datasets) | 6.3 GB | 12,794,627 | 85 | 12.648s |
+| [Seattle Checkouts by Title](https://www.kaggle.com/city-of-seattle/seattle-checkouts-by-title) | 7.1 GB | 34,892,623 | 11 | 12.883s |
+| [SHA-1 password hash dump](https://www.kaggle.com/urvishramaiya/have-i-been-pwnd) | 11 GB | 2,62,974,241 | 2 | 19.505s |
+| [DOHUI NOH scaled_data](https://www.kaggle.com/seaa0612/scaled-data) | 16 GB | 496,782 | 3213 | 32.780s |
 
 ## API
 

benchmark/csv-game/README.md

@@ -36,5 +36,5 @@ Operating System: macOS Catalina version 10.15.3
 
 | Test | Time |
 | --- | --- |
-| fieldcount | 47.2 ms ± 1.0 ms |
-| csv_count | 98.7 ms ± 1.9 ms |
+| fieldcount | 36.2 ms ± 3.1 ms |
+| csv_count | 92.5 ms ± 1.8 ms |

benchmark/csv-game/csv_count.cpp

@@ -18,14 +18,11 @@ int main(int argc, char **argv) {
     for (const auto row : csv) {
       size_t col{0};
       for (const auto cell : row) {
-        cell_value.clear();
-        cell.read_raw_value(cell_value);
-        std::cout << "Cell value: " << cell_value << "\n";
         col += 1;
         if (col == column_index) {
-          // cell.read_raw_value(cell_value);
+          cell.read_raw_value(cell_value);
           sum += std::stoi(cell_value);
-          // cell_value.clear();
+          cell_value.clear();
         }
       }
     }

include/csv2/reader.hpp

@@ -165,10 +165,10 @@ class Reader {
         size_t last_quote_location = 0;
         bool quote_opened = false;
         for (auto i = current_; i < end_; i++) {
+          current_ = i;
           if (buffer_[i] == delimiter::value && !quote_opened) {
             // actual delimiter
             // end of cell
-            current_ = i;
             cell.end_ = current_;
             cell.escaped_ = escaped;
             return cell;
@@ -179,22 +179,10 @@ class Reader {
                 quote_opened = true;
                 last_quote_location = i;
               } else {
-                // quote previously opened for this cell
-                // check last quote location
-                if (last_quote_location == i - 1) {
-                  // previous character was quote too!
-                  escaped = true;
-                } else {
-                  last_quote_location = i;
-                  if (i + 1 < end_ && buffer_[i + 1] == delimiter::value) {
-                    quote_opened = false;
-                  }
-                }
+                escaped = (last_quote_location == i - 1);
+                last_quote_location += (i - last_quote_location) * size_t(!escaped);
+                quote_opened = escaped || (buffer_[i + 1] != delimiter::value);
               }
-              current_ = i;
-            } else {
-              // Not delimiter or quote
-              current_ = i;
             }
           }
         }
@@ -236,8 +224,7 @@ class Reader {
               static_cast<const char *>(memchr(&buffer_[start_], '\n', (buffer_size_ - start_)))) {
         end_ = start_ + (ptr - &buffer_[start_]);
         result.end_ = end_;
-        if (end_ + 1 < buffer_size_)
-          start_ = end_ + 1;
+        start_ = end_ + 1;
       } else {
         // last row
         end_ = buffer_size_;