diff options
| author | Douglas Rumbaugh <dbr4@psu.edu> | 2023-07-24 18:51:12 -0400 |
|---|---|---|
| committer | Douglas Rumbaugh <dbr4@psu.edu> | 2023-07-24 18:51:12 -0400 |
| commit | ad95b1312b86426e149362166a560dea0ba920fe (patch) | |
| tree | f1ceed253329803b77aa6d095b4540d7ad402428 /benchmarks/include | |
| parent | 5ab408321dd45865a88fed71d11efe01dd7715d9 (diff) | |
| download | dynamic-extension-ad95b1312b86426e149362166a560dea0ba920fe.tar.gz | |
M-Tree benchmarks
Diffstat (limited to 'benchmarks/include')
| -rw-r--r-- | benchmarks/include/bench.h | 8 | ||||
| -rw-r--r-- | benchmarks/include/bench_utility.h | 32 |
2 files changed, 36 insertions, 4 deletions
diff --git a/benchmarks/include/bench.h b/benchmarks/include/bench.h index d90bc3f..12d0a7e 100644 --- a/benchmarks/include/bench.h +++ b/benchmarks/include/bench.h @@ -51,6 +51,8 @@ static bool insert_tput_bench(DE &de_index, std::fstream &file, size_t insert_cn if (applied_deletes < delete_cnt && delete_idx < delete_vec.size() && gsl_rng_uniform(g_rng) < delete_prop) { if constexpr (std::is_same_v<TreeMap, DE>) { de_index.erase_one(delete_vec[delete_idx++].key); + } else if constexpr (std::is_same_v<MTree, DE>) { + de_index.remove(delete_vec[delete_idx++]); } else { de_index.erase(delete_vec[delete_idx++]); } @@ -58,7 +60,11 @@ static bool insert_tput_bench(DE &de_index, std::fstream &file, size_t insert_cn } // insert the record; - de_index.insert(insert_vec[i]); + if constexpr (std::is_same_v<MTree, DE>) { + de_index.add(insert_vec[i]); + } else { + de_index.insert(insert_vec[i]); + } applied_inserts++; } auto insert_stop = std::chrono::high_resolution_clock::now(); diff --git a/benchmarks/include/bench_utility.h b/benchmarks/include/bench_utility.h index b728cbd..6610ab4 100644 --- a/benchmarks/include/bench_utility.h +++ b/benchmarks/include/bench_utility.h @@ -16,6 +16,7 @@ #include "shard/WIRS.h" #include "ds/BTree.h" #include "shard/VPTree.h" +#include "mtree.h" #include <cstdlib> #include <cstdio> @@ -38,7 +39,8 @@ typedef uint64_t weight_type; typedef de::WeightedRecord<key_type, value_type, weight_type> WRec; typedef de::Record<key_type, value_type> Rec; -typedef de::CosinePoint<double, 300> Word2VecRec; +const size_t W2V_SIZE = 300; +typedef de::CosinePoint<double, W2V_SIZE> Word2VecRec; typedef de::DynamicExtension<WRec, de::WSS<WRec>, de::WSSQuery<WRec>> ExtendedWSS; typedef de::DynamicExtension<Rec, de::TrieSpline<Rec>, de::TrieSplineRangeQuery<Rec>> ExtendedTSRQ; @@ -66,8 +68,25 @@ struct btree_key_extract { } }; -typedef tlx::BTree<key_type, btree_record, btree_key_extract> TreeMap; +struct cosine_similarity { + double operator()(const Word2VecRec &first, const Word2VecRec &second) const { + double prod = 0; + double asquared = 0; + double bsquared = 0; + + for (size_t i=0; i<W2V_SIZE; i++) { + prod += first.data[i] * second.data[i]; + asquared += first.data[i]*first.data[i]; + bsquared += second.data[i]*second.data[i]; + } + + return prod / std::sqrt(asquared * bsquared); + } +}; + +typedef tlx::BTree<key_type, btree_record, btree_key_extract> TreeMap; +typedef mt::mtree<Word2VecRec, cosine_similarity> MTree; static gsl_rng *g_rng; static std::set<WRec> *g_to_delete; @@ -347,13 +366,20 @@ static bool warmup(std::fstream &file, DE &extended_index, size_t count, if (delete_idx < delete_vec.size() && gsl_rng_uniform(g_rng) < delete_prop) { if constexpr (std::is_same_v<TreeMap, DE>) { extended_index.erase_one(delete_vec[delete_idx++].key); + } + else if constexpr (std::is_same_v<MTree, DE>) { + extended_index.remove(delete_vec[delete_idx++]); } else { extended_index.erase(delete_vec[delete_idx++]); } } // insert the record; - extended_index.insert(insert_vec[i]); + if constexpr (std::is_same_v<MTree, DE>) { + extended_index.add(insert_vec[i]); + } else { + extended_index.insert(insert_vec[i]); + } inserted++; if (progress) { |