From bd74e27b28bd95267ce50d2e4b6f12b51d9b6aae Mon Sep 17 00:00:00 2001 From: Douglas Rumbaugh Date: Wed, 7 Feb 2024 17:23:23 -0500 Subject: Cleaned up shard files (except VPTree) Cleaned up shard implementations, fixed a few bugs, and set up some tests. There's still some work to be done in creating tests for the weighted sampling operations for the alias and aug btree shards. --- include/shard/AugBTree.h | 150 +++++++++++++---------------------------------- 1 file changed, 42 insertions(+), 108 deletions(-) (limited to 'include/shard/AugBTree.h') diff --git a/include/shard/AugBTree.h b/include/shard/AugBTree.h index be664ac..58bd098 100644 --- a/include/shard/AugBTree.h +++ b/include/shard/AugBTree.h @@ -16,28 +16,21 @@ #include #include -#include -#include -#include #include "framework/ShardRequirements.h" -#include "psu-ds/PriorityQueue.h" -#include "util/Cursor.h" #include "psu-ds/Alias.h" #include "psu-ds/BloomFilter.h" #include "util/bf_config.h" +#include "util/SortedMerge.h" using psudb::CACHELINE_SIZE; using psudb::BloomFilter; -using psudb::PriorityQueue; -using psudb::queue_record; using psudb::Alias; +using psudb::byte; namespace de { -thread_local size_t wirs_cancelations = 0; - template struct AugBTreeNode { struct AugBTreeNode *left, *right; @@ -54,108 +47,52 @@ private: typedef decltype(R::weight) W; public: - AugBTree(MutableBuffer* buffer) - : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0), m_root(nullptr) { - m_alloc_size = (buffer->get_record_count() * sizeof(Wrapped)) + (CACHELINE_SIZE - (buffer->get_record_count() * sizeof(Wrapped)) % CACHELINE_SIZE); - assert(m_alloc_size % CACHELINE_SIZE == 0); - m_data = (Wrapped*)std::aligned_alloc(CACHELINE_SIZE, m_alloc_size); - - m_bf = new BloomFilter(BF_FPR, buffer->get_tombstone_count(), BF_HASH_FUNCS); - - size_t offset = 0; - m_reccnt = 0; - auto base = buffer->get_data(); - auto stop = base + buffer->get_record_count(); - - std::sort(base, stop, std::less>()); - - while (base < stop) { - if (!(base->is_tombstone()) && (base + 1) < stop) { - if (base->rec == (base + 1)->rec && (base + 1)->is_tombstone()) { - base += 2; - wirs_cancelations++; - continue; - } - } else if (base->is_deleted()) { - base += 1; - continue; - } - - // FIXME: this shouldn't be necessary, but the tagged record - // bypass doesn't seem to be working on this code-path, so this - // ensures that tagged records from the buffer are able to be - // dropped, eventually. It should only need to be &= 1 - base->header &= 3; - m_data[m_reccnt++] = *base; - m_total_weight+= base->rec.weight; - - if (m_bf && base->is_tombstone()) { - m_tombstone_cnt++; - m_bf->insert(base->rec); - } - - base++; - } + AugBTree(BufferView buffer) + : m_data(nullptr) + , m_root(nullptr) + , m_reccnt(0) + , m_tombstone_cnt(0) + , m_group_size(0) + , m_alloc_size(0) + , m_node_cnt(0) + , m_bf(new BloomFilter(BF_FPR, buffer.get_tombstone_count(), BF_HASH_FUNCS)) + { + m_alloc_size = psudb::sf_aligned_alloc(CACHELINE_SIZE, + buffer.get_record_count() * + sizeof(Wrapped), + (byte**) &m_data); + + auto res = sorted_array_from_bufferview(std::move(buffer), m_data, m_bf); + m_reccnt = res.record_count; + m_tombstone_cnt = res.tombstone_count; if (m_reccnt > 0) { build_wirs_structure(); } } - AugBTree(AugBTree** shards, size_t len) - : m_reccnt(0), m_tombstone_cnt(0), m_total_weight(0), m_root(nullptr) { - std::vector>> cursors; - cursors.reserve(len); - - PriorityQueue> pq(len); - + AugBTree(std::vector shards) + : m_data(nullptr) + , m_root(nullptr) + , m_reccnt(0) + , m_tombstone_cnt(0) + , m_group_size(0) + , m_alloc_size(0) + , m_node_cnt(0) + , m_bf(nullptr) + { size_t attemp_reccnt = 0; size_t tombstone_count = 0; - - for (size_t i = 0; i < len; ++i) { - if (shards[i]) { - auto base = shards[i]->get_data(); - cursors.emplace_back(Cursor{base, base + shards[i]->get_record_count(), 0, shards[i]->get_record_count()}); - attemp_reccnt += shards[i]->get_record_count(); - tombstone_count += shards[i]->get_tombstone_count(); - pq.push(cursors[i].ptr, i); - } else { - cursors.emplace_back(Cursor>{nullptr, nullptr, 0, 0}); - } - } + auto cursors = build_cursor_vec(shards, &attemp_reccnt, &tombstone_count); m_bf = new BloomFilter(BF_FPR, tombstone_count, BF_HASH_FUNCS); + m_alloc_size = psudb::sf_aligned_alloc(CACHELINE_SIZE, + attemp_reccnt * sizeof(Wrapped), + (byte **) &m_data); - m_alloc_size = (attemp_reccnt * sizeof(Wrapped)) + (CACHELINE_SIZE - (attemp_reccnt * sizeof(Wrapped)) % CACHELINE_SIZE); - assert(m_alloc_size % CACHELINE_SIZE == 0); - m_data = (Wrapped*)std::aligned_alloc(CACHELINE_SIZE, m_alloc_size); - - while (pq.size()) { - auto now = pq.peek(); - auto next = pq.size() > 1 ? pq.peek(1) : queue_record>{nullptr, 0}; - if (!now.data->is_tombstone() && next.data != nullptr && - now.data->rec == next.data->rec && next.data->is_tombstone()) { - - pq.pop(); pq.pop(); - auto& cursor1 = cursors[now.version]; - auto& cursor2 = cursors[next.version]; - if (advance_cursor>(cursor1)) pq.push(cursor1.ptr, now.version); - if (advance_cursor>(cursor2)) pq.push(cursor2.ptr, next.version); - } else { - auto& cursor = cursors[now.version]; - if (!cursor.ptr->is_deleted()) { - m_data[m_reccnt++] = *cursor.ptr; - m_total_weight += cursor.ptr->rec.weight; - if (m_bf && cursor.ptr->is_tombstone()) { - ++m_tombstone_cnt; - if (m_bf) m_bf->insert(cursor.ptr->rec); - } - } - pq.pop(); - - if (advance_cursor>(cursor)) pq.push(cursor.ptr, now.version); - } - } + auto res = sorted_array_merge(cursors, m_data, m_bf); + m_reccnt = res.record_count; + m_tombstone_cnt = res.tombstone_count; if (m_reccnt > 0) { build_wirs_structure(); @@ -163,13 +100,12 @@ public: } ~AugBTree() { - if (m_data) free(m_data); + free(m_data); for (size_t i=0; i>); } size_t get_aux_memory_usage() { - return 0; + return (m_bf) ? m_bf->memory_usage() : 0; } size_t get_lower_bound(const K& key) const { @@ -364,7 +299,6 @@ private: Wrapped* m_data; std::vector m_alias; AugBTreeNode* m_root; - W m_total_weight; size_t m_reccnt; size_t m_tombstone_cnt; size_t m_group_size; -- cgit v1.2.3