diff options
author | Henning Baldersheim <balder@yahoo-inc.com> | 2023-12-12 15:44:52 +0100 |
---|---|---|
committer | GitHub <noreply@github.com> | 2023-12-12 15:44:52 +0100 |
commit | efc9e84a4678ce90574cdf3d1dba99b9e04c0a87 (patch) | |
tree | a36dd7c01f09fa82bc8cbbc41ee4e1279f80cf8f /searchlib/src | |
parent | 96e19ab8cf802a1ddf5087599201511c94ea8814 (diff) | |
parent | 81487d4843b65c66f2f14a3346f3b6b51b85e245 (diff) |
Merge pull request #29617 from vespa-engine/balder/wire-thread-bundle-to-executeinfo-and-requestcontext
Wire in thread bundle to execute info and request context.
Diffstat (limited to 'searchlib/src')
12 files changed, 62 insertions, 47 deletions
diff --git a/searchlib/src/tests/queryeval/blueprint/intermediate_blueprints_test.cpp b/searchlib/src/tests/queryeval/blueprint/intermediate_blueprints_test.cpp index eea6c898aa2..7d994ee4cd6 100644 --- a/searchlib/src/tests/queryeval/blueprint/intermediate_blueprints_test.cpp +++ b/searchlib/src/tests/queryeval/blueprint/intermediate_blueprints_test.cpp @@ -39,13 +39,16 @@ struct WeightOrder { }; struct RememberExecuteInfo : public MyLeaf { - ExecuteInfo executeInfo; + bool is_strict; + double hit_rate; - using MyLeaf::MyLeaf; + RememberExecuteInfo() : MyLeaf(), is_strict(false), hit_rate(0.0) {} + RememberExecuteInfo(FieldSpecBaseList fields) : MyLeaf(std::move(fields)), is_strict(false), hit_rate(0.0) {} void fetchPostings(const ExecuteInfo &execInfo) override { LeafBlueprint::fetchPostings(execInfo); - executeInfo = execInfo; + is_strict = execInfo.is_strict(); + hit_rate = execInfo.hit_rate(); } }; @@ -134,11 +137,11 @@ TEST("test And propagates updated histestimate") { EXPECT_EQUAL(3u, bp->childCnt()); for (uint32_t i = 0; i < bp->childCnt(); i++) { const auto & child = dynamic_cast<const RememberExecuteInfo &>(bp->getChild(i)); - EXPECT_EQUAL((i == 0), child.executeInfo.isStrict()); + EXPECT_EQUAL((i == 0), child.is_strict); } - EXPECT_EQUAL(1.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(0)).executeInfo.hitRate()); - EXPECT_EQUAL(1.0/250, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(1)).executeInfo.hitRate()); - EXPECT_EQUAL(1.0/(250*25), dynamic_cast<const RememberExecuteInfo &>(bp->getChild(2)).executeInfo.hitRate()); + EXPECT_EQUAL(1.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(0)).hit_rate); + EXPECT_EQUAL(1.0/250, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(1)).hit_rate); + EXPECT_EQUAL(1.0/(250*25), dynamic_cast<const RememberExecuteInfo &>(bp->getChild(2)).hit_rate); } TEST("test Or propagates updated histestimate") { @@ -154,12 +157,12 @@ TEST("test Or propagates updated histestimate") { EXPECT_EQUAL(4u, bp->childCnt()); for (uint32_t i = 0; i < bp->childCnt(); i++) { const auto & child = dynamic_cast<const RememberExecuteInfo &>(bp->getChild(i)); - EXPECT_TRUE(child.executeInfo.isStrict()); + EXPECT_TRUE(child.is_strict); } - EXPECT_EQUAL(1.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(0)).executeInfo.hitRate()); - EXPECT_APPROX(0.5, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(1)).executeInfo.hitRate(), 1e-6); - EXPECT_APPROX(0.5*3.0/5.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(2)).executeInfo.hitRate(), 1e-6); - EXPECT_APPROX(0.5*3.0*42.0/(5.0*50.0), dynamic_cast<const RememberExecuteInfo &>(bp->getChild(3)).executeInfo.hitRate(), 1e-6); + EXPECT_EQUAL(1.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(0)).hit_rate); + EXPECT_APPROX(0.5, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(1)).hit_rate, 1e-6); + EXPECT_APPROX(0.5*3.0/5.0, dynamic_cast<const RememberExecuteInfo &>(bp->getChild(2)).hit_rate, 1e-6); + EXPECT_APPROX(0.5*3.0*42.0/(5.0*50.0), dynamic_cast<const RememberExecuteInfo &>(bp->getChild(3)).hit_rate, 1e-6); } TEST("test And Blueprint") { diff --git a/searchlib/src/tests/queryeval/blueprint/mysearch.h b/searchlib/src/tests/queryeval/blueprint/mysearch.h index 6cfc7a04368..db7dd2adae6 100644 --- a/searchlib/src/tests/queryeval/blueprint/mysearch.h +++ b/searchlib/src/tests/queryeval/blueprint/mysearch.h @@ -58,10 +58,10 @@ public: if (!_isLeaf) { ok &= (_md == &md); } - for (size_t i = 0; i < _children.size(); ++i) { - MySearch *child = dynamic_cast<MySearch *>(_children[i].get()); - ok &= (child != 0); - if (child != 0) { + for (const auto & my_search : _children) { + auto *child = dynamic_cast<MySearch *>(my_search.get()); + ok &= (child != nullptr); + if (child != nullptr) { ok &= child->verifyAndInferImpl(md); } } @@ -80,8 +80,8 @@ public: } static bool verifyAndInfer(SearchIterator *search, MatchData &md) { - MySearch *self = dynamic_cast<MySearch *>(search); - if (self == 0) { + auto *self = dynamic_cast<MySearch *>(search); + if (self == nullptr) { return false; } else { return self->verifyAndInferImpl(md); diff --git a/searchlib/src/vespa/searchlib/attribute/attribute_weighted_set_blueprint.cpp b/searchlib/src/vespa/searchlib/attribute/attribute_weighted_set_blueprint.cpp index 42b2cca06f7..99d3ba3f7aa 100644 --- a/searchlib/src/vespa/searchlib/attribute/attribute_weighted_set_blueprint.cpp +++ b/searchlib/src/vespa/searchlib/attribute/attribute_weighted_set_blueprint.cpp @@ -163,7 +163,7 @@ AttributeWeightedSetBlueprint::createFilterSearch(bool strict, FilterConstraint) void AttributeWeightedSetBlueprint::fetchPostings(const queryeval::ExecuteInfo &execInfo) { - if (execInfo.isStrict()) { + if (execInfo.is_strict()) { for (auto * context : _contexts) { context->fetchPostings(execInfo); } diff --git a/searchlib/src/vespa/searchlib/attribute/imported_search_context.cpp b/searchlib/src/vespa/searchlib/attribute/imported_search_context.cpp index b49ec1bfc9e..b27a67e893b 100644 --- a/searchlib/src/vespa/searchlib/attribute/imported_search_context.cpp +++ b/searchlib/src/vespa/searchlib/attribute/imported_search_context.cpp @@ -308,7 +308,7 @@ void ImportedSearchContext::fetchPostings(const queryeval::ExecuteInfo &execInfo) { if (!_searchCacheLookup) { _target_search_context->fetchPostings(execInfo); - if (!_merger.merge_done() && (execInfo.isStrict() || (_target_attribute.getIsFastSearch() && execInfo.hitRate() > 0.01))) { + if (!_merger.merge_done() && (execInfo.is_strict() || (_target_attribute.getIsFastSearch() && execInfo.hit_rate() > 0.01))) { makeMergedPostings(_target_attribute.getIsFilter()); considerAddSearchCacheEntry(); } diff --git a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h index eefb2ea77d5..a291318e837 100644 --- a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h +++ b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.h @@ -318,7 +318,7 @@ bool StringPostingSearchContext<BaseSC, AttrT, DataT>::use_posting_lists_when_non_strict(const queryeval::ExecuteInfo& info) const { if (this->isFuzzy()) { - uint32_t exp_doc_hits = this->_docIdLimit * info.hitRate(); + uint32_t exp_doc_hits = this->_docIdLimit * info.hit_rate(); constexpr uint32_t fuzzy_use_posting_lists_doc_limit = 10000; /** * The above constant was derived after a query latency experiment with fuzzy matching @@ -441,7 +441,7 @@ NumericPostingSearchContext<BaseSC, AttrT, DataT>::use_posting_lists_when_non_st constexpr float lookup_match_constant = 5.0; constexpr float posting_list_merge_constant = 1.0; - uint32_t exp_doc_hits = this->_docIdLimit * info.hitRate(); + uint32_t exp_doc_hits = this->_docIdLimit * info.hit_rate(); float avg_values_per_document = static_cast<float>(this->_numValues) / static_cast<float>(this->_docIdLimit); float lookup_match_cost = exp_doc_hits * avg_values_per_document * lookup_match_constant; float posting_list_cost = this->estimated_hits_in_range() * posting_list_merge_constant; diff --git a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.hpp b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.hpp index 3ff5c749899..ecdfad433ee 100644 --- a/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.hpp +++ b/searchlib/src/vespa/searchlib/attribute/postinglistsearchcontext.hpp @@ -62,8 +62,7 @@ void PostingListSearchContextT<DataT>::fillArray() { for (auto it(_lowerDictItr); it != _upperDictItr; ++it) { - _merger.addToArray(PostingListTraverser<PostingStore>(_posting_store, - it.getData().load_acquire())); + _merger.addToArray(PostingListTraverser<PostingStore>(_posting_store, it.getData().load_acquire())); } _merger.merge(); } @@ -110,7 +109,7 @@ PostingListSearchContextT<DataT>::fetchPostings(const queryeval::ExecuteInfo & e // The threshold for when to use array merging is therefore 0.0025 (0.08 / 32). constexpr float threshold_for_using_array = 0.0025; if (!_merger.merge_done() && _uniqueValues >= 2u && this->_dictionary.get_has_btree_dictionary()) { - if (execInfo.isStrict() || use_posting_lists_when_non_strict(execInfo)) { + if (execInfo.is_strict() || use_posting_lists_when_non_strict(execInfo)) { size_t sum = estimated_hits_in_range(); if (sum < (_docIdLimit * threshold_for_using_array)) { _merger.reserveArray(_uniqueValues, sum); diff --git a/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp b/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp index 16a2ba64fa7..043b006d3bd 100644 --- a/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp +++ b/searchlib/src/vespa/searchlib/queryeval/blueprint.cpp @@ -628,10 +628,10 @@ IntermediateBlueprint::visitMembers(vespalib::ObjectVisitor &visitor) const void IntermediateBlueprint::fetchPostings(const ExecuteInfo &execInfo) { - double nextHitRate = execInfo.hitRate(); + double nextHitRate = execInfo.hit_rate(); for (size_t i = 0; i < _children.size(); ++i) { Blueprint & child = *_children[i]; - child.fetchPostings(ExecuteInfo::create(execInfo.isStrict() && inheritStrict(i), nextHitRate, execInfo)); + child.fetchPostings(ExecuteInfo::create(execInfo.is_strict() && inheritStrict(i), nextHitRate, execInfo)); nextHitRate = computeNextHitRate(child, nextHitRate, execInfo.use_estimate_for_fetch_postings()); } } diff --git a/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp b/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp index ac3dfcc3b20..858cb92331a 100644 --- a/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp +++ b/searchlib/src/vespa/searchlib/queryeval/executeinfo.cpp @@ -4,7 +4,7 @@ namespace search::queryeval { -const ExecuteInfo ExecuteInfo::TRUE(true, 1.0, nullptr, true, true); -const ExecuteInfo ExecuteInfo::FALSE(false, 1.0, nullptr, true, true); +const ExecuteInfo ExecuteInfo::TRUE(true, 1.0, nullptr, vespalib::ThreadBundle::trivial(), true, true); +const ExecuteInfo ExecuteInfo::FALSE(false, 1.0, nullptr, vespalib::ThreadBundle::trivial(), true, true); } diff --git a/searchlib/src/vespa/searchlib/queryeval/executeinfo.h b/searchlib/src/vespa/searchlib/queryeval/executeinfo.h index 01b3ddfd182..3300a2aea4d 100644 --- a/searchlib/src/vespa/searchlib/queryeval/executeinfo.h +++ b/searchlib/src/vespa/searchlib/queryeval/executeinfo.h @@ -3,6 +3,7 @@ #pragma once #include <vespa/vespalib/util/doom.h> +#include <vespa/vespalib/util/thread_bundle.h> namespace search::queryeval { @@ -12,26 +13,27 @@ namespace search::queryeval { */ class ExecuteInfo { public: - ExecuteInfo() noexcept : ExecuteInfo(false, 1.0, nullptr, true, true) { } - bool isStrict() const noexcept { return _strict; } + ExecuteInfo() noexcept : ExecuteInfo(false, 1.0, nullptr, vespalib::ThreadBundle::trivial(), true, true) { } + bool is_strict() const noexcept { return _strict; } bool create_postinglist_when_non_strict() const noexcept { return _create_postinglist_when_non_strict; } bool use_estimate_for_fetch_postings() const noexcept { return _use_estimate_for_fetch_postings; } - double hitRate() const noexcept { return _hitRate; } + double hit_rate() const noexcept { return _hitRate; } bool soft_doom() const noexcept { return _doom && _doom->soft_doom(); } - const vespalib::Doom * getDoom() const { return _doom; } + vespalib::ThreadBundle & thread_bundle() const noexcept { return _thread_bundle; } + static const ExecuteInfo TRUE; static const ExecuteInfo FALSE; static ExecuteInfo create(bool strict, const ExecuteInfo & org) noexcept { return create(strict, org._hitRate, org); } static ExecuteInfo create(bool strict, double hitRate, const ExecuteInfo & org) noexcept { - return {strict, hitRate, org.getDoom(), org.create_postinglist_when_non_strict(), org.use_estimate_for_fetch_postings()}; + return {strict, hitRate, org._doom, org.thread_bundle(), org.create_postinglist_when_non_strict(), org.use_estimate_for_fetch_postings()}; } - static ExecuteInfo create(bool strict, double hitRate, const vespalib::Doom * doom, bool postinglist_when_non_strict, - bool use_estimate_for_fetch_postings) noexcept + static ExecuteInfo create(bool strict, double hitRate, const vespalib::Doom * doom, vespalib::ThreadBundle & thread_bundle_in, + bool postinglist_when_non_strict, bool use_estimate_for_fetch_postings) noexcept { - return {strict, hitRate, doom, postinglist_when_non_strict, use_estimate_for_fetch_postings}; + return {strict, hitRate, doom, thread_bundle_in, postinglist_when_non_strict, use_estimate_for_fetch_postings}; } static ExecuteInfo createForTest(bool strict) noexcept { return createForTest(strict, 1.0); @@ -40,22 +42,24 @@ public: return createForTest(strict, hitRate, nullptr); } static ExecuteInfo createForTest(bool strict, double hitRate, const vespalib::Doom * doom) noexcept { - return create(strict, hitRate, doom, true, true); + return create(strict, hitRate, doom, vespalib::ThreadBundle::trivial(), true, true); } private: - ExecuteInfo(bool strict, double hitRate_in, const vespalib::Doom * doom, bool postinglist_when_non_strict, - bool use_estimate_for_fetch_postings) noexcept + ExecuteInfo(bool strict, double hitRate_in, const vespalib::Doom * doom, vespalib::ThreadBundle & thread_bundle_in, + bool postinglist_when_non_strict, bool use_estimate_for_fetch_postings) noexcept : _doom(doom), + _thread_bundle(thread_bundle_in), _hitRate(hitRate_in), _strict(strict), _create_postinglist_when_non_strict(postinglist_when_non_strict), _use_estimate_for_fetch_postings(use_estimate_for_fetch_postings) { } - const vespalib::Doom * _doom; - double _hitRate; - bool _strict; - bool _create_postinglist_when_non_strict; - bool _use_estimate_for_fetch_postings; + const vespalib::Doom * _doom; + vespalib::ThreadBundle & _thread_bundle; + double _hitRate; + bool _strict; + bool _create_postinglist_when_non_strict; + bool _use_estimate_for_fetch_postings; }; } diff --git a/searchlib/src/vespa/searchlib/queryeval/fake_requestcontext.h b/searchlib/src/vespa/searchlib/queryeval/fake_requestcontext.h index e536727169a..0122352656e 100644 --- a/searchlib/src/vespa/searchlib/queryeval/fake_requestcontext.h +++ b/searchlib/src/vespa/searchlib/queryeval/fake_requestcontext.h @@ -11,6 +11,7 @@ #include <vespa/searchlib/attribute/attributevector.h> #include <vespa/searchlib/attribute/attribute_blueprint_params.h> #include <vespa/vespalib/util/doom.h> +#include <vespa/vespalib/util/thread_bundle.h> #include <limits> namespace vespalib { class TestClock; } @@ -25,6 +26,7 @@ public: vespalib::steady_time hard=vespalib::steady_time::max()); ~FakeRequestContext() override; const vespalib::Doom & getDoom() const override { return _doom; } + vespalib::ThreadBundle & thread_bundle() const override { return vespalib::ThreadBundle::trivial(); } const attribute::IAttributeVector *getAttribute(const vespalib::string &name) const override { return _attributeContext ? _attributeContext->getAttribute(name) diff --git a/searchlib/src/vespa/searchlib/queryeval/irequestcontext.h b/searchlib/src/vespa/searchlib/queryeval/irequestcontext.h index f4c3ae547f0..ad25385be4b 100644 --- a/searchlib/src/vespa/searchlib/queryeval/irequestcontext.h +++ b/searchlib/src/vespa/searchlib/queryeval/irequestcontext.h @@ -8,7 +8,10 @@ namespace search::attribute { struct AttributeBlueprintParams; } namespace search::attribute { class IAttributeVector; } namespace vespalib::eval { struct Value; } -namespace vespalib { class Doom; } +namespace vespalib { + class Doom; + struct ThreadBundle; +} namespace search::queryeval { @@ -26,6 +29,10 @@ public: * @return time of soft doom. */ virtual const vespalib::Doom & getDoom() const = 0; + /** + * Provide an optional thread bundle that can be used for multithreading parts of the query. + */ + virtual vespalib::ThreadBundle & thread_bundle() const = 0; /** * Provide access to attributevectors diff --git a/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp b/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp index eb1d217ad6a..f0c75173671 100644 --- a/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp +++ b/searchlib/src/vespa/searchlib/queryeval/same_element_blueprint.cpp @@ -61,7 +61,7 @@ SameElementBlueprint::fetchPostings(const ExecuteInfo &execInfo) if (_terms.empty()) return; _terms[0]->fetchPostings(execInfo); double estimate = execInfo.use_estimate_for_fetch_postings() ? _terms[0]->hit_ratio() : _terms[0]->estimate(); - double hit_rate = execInfo.hitRate() * estimate; + double hit_rate = execInfo.hit_rate() * estimate; for (size_t i = 1; i < _terms.size(); ++i) { Blueprint & term = *_terms[i]; term.fetchPostings(ExecuteInfo::create(false, hit_rate, execInfo)); |