aboutsummaryrefslogtreecommitdiffstats
path: root/eval
Commit message (Collapse)AuthorAgeFilesLines
* Add support for partial update remove operation where address is not fully ↵Geir Storli2020-11-172-9/+93
| | | | specified.
* drop un-needed optimizerArne Juul2020-11-121-1/+0
| | | | | * GenericLambda has the same code that the optimizer would insert, so there is no need for it in the new ValueBuilderFactory world
* Merge branch 'master' into arnej/add-join-with-number-optimizationArne Juul2020-11-1246-462/+494
|\ | | | | | | | | | | Conflicts: eval/CMakeLists.txt eval/src/vespa/eval/instruction/CMakeLists.txt
| * Merge pull request #15305 from vespa-engine/arnej/move-optimizersArne H Juul2020-11-1240-387/+408
| |\ | | | | | | Arnej/move optimizers
| | * test with FastValueBuilderFactory alsoArne Juul2020-11-125-23/+119
| | |
| | * drop un-needed includesArne Juul2020-11-125-39/+22
| | |
| | * avoid using namespace vespalib::tensorArne Juul2020-11-125-10/+5
| | |
| | * fix misleading commentsArne Juul2020-11-1216-24/+24
| | |
| | * move DenseCellRangeFunction to vespalib::eval namespaceArne Juul2020-11-123-22/+17
| | |
| | * also move dense_cell_range_functionArne Juul2020-11-126-4/+4
| | |
| | * move to vespalib::eval namespaceArne Juul2020-11-1218-182/+128
| | |
| | * move "keep as-is" optimizersArne Juul2020-11-1235-44/+50
| | | | | | | | | | | | | | | * from eval/tensor/dense to eval/instruction * minimal changes to track move in this commit
| * | Merge pull request #15286 from vespa-engine/arnej/more-specific-anonymous-namesArne H Juul2020-11-123-18/+18
| |\ \ | | |/ | |/| more specific names for View implementations
| | * more specific names for View implementationsArne Juul2020-11-113-18/+18
| | |
| * | Merge pull request #15278 from vespa-engine/arnej/move-dense-xwArne H Juul2020-11-119-58/+69
| |\ \ | | |/ | |/| Arnej/move dense xw
| | * complete the test matrixArne Juul2020-11-111-4/+6
| | | | | | | | | | | | | | | * test with prod_factory both optimized and not * test with old_engine both optimized and not
| | * move unit test for DenseXWProductFunctionArne Juul2020-11-113-16/+29
| | | | | | | | | | | | * and verify with FastValueBuilderFactory also
| | * still in same directory after moveArne Juul2020-11-111-1/+1
| | |
| | * move DenseXWProductFunction into vespalib::eval namespaceArne Juul2020-11-113-37/+33
| | |
| | * move dense_xw_product_function to eval/instructionArne Juul2020-11-117-7/+7
| | |
* | | update add_variants after reviewArne Juul2020-11-122-25/+25
| | |
* | | try adding a utility and more testsArne Juul2020-11-112-9/+52
| | |
* | | optimize join with number, with unit testArne Juul2020-11-117-2/+299
|/ /
* | Merge pull request #15267 from vespa-engine/toregge/avoid-clang-11-warningsArne H Juul2020-11-112-4/+4
|\ \ | | | | | | Avoid clang 11 warnings.
| * | Avoid clang 11 warnings.Tor Egge2020-11-102-4/+4
| |/
* | Merge pull request #15257 from ↵Arne H Juul2020-11-111-0/+1
|\ \ | |/ |/| | | | | vespa-engine/arnej/add-java-tensor-conformance-app Arnej/add java tensor conformance app
| * install binary for vespa-tensor-conformanceArne Juul2020-11-101-0/+1
| |
* | Include cstddef to get definition of size_tTor Egge2020-11-101-0/+1
| |
* | benchmark some forms of join with numberArne Juul2020-11-101-0/+18
|/
* minor refactoring:Arne Juul2020-11-093-3/+6
| | | | | | * have a static const empty_label in FastSparseMap (may be useful more places) * avoid re-computing num_mapped_dims needlessly
* pre-compute how to merge left and right sparse address componentsArne Juul2020-11-061-17/+28
|
* just reorder so optimized join implementations are togetherArne Juul2020-11-061-42/+42
|
* Merge pull request #15168 from vespa-engine/arnej/add-generic-lambdaArne H Juul2020-11-056-9/+211
|\ | | | | Arnej/add generic lambda
| * rename method counting typed nodesArne Juul2020-11-051-3/+3
| |
| * fix lambda traversing and extend testArne Juul2020-11-042-4/+18
| |
| * also add test of type exporting for inner lambdasArne Juul2020-11-041-0/+12
| |
| * use stash.create_uninitialized_arrayArne Juul2020-11-041-1/+1
| |
| * use as<TensorLambda>Arne Juul2020-11-031-1/+1
| |
| * GC unused includeArne Juul2020-11-031-1/+0
| |
| * test with simple factory alsoArne Juul2020-11-031-0/+5
| |
| * export nodes belonging to inner lambdasArne Juul2020-11-031-1/+7
| |
| * add GenericLambda instructionArne Juul2020-11-034-8/+174
| |
* | Merge pull request #15171 from vespa-engine/arnej/inline-sparse-no-overlapArne H Juul2020-11-044-49/+179
|\ \ | | | | | | Arnej/inline sparse no overlap
| * | just loop over subspacesArne Juul2020-11-041-32/+28
| | |
| * | move detect_type to its own header fileArne Juul2020-11-043-83/+77
| | |
| * | make a specific inline version of sparse no-overlap joinArne Juul2020-11-042-8/+148
| | |
* | | Merge pull request #15169 from vespa-engine/arnej/use-output-cell-type-in-reduceArne H Juul2020-11-041-3/+3
|\ \ \ | |/ / |/| | template aggregator on output cell type
| * | template aggregator on output cell typeArne Juul2020-11-031-3/+3
| |/
* | Merge pull request #15162 from vespa-engine/havardpe/optimize-tensor-functionArne H Juul2020-11-0310-70/+153
|\ \ | |/ |/| untangle factory-based optimization pipeline from DefaultTensorEngine
| * untangle factory-based optimization pipeline from DefaultTensorEngineHåvard Pettersen2020-11-0310-70/+153
| |