summaryrefslogtreecommitdiffstats
path: root/searchlib/src/tests
Commit message (Collapse)AuthorAgeFilesLines
* Take dead elements into account (and subtract them) when calculating how ↵Geir Storli2021-03-011-4/+29
| | | | | | | | | | | | many arrays to allocate in a datastore buffer. This avoids a problem were allocated memory can accumulate over time in components using an ArrayStore. If all documents in an array attribute vector changes from one value class to another, all elements in the buffers of the previous value class are marked dead. Those buffers will eventually be compacted. Without this fix the wanted size of the resulting compacted buffer is calculated too high, and we allocate memory we are not going to use. If we move to yet another value class later, the same problem occurs again and more memory is allocated.
* Merge pull request #16690 from vespa-engine/geirst/refactor-datastore-classesGeir Storli2021-02-261-1/+1
|\ | | | | Refactor datastore classes
| * Rename 'active' buffer to 'primary' buffer in DataStoreBase.Geir Storli2021-02-261-1/+1
| | | | | | | | | | | | This should avoid confusion with the buffer state ACTIVE. Each buffer type has exactly one primary buffer that is used for allocations when no slots in free lists exist. A buffer type will typically use several buffers to store data, all of them being on state ACTIVE when in use.
* | Remove includes no longer necessary.Henning Baldersheim2021-02-258-33/+8
| |
* | Avoid including singlenumericattribute.hppHenning Baldersheim2021-02-253-7/+4
|/
* Eliminate array bounds warning.Tor Egge2021-02-211-0/+5
|
* Merge pull request #16495 from vespa-engine/arnej/add-size-literalsArne H Juul2021-02-1619-61/+80
|\ | | | | add literals for sizes in KB, MB, and GB
| * use size literals in searchlibArne Juul2021-02-1519-61/+80
| |
* | Use mmap file allocator in dense tensor attribute if huge flag isTor Egge2021-02-152-1/+25
|/ | | | set in config.
* Add alloc_aligned member function.Tor Egge2021-02-131-2/+1
|
* Forward declare MemoryAllocator in vespalib/util/alloc.h.Tor Egge2021-02-131-0/+1
|
* Less asserts in header files.Henning Baldersheim2021-02-132-0/+2
|
* Add unit tests for active boundary checks.Tor Egge2021-02-123-0/+36
|
* Fix onhold accounting for SerializedFastValueAttribute.Tor Egge2021-02-051-0/+15
|
* Improve handling of changed alloc config.Tor Egge2021-02-021-0/+20
|
* Use vespalib::duration for timeoutsHenning Baldersheim2021-01-311-0/+1
|
* libzstd 1.4.7 needs more bytes for compressed data.Tor Egge2021-01-211-1/+3
|
* Move idestructorcallback.h to vespalib module and ↵Henning Baldersheim2021-01-143-4/+4
| | | | search::IDestructorCallback to vespalib::IdestructorCallback.
* Reduce runtime by using 500 -> 300 children.Henning Baldersheim2021-01-141-11/+14
|
* - Add the cost to expensive tests.Henning Baldersheim2021-01-133-3/+3
| | | | - Use less resources on the fnet_parallel_rpc test bu just using as many threads as you have cores.
* Start expensive tests earlier.Henning Baldersheim2021-01-133-3/+3
|
* Have the BufferType::_emptyEntry be static.Henning Baldersheim2021-01-102-0/+2
| | | | Use an array of buffer types in the array class.
* make check_with_distance_threshold methodArne Juul2021-01-081-8/+16
|
* add distanceThreshold option for nearestNeighbor operatorArne Juul2021-01-087-13/+81
|
* Merge pull request #15913 from ↵Arne H Juul2021-01-081-1/+1
|\ | | | | | | | | vespa-engine/havardpe/use-string-ids-as-tensor-labels Havardpe/use string ids as tensor labels
| * use string ids as tensor labelsHåvard Pettersen2021-01-051-1/+1
| |
* | Stop fusion when closing flush engine.Tor Egge2021-01-061-3/+3
| |
* | Detect NUL byte in word when inverting field. Truncate the word andTor Egge2021-01-061-0/+49
| | | | | | | | drop it if the reduced length is zero.
* | Add low level support for stopping a running disk index fusion.Tor Egge2021-01-051-10/+71
|/
* Use stricter version of dynamic_cast.Tor Egge2021-01-041-1/+1
|
* move class to the only place it is usedArne Juul2020-12-101-1/+0
|
* cleanup old includesArne Juul2020-12-081-1/+0
|
* Deinline destructor.Tor Egge2020-12-061-0/+4
|
* Some -std=c++20a fixes:Arne Juul2020-12-061-2/+2
| | | | | | | add noexcept de-inline destructor avoid comparison warning (and revert changes already made in another branch)
* Add noexcept specifiers to non-throwing constructors and operators.Tor Egge2020-12-051-2/+2
|
* Avoid bitwise or of different enum types.Tor Egge2020-12-042-4/+4
|
* Eliminate inlining warnings.Tor Egge2020-12-041-0/+4
|
* getTensor with mutable view no longer neededArne Juul2020-12-031-3/+2
|
* Make it compatible with c++2aHenning Baldersheim2020-12-021-1/+1
|
* use SimpleValue instead of EngineOrFactoryArne Juul2020-12-0211-51/+59
|
* Merge pull request #15484 from ↵Arne H Juul2020-12-011-3/+3
|\ | | | | | | | | vespa-engine/arnej/add-serialized_fast_value_attribute Arnej/add serialized fast value attribute
| * Merge branch 'master' into arnej/add-serialized_fast_value_attributeArne Juul2020-11-275-38/+43
| |\ | | | | | | | | | | | | | | | | | | fix trivial conflict. Conflicts: searchlib/src/tests/features/tensor/tensor_test.cpp
| * | test SerializedFastValueAttribute insteadArne Juul2020-11-271-3/+3
| | |
| * | add SerializedFastValueAttributeArne Juul2020-11-261-3/+3
| | |
* | | Misc cleanup of includes and code healthHenning Baldersheim2020-12-011-2/+4
| |/ |/|
* | avoid using vespalib::tensor directlyArne Juul2020-11-271-6/+14
| | | | | | | | | | | | | | * do not construct DenseTensor instances directly * instead, use TensorSpec::from_expr to make a spec * and factory.from_spec() to make a value * also, use value codec directly to encode a value
* | use factory.from_spec() instead of tensor::makeTensor<T>Arne Juul2020-11-271-29/+29
| |
* | cleanup unused includeArne Juul2020-11-273-3/+0
|/
* track CellType moveArne Juul2020-11-203-8/+9
|
* Name the threads so it is easier to see who is doing what.Henning Baldersheim2020-11-124-12/+21
|