Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Update Verizon Media copyright notices. | gjoranv | 2021-10-07 | 2 | -2/+2 |
| | |||||
* | all scalars must be double | Håvard Pettersen | 2021-03-04 | 1 | -2/+3 |
| | | | | and all operation results must be at least float | ||||
* | Merge pull request #16788 from ↵ | Håvard Pettersen | 2021-03-04 | 1 | -2/+2 |
|\ | | | | | | | | | vespa-engine/arnej/move-normalize-to-reference-operations Arnej/move normalize to reference operations | ||||
| * | no need to normalize output from reference operations anymore | Arne Juul | 2021-03-04 | 1 | -2/+2 |
| | | |||||
* | | send result type as first arg to all make_instruction methods | Arne Juul | 2021-03-04 | 1 | -1/+2 |
|/ | |||||
* | use CellTypeUtils::list_types to loop over possible cell types in tests | Arne Juul | 2021-03-03 | 1 | -3/+2 |
| | |||||
* | use SmallVector in GenericReduce | Arne Juul | 2021-03-01 | 1 | -4/+4 |
| | |||||
* | adjust param repo add options and remove seq_bias | Håvard Pettersen | 2021-02-05 | 1 | -2/+2 |
| | | | | GenSpec is now implicitly convertible to TensorSpec | ||||
* | Merge pull request #16347 from vespa-engine/arnej/genspec-review-followup | Arne H Juul | 2021-02-03 | 1 | -3/+5 |
|\ | | | | | followup on GenSpec after review | ||||
| * | followup on GenSpec after review | Arne Juul | 2021-02-02 | 1 | -3/+5 |
| | | | | | | | | | | | | * avoid changing layouts in-place, call cpy() first * do cells_float() vs cells_double() more equally * minor cosmetic fixes | ||||
* | | Use noexcept specifier for simple lambdas. | Tor Egge | 2021-02-02 | 1 | -1/+1 |
|/ | |||||
* | copy gen spec to avoid sticky changes | Håvard Pettersen | 2021-02-01 | 1 | -2/+2 |
| | |||||
* | use GenSpec in generic_reduce_test | Arne Juul | 2021-02-01 | 1 | -28/+31 |
| | |||||
* | forward or ignore index in relevant mixed tensor reduce cases | Håvard Pettersen | 2021-01-19 | 1 | -1/+4 |
| | |||||
* | remove immediate API | Arne Juul | 2020-12-03 | 1 | -23/+0 |
| | |||||
* | put aggregator before dimensions in reference reduce() | Arne Juul | 2020-11-26 | 1 | -11/+11 |
| | |||||
* | add normalize() method to tensorspec | Arne Juul | 2020-11-25 | 1 | -9/+4 |
| | |||||
* | move spec normalization out from reference operation to unit test | Arne Juul | 2020-11-25 | 1 | -4/+9 |
| | |||||
* | move reference operations to a common place | Arne Juul | 2020-11-25 | 1 | -33/+5 |
| | | | | | | | * lift reference implementations from unit tests * write a reference TensorCreate implementation * use these from unit tests * some minor cosmetic changes with structured bindings | ||||
* | Merge pull request #15012 from ↵ | Arne H Juul | 2020-10-22 | 1 | -8/+6 |
|\ | | | | | | | | | vespa-engine/havardpe/better-dense-plan-for-generic-reduce improve generic dense reduce with more robust cell ordering | ||||
| * | improve generic dense reduce with more robust cell ordering | Håvard Pettersen | 2020-10-22 | 1 | -8/+6 |
| | | | | | | | | | | | | | | - unroll reduce all loop - extend benchmark with optimized instructions (to catch slow ones) - drop SimpleTensorEngine from benchmark (always very slow) - tweak API on templated aggregators | ||||
* | | add unit tests with FastValueBuilderFactory also | Arne Juul | 2020-10-22 | 1 | -5/+15 |
|/ | |||||
* | test immediate forms | Arne Juul | 2020-10-19 | 1 | -0/+24 |
| | |||||
* | reduce empty mixed to dense must add cells with default value | Håvard Pettersen | 2020-10-16 | 1 | -1/+5 |
| | |||||
* | allow interpreted function to use new generic operations | Håvard Pettersen | 2020-10-12 | 1 | -1/+1 |
| | | | | update conformance test to also test with SimpleValue and FastValue | ||||
* | generic reduce | Håvard Pettersen | 2020-10-02 | 2 | -0/+118 |