Commit message (Collapse) | Author | Age | Files | Lines | |
---|---|---|---|---|---|
* | Use find_package to find gtest library. | Tor Egge | 2020-06-29 | 9 | -9/+18 |
| | |||||
* | ignore trivial dimensions | Håvard Pettersen | 2020-06-26 | 7 | -57/+75 |
| | |||||
* | GC old way of selecting functions by cell type | Arne Juul | 2020-06-25 | 1 | -27/+0 |
| | |||||
* | added 'erf' function | Håvard Pettersen | 2020-06-18 | 17 | -0/+24 |
| | |||||
* | observe the results of multiply-add across implementations | Håvard Pettersen | 2020-06-18 | 3 | -3/+56 |
| | | | | | | - add specific test showing diverging results when using FMA - re-write eval test to not fail due to FMA - re-write hitcollector test to not fail due to FMA | ||||
* | dense simple expand function | Håvard Pettersen | 2020-06-16 | 6 | -0/+321 |
| | |||||
* | pow 2/3 as map for dense tensors | Håvard Pettersen | 2020-06-15 | 14 | -57/+278 |
| | | | | | - let tensor_function::Node be an implementation detail and not an interface requirement. | ||||
* | use typify_invoke instead of dispatch_0/select_2 | Arne Juul | 2020-06-15 | 2 | -6/+9 |
| | |||||
* | Merge pull request #13579 from vespa-engine/havardpe/auto-detect-type-existence | Håvard Pettersen | 2020-06-15 | 1 | -4/+2 |
|\ | | | | | auto-detect actual type typify results | ||||
| * | auto-detect actual type typify results | Håvard Pettersen | 2020-06-13 | 1 | -4/+2 |
| | | |||||
* | | avoid deadlock when using compile cache with blocking executors | Håvard Pettersen | 2020-06-14 | 3 | -52/+98 |
| | | | | | | | | | | | | | | | | - avoid taking cache lock in compile task destructor - avoid holding cache lock while posting compile task - add unit test trying to provoke deadlock - bonus: fix blocking executor implementation (did not work with threads <= task limit) | ||||
* | | Merge pull request #13577 from ↵ | Henning Baldersheim | 2020-06-13 | 2 | -5/+6 |
|\ \ | |/ |/| | | | | | vespa-engine/havardpe/separate-result-locks-for-compile-cache use separate locks for result value propagation | ||||
| * | use separate locks for result value propagation | Håvard Pettersen | 2020-06-13 | 2 | -5/+6 |
| | | | | | | | | ... to avoid deadlock when using blocking executors | ||||
* | | Merge pull request #13571 from vespa-engine/arnej/use-typify-invoke-for-matmul | Håvard Pettersen | 2020-06-13 | 1 | -43/+16 |
|\ \ | | | | | | | use typify_invoke to select matmul implementation | ||||
| * | | replace template magic with if statement | Håvard Pettersen | 2020-06-13 | 1 | -28/+11 |
| | | | |||||
| * | | use typify_invoke to select matmul implementation | Arne Juul | 2020-06-12 | 1 | -43/+33 |
| | | | | | | | | | | | | | | | | | | * instead of a chain of select functions, use typify_invoke; use partial specialization on a struct to handle the double*double and float*float cases specially. | ||||
* | | | Merge pull request #13570 from vespa-engine/arnej/use-typify-invoke-for-select | Håvard Pettersen | 2020-06-13 | 8 | -45/+41 |
|\ \ \ | | | | | | | | | Arnej/use typify invoke for select | ||||
| * | | | typify_invoke instead of multiple select levels | Arne Juul | 2020-06-13 | 1 | -31/+16 |
| | | | | | | | | | | | | | | | | * follow pattern suggested in matmul PR | ||||
| * | | | use typify_invoke instead of select_2 | Arne Juul | 2020-06-12 | 3 | -6/+10 |
| | | | | |||||
| * | | | use typify_invoke instead of select_1 | Arne Juul | 2020-06-12 | 4 | -10/+16 |
| | | | | |||||
| * | | | use typify_invoke instead of dispatch_0 | Arne Juul | 2020-06-12 | 1 | -2/+3 |
| |/ / | |||||
* | / | add aggr typifier and use it | Håvard Pettersen | 2020-06-13 | 3 | -30/+33 |
| |/ |/| | |||||
* | | even better Inv handling, and testing | Håvard Pettersen | 2020-06-12 | 2 | -2/+2 |
| | | |||||
* | | better handling of Inv | Håvard Pettersen | 2020-06-12 | 2 | -2/+4 |
| | | |||||
* | | inline more operations | Håvard Pettersen | 2020-06-12 | 4 | -29/+241 |
|/ | |||||
* | use common apply loops | Håvard Pettersen | 2020-06-12 | 4 | -33/+28 |
| | |||||
* | Use function pointers in inline operations test. | Tor Egge | 2020-06-12 | 1 | -47/+47 |
| | |||||
* | use common code for simple map and number join | Håvard Pettersen | 2020-06-11 | 2 | -86/+22 |
| | |||||
* | common code for operation inlining | Håvard Pettersen | 2020-06-11 | 11 | -126/+390 |
| | | | | | | | | | | - add common code to make selecting the appropriate template function easier (vespa/vespalib/util/typify.h) - enable detection of lambda functions matching all low-level operations. (lookup_op1, lookup_op2) - add typifiers to decide which low-level operations should be inlined (TypifyOp1, TypifyOp2) - integrate into dense_simple_join as a pilot customer | ||||
* | Revert "Revert "When we pull in a cacheline, we should use it too."" | Henning Baldersheim | 2020-06-08 | 1 | -1/+1 |
| | |||||
* | Revert "When we pull in a cacheline, we should use it too." | Harald Musum | 2020-06-08 | 1 | -1/+1 |
| | |||||
* | - Stick to void * and byte offsets. | Henning Baldersheim | 2020-06-07 | 1 | -1/+1 |
| | | | | - Correct spelling error. | ||||
* | added some missing thats | Håvard Pettersen | 2020-05-29 | 1 | -2/+2 |
| | |||||
* | revert unintended change | Håvard Pettersen | 2020-05-29 | 1 | -1/+1 |
| | |||||
* | dense number join | Håvard Pettersen | 2020-05-29 | 7 | -4/+346 |
| | |||||
* | dense tensor lambda | Håvard Pettersen | 2020-05-28 | 7 | -56/+305 |
| | |||||
* | dense simple map | Håvard Pettersen | 2020-05-28 | 10 | -179/+230 |
| | |||||
* | use index lookup table with shared cache | Håvard Pettersen | 2020-05-27 | 6 | -22/+306 |
| | |||||
* | Move streaming operators to namespace searched by ADL. | Tor Egge | 2020-05-25 | 1 | -0/+4 |
| | |||||
* | simple dense join | Håvard Pettersen | 2020-05-22 | 11 | -195/+625 |
| | |||||
* | let compile cache use shared proton executor | Håvard Pettersen | 2020-05-19 | 3 | -1/+24 |
| | |||||
* | dense single reduce | Håvard Pettersen | 2020-05-07 | 11 | -92/+449 |
| | |||||
* | Avoid making copies of container elements. | Tor Egge | 2020-05-04 | 1 | -2/+2 |
| | |||||
* | include local file first | Håvard Pettersen | 2020-05-04 | 1 | -1/+1 |
| | |||||
* | multi-matmul | Håvard Pettersen | 2020-05-04 | 10 | -23/+520 |
| | |||||
* | fix PR comments | Håvard Pettersen | 2020-04-30 | 7 | -13/+8 |
| | |||||
* | added float cell range tests | Håvard Pettersen | 2020-04-30 | 1 | -0/+2 |
| | |||||
* | lambda peek optimizer | Håvard Pettersen | 2020-04-30 | 16 | -30/+880 |
| | |||||
* | Unwrap reference wrappers to avoid extra indirections via | Tor Egge | 2020-04-23 | 1 | -12/+12 |
| | | | | invalid memory. | ||||
* | truncate doubles when converting to labels/indexes | Håvard Pettersen | 2020-04-15 | 5 | -9/+9 |
| |