summaryrefslogtreecommitdiff
path: root/benchmarks
Commit message (Collapse)AuthorAgeFilesLines
* STY: Add blank line for PEP8.Charles Harris2017-09-121-0/+1
| | | [ci skip]
* [MAINT] Add block benchmarksJamie Townsend2017-09-121-0/+88
| | | | Note I've parameterised the benchmarks to run at three sizes.
* MAINT/DOC: Use builtin when np.{x} is builtins.{x}.Eric Wieser2017-08-053-4/+4
| | | | | | | This is the case for x in {int, bool, str, float, complex, object}. Using the np.{x} version is deceptive as it suggests that there is a difference. This change doesn't affect any external behaviour. The `long` type is missing in python 3, so np.long is still useful
* BENCH: Add some basic ma benchmarksEric Wieser2017-04-121-0/+93
|
* MAINT: Move ma benchmarks to new fileEric Wieser2017-04-112-15/+20
|
* BENCH: use initialized memory for count_nonzero benchmarkJulian Taylor2017-03-251-2/+2
| | | | | For the bool case uninitialized memory can randomly go into the slow path handling values different than 0 or 1.
* BENCH: add benchmarks for operations with temporariesJulian Taylor2017-02-241-0/+20
|
* Merge pull request #8629 from eric-wieser/speedup-indicesMarten van Kerkwijk2017-02-191-0/+5
|\ | | | | ENH: Improve the efficiency of indices
| * TST: add a benchmark for indicesEric Wieser2017-02-181-0/+5
| |
* | MAINT: ensure benchmark suite is importable on old numpy versionsPauli Virtanen2017-02-181-1/+1
|/
* TST: add extended packbits testsJulian Taylor2017-01-121-0/+32
| | | | | Larger data to account for future vectorization. Also add benchmarks for packbits and unpackbits
* DOC: change Numpy to NumPy in remaining filesPierre de Buyl2016-09-061-6/+6
| | | | the files in doc/ and numpy/ were covered in previous commits
* ENH: add inplace cases to fast ufunc loop macrosJulian Taylor2016-09-011-0/+39
| | | | | | | | Both gcc and clang don't automatically specialize the inplace case, so add extra conditions to the loop macros to get the compilers to emit decent code. Without them inplace code ends up much slower than the out of place code.
* ENH: added axis param for np.count_nonzerogfyoung2016-08-042-3/+24
| | | | Closes gh-391.
* BENCH: add more sort benchmarksJulian Taylor2016-07-291-0/+28
| | | | | benchmark for random, all equal elements, many equal elements and median of 3 worst case
* BENCH: add correlate/convolve benchmarks.Ralf Gommers2016-05-281-27/+44
| | | | Related to gh-5978, which makes significant changes to these functions.
* ENH: make memmap file in temporary directoryMatthew Brett2016-02-141-1/+10
| | | | | | Memmap test can fail if run from a directory for which we don't have write permission. Avoid NamedTemporaryFile because of problems reopening files on Windows.
* BUG: only benchmark complex256 if it existsMatthew Brett2016-02-141-1/+2
| | | | | Only benchmark complex256 type if numpy has it (MSVC windows builds do not).
* BENCH: Add some benchmarks for `matmul`.John Kirkham2016-01-161-0/+18
|
* BENCH: Reorganize existing benchmarks by the order they show up when run in ↵John Kirkham2016-01-161-11/+11
| | | | the benchmarking suite.
* Merge pull request #6969 from charris/bench-randintCharles Harris2016-01-121-0/+38
|\ | | | | ENH: Add benchmark tests for numpy.random.randint.
| * ENH: Add benchmark tests for numpy.random.randint.Charles Harris2016-01-071-0/+38
| | | | | | | | | | | | This add benchmarks randint. There is one set of benchmarks for the default dtype, 'l', that can be tracked back, and another set for the new dtypes 'bool', 'uint8', 'uint16', 'uint32', and 'uint64'.
* | BENCH: Perform benchmarking for the computation of `inner` on a matrix with ↵John Kirkham2016-01-081-0/+7
|/ | | | itself.
* BENCH: Add benchmarks between an array and its transpose that share data, ↵John Kirkham2016-01-061-0/+14
| | | | which should be optimized, versus the same configuration without shared data.
* BENCH: speed up benchmark suite import time; bump bench_ufunc timeout upwardPauli Virtanen2015-12-146-51/+107
| | | | | | The input data generation in benchmarks/common.py takes ~ 1s, and it is not used by most benchmarks. Generate it lazily instead, making sure the generation is done in the setup() routines.
* BENCH: allow benchmark suite to run on Python 3Pauli Virtanen2015-12-062-5/+9
|
* PERF: add 0d structured indexing benchmarkPauli Virtanen2015-10-181-0/+23
|
* WHT: break long lines + pep8Pauli Virtanen2015-08-244-14/+21
|
* DOC: mention asv in release notes and add a readmePauli Virtanen2015-08-221-0/+62
|
* MAINT: add airspeed velocity benchmarksPauli Virtanen2015-08-2212-0/+809
| | | | Converted from numpy-vbench suite
* REM: Remove benchmarks files.Charles Harris2013-02-285-146/+0
| | | | | | | | | | | The files are very basic, old benchmarks testing numpy against numeric and numarray. The competitors are almost defunct and, while benchmarks are awesome, we really need a more polished and complete framework that runs against the current competition. I think the early results from these benchmarks were posted, maybe even presented, and could be found in a search. Closes #3088 ;) So old a tuple parameter was used.
* 2to3: Updata `except Exception, msg:` syntax in files not in numpy/ .Charles Harris2013-02-271-1/+1
| | | | This should finish the updating of the exception syntax.
* ran reindentJarrod Millman2008-02-081-1/+0
|
* updating to us import numpy as np conventionJarrod Millman2007-12-154-25/+32
|
* ran reindent.py to cleanup whitespacesJarrod Millman2007-11-191-1/+1
|
* Make sorting benchmarks all sort integersCharles Harris2007-04-071-9/+9
|
* Refactor benchmarks.Stefan van der Walt2007-02-174-41/+90
|
* Fix to example.Travis Oliphant2006-07-181-1/+1
|
* Change occurrences of N.rand to N.random.rand.Stefan van der Walt2006-07-121-4/+4
|
* Remove dependency on _internal.py from pickles. Clean up Py_ssize_t usage. ↵Travis Oliphant2006-07-022-8/+8
| | | | Add .ctypes attribute for use with the ctypes module if it's available.
* Add simple indexing test.Travis Oliphant2006-06-281-0/+42
|
* Add benchmark for creating new arrays.Travis Oliphant2006-06-281-0/+9
|
* Add a function to retrieve a user-defined type number from the name of the ↵Travis Oliphant2006-06-271-0/+9
| | | | associated type-object
* Run reindent.py (script distributed with Python) over the source to remove ↵cookedm2006-03-101-5/+0
| | | | extraneous whitespace
* More documentation updatesedschofield2006-01-051-3/+3
|
* Changed all references to scipy to numpyTravis Oliphant2006-01-041-3/+3
|
* Added benchmarks folderTravis Oliphant2005-12-311-0/+36