summaryrefslogtreecommitdiff
path: root/numpy
Commit message (Collapse)AuthorAgeFilesLines
...
| * | | | MAINT: use sse header macros for einsum sse activationJulian Taylor2013-05-252-3/+7
| | | | |
| * | | | ENH: vectorize sqrt ufunc using SSE2Julian Taylor2013-05-2510-176/+417
| | |/ / | |/| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | specialize the sqrt ufunc for float and double and vectorize it using SSE2. improves performance by 4/2 for float/double if one is not memory bound due to non-cached data. performance is always better on all tested machines (amd phenom X2, intel xeon 5xxx/7xxx, core2duo, corei7) This version will not set errno on invalid input, but numpy only checks the fpu flags so the behavior is the same. In principle the compiler could autovectorize it when setting ffast-math (for no errno) and specializing the loop for the vectorizable strides and giving it some hints (restrict, __builtin_assume_aligned, etc.), but its simpler and more reliable to simply vectorize it by hand.
* | | | MAINT: fix some compiler warningsJulian Taylor2013-05-2513-17/+19
| |/ / |/| | | | | | | | | | | implicit declarations, wrong declarations, unused variables and fixes a comparison typo bug in multiarraymodule.c
* | | Merge pull request #3324 from nouiz/prio_cmpseberg2013-05-233-40/+194
|\ \ \ | |/ / |/| | Make comparison function (gt, ge, ...) respect __array_priority__.
| * | code clean up. Remove code that was never executed.Frederic2013-05-231-14/+2
| | | | | | | | | | | | @seberg saw this.
| * | updated the comment.Frederic2013-05-171-6/+4
| | |
| * | use assert function instead of python keywordFrederic2013-05-171-42/+42
| | |
| * | Move computation inside an if to speed it up.Frederic2013-05-161-36/+35
| | |
| * | remove comparison not needed as this is check above and we break in that case.Frederic2013-05-161-4/+2
| | |
| * | fix comment style.Frederic2013-05-161-6/+10
| | |
| * | reorder code to remove not needed computation in some cases.Frederic2013-05-161-5/+19
| | |
| * | Add test for the array_priority for comparison function.Frederic2013-05-151-0/+146
| | |
| * | Commit from @seberg to make comparison use respect __array_priority__Frederic2013-05-152-2/+9
| | |
| * | Revert "Make comparison function (gt, ge, ...) respect __array_priority__."Frederic2013-05-151-44/+0
| | | | | | | | | | | | This reverts commit aa7a04741146c44ab1735782a89a60b038028bf2.
| * | Make comparison function (gt, ge, ...) respect __array_priority__.Frederic2013-05-101-0/+44
| | |
* | | BUG: f2py script shebang may refer to wrong pythonCarwyn Pelley2013-05-231-2/+2
| | | | | | | | | | | | | | | | | | | | | | | | | | | The f2py executable has a shebang which uses the default python, rather than the python it was compiled for. This causes issues for deployment of numpy (+f2py) across systems which have different environments. This fix uses sys.executable to determine the resulting hardcoded python to use.
* | | MAINT: Remove msvc_on_amd64 functionChristoph Gohlke2013-05-222-15/+1
| | | | | | | | | | | | This function is no longer required for building with msvc on AMD64
* | | Merge pull request #3342 from rgommers/easyinstall-issueCharles Harris2013-05-201-2/+4
|\ \ \ | | | | | | | | BLD: fix setuptools-specific easy_install issue. Closes gh-3160.
| * | | BLD: fix setuptools-specific easy_install issue. Closes gh-3160.Ralf Gommers2013-05-191-2/+4
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | An error is raised by setuptools when trying to write to /dev/null. Was fixed in distribute, but not in setuptools. No multi-arch support with plain setuptools should be OK, because multi-arch is Ubuntu specific (at least for now), and they ship distribute.
* | | | Merge pull request #3339 from endolith/min_max_docstringsCharles Harris2013-05-203-73/+131
|\ \ \ \ | |/ / / |/| | | DOC: Min max docstrings
| * | | DOC: Harmonize max and min docstrings with each otherendolith2013-05-173-38/+40
| | | |
| * | | DOC: Mention all min/max functions from all others, clarify differencesendolith2013-05-163-35/+91
| | | |
* | | | Merge pull request #3329 from JDWarner/faster_arraypadCharles Harris2013-05-192-429/+1102
|\ \ \ \ | | | | | | | | | | ENH: improved, faster algorithm for array padding
| * | | | ENH: improved, faster algorithm for array paddingJosh Warner (Mac)2013-05-192-429/+1102
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | New padding method which scales much better with dimensionality. This new implementation is fully vectorized, builds each abstracted n-dimensional padding block in a single step, and takes advantage of separability. The API is completely preserved, and the old algorithm is used if a vector function is input for `mode`. The new algorithm is faster for all tested combinations of inputs, and scales much better with dimensionality. Execution time reductions from ~25% for small rank 1 arrays to >99% for rank 4+ arrays observed.
* | | | | Merge pull request #3337 from seberg/issue-3336Charles Harris2013-05-172-4/+11
|\ \ \ \ \ | | | | | | | | | | | | BUG: Correctly pass on ddof paramter on inside np.ma.corrcoef
| * | | | | BUG: Correctly pass on ddof paramter on inside np.ma.corrcoefSebastian Berg2013-05-172-4/+11
| | |_|_|/ | |/| | | | | | | | | | | | | | | | | | | | | | | While ddof has basically no effect on corrcoef, it exists, but it was not passed on correctly (instead only bias would be passed on). Fixes gh-3336
* | | | | ENH: optimize byteswapping via intrinsicsJulian Taylor2013-05-152-8/+24
| |_|/ / |/| | | | | | | | | | | | | | | | | | | | | | | Byteswapping is implemented in hardware on x86 cpus which is more than double as fast than the existing implementation. The masked version used for 32bit plays better with compiler pattern matching so it can be optimized also by non-gcc/clang compilers.
* | | | Merge pull request #3308 from juliantaylor/optimize-isnannjsmith2013-05-154-17/+68
|\ \ \ \ | | | | | | | | | | ENH: use intrinsics for isnan, isfinite and isinf
| * | | | MAINT: document why OPTIONAL_INTRINSICS needs argumentsJulian Taylor2013-05-091-0/+1
| | | | |
| * | | | MAINT: document intrinsic macro guardsJulian Taylor2013-05-091-2/+4
| | | | |
| * | | | MAINT: mask npy_config.h include with HAVE_NPY_CONFIG_HJulian Taylor2013-05-092-0/+4
| | | | |
| * | | | ENH: add possibility to add define_macros to extensionsJulian Taylor2013-05-091-1/+19
| | | | |
| * | | | MAINT: check for intrinsics at configure timeJulian Taylor2013-05-083-3/+13
| | | | |
| * | | | ENH: use intrinsics for isnan, isfinite and isinfJulian Taylor2013-05-061-16/+32
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Use of intrinsics avoids expensive function calls in tight loops on x86 cpus where these operations are implemented in hardware while still retaining same portability. More than doubles performance of np.isnan/isinf/isfinite.
* | | | | Merge pull request #462 from ContinuumIO/ufunc-structCharles Harris2013-05-148-20/+375
|\ \ \ \ \ | | | | | | | | | | | | Create ufunc for struct array
| * | | | | Another name change to RegisterLoopForDescrJay Bourque2013-05-144-5/+5
| | | | | |
| * | | | | Code cleanupJay Bourque2013-05-141-13/+15
| | | | | |
| * | | | | Add test for RegisterLoopByDescr and custom dtypesJay Bourque2013-05-142-0/+42
| | | | | |
| * | | | | Change name of RegisterLoopForStructType to RegisterLoopByDescrJay Bourque2013-05-143-12/+15
| | | | | |
| * | | | | Add null checkJay Bourque2013-05-141-0/+4
| | | | | |
| * | | | | Remove unneeded codeJay Bourque2013-05-141-8/+0
| | | | | |
| * | | | | Whitespace changeJay Bourque2013-05-141-2/+3
| | | | | |
| * | | | | Change free to PyArray_free, and formatting changesJay Bourque2013-05-141-9/+13
| | | | | |
| * | | | | Update c-api documentationJay Bourque2013-05-141-1/+12
| | | | | |
| * | | | | Add support for python 3Jay Bourque2013-05-141-24/+17
| | | | | |
| * | | | | Add test for struct array ufuncsJay Bourque2013-05-143-0/+142
| | | | | |
| * | | | | Code cleanupJay Bourque2013-05-142-10/+15
| | | | | |
| * | | | | Fix build errorJay Bourque2013-05-141-1/+1
| | | | | |
| * | | | | Add comments for new RegisterLoopForStructType API methodJay Bourque2013-05-141-0/+10
| | | | | |
| * | | | | Add support for structured array ufuncsJay Bourque2013-05-144-20/+166
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | - Add New api method PyUFunc_RegisterLoopForStructType for creating new ufunc for structured array - Add arg_dtypes array to PyUFunc_Loop1d object. This keeps track of each struct dtype that the ufunc can handle. - Add arg_dtypes parameter to ufunc_loop_matches function so that struct dtypes can be compared for NPY_VOID type num
View Lorry Controller Status