| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| ... | |
| |\ \ \ \
| |/ / /
|/| | | |
BUG: add module extensions to load_library search list
|
| |/ / /
| | |
| | |
| | |
| | | |
on mac, windows and linux (with >= py3.2) shared libraries and loadable
module have different extensions, so check both for all platforms.
|
| |\ \ \
| | | |
| | | | |
vectorize boolean reductions
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
Improves performance by a factor of 5 to 10 depending on the cpu.
These operations currently can't be autovectorized by gcc 4.8.
|
| | | | | |
|
| |\ \ \ \
| |_|/ /
|/| | | |
ENH: tell gcc to unroll strided copy loops
|
| | | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The strided copy loops profit a lot from unrolling as the number of
operations executed is in each iterations very small.
GCC needs to be told explicitly to do unrolling even on O3.
Unrolling is only profitable if the move can be done in a single
instruction, else the increased code size makes it slower, thus the flag
is only used for operations on element sizes less equal the native
pointer size.
Tested to improve performance of by 20-50% on intel core2duo, xeon
5xxx/7xxx and amd phenom x4.
|
| |\ \ \ \
| | |/ /
| |/| | |
BUG: Campanion Matrix was scalar, not matrix for degree 1.
|
| |/ / /
| | |
| | |
| | |
| | |
| | |
| | |
| | | |
The companion matrices returned by the various polynomial types was
a scalar in the degree one case instead of a 2-D array. Fix that and
add a test to check for that result.
Closes #3459.
|
| |\ \ \
| | | |
| | | | |
Possible memory overwrite in copy_and_swap
|
| | | | |
| | | |
| | | |
| | | |
| | | | |
modified by the call to the strided memcpy and will end up pointing
after the end of the destination buffer in that case.
|
| |\ \ \ \
| | | | |
| | | | | |
BUG: fix uninitialized-use on invalid sort kind
|
| | | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
orign is unset at that point but used in the fail goto,
can probably only be triggered by using the C-api wrong.
|
| |\ \ \ \ \
| |/ / / /
| | | | | |
BUG: Fix failing python long behaviour and possible heisen bug
|
| |/ / / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
The priority error fixes np.float_(64) + python_long not working
anymore. The second change reformats the
PyArray_GetAttrString_SuppressException to avoid a possible
heisenbug decrefing an object before use.
|
| |\ \ \ \
| | | | |
| | | | | |
BUG: np.ma.compress treated inputs in wrong order; closes #2495
|
| | | | | | |
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
BUG: add __len__ method to ma.mvoid; closes #576
|
| |/ / / / / |
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
ENH: vectorize boolean logical &&, ||, abs and not
|
| | | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
The code ensures the result is identical to a boolean operation even
though this might be unnecessary if bools are used correctly everywhere.
The overhead doesn't matter much as vectorizing single byte operations
hits the memory bandwidth limit very quickly.
Improves performance by about a factor of 5 to 10 depending on the cpu.
These operations currently can't be autovectorized by gcc 4.8.
|
| |\ \ \ \ \ \
| |_|_|/ / /
|/| | | | | |
BUG: field assignment in masked array did not reset mask; closes #2403
|
| | | |/ / /
| |/| | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | |
| | | | | |
The previous behavior when setting a field after indexing to
select an element was suitable for the hard mask case, but
not for the default soft mask. In addition, the _hardmask
value was not being set at all in the mvoid instance. With
this changeset, the _hardmask is passed in and __setitem__
takes it into account.
|
| |\ \ \ \ \
| | | | | |
| | | | | |
| | | | | |
| | | | | | |
https://github.com/numpy/numpy/pull/3444
Fixed compilation with Cygwin.
|
| |/ / / / / |
|
| |\ \ \ \ \
| |/ / / /
|/| | | | |
BUG: Revert "ENH: enable unaligned loads on x86"
|
| | |/ / /
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | |
| | | | |
This reverts commit aef286debfd11a62f1c337dea55624cee7fd4d9e.
x86 platform works with unaligned access but the compiler is allowed to assume
all data is aligned to its size by the C standard. This means it can vectorize
instructions peeling only by the size of the type, if the data is not aligned
to this size one ends up with data not correctly aligned for SSE instructions
(16 byte).
So this flag can only be enabled if autovectorization is disabled.
|
| |\ \ \ \
| | | | |
| | | | | |
BUG: ma: ma.average didn't handle complex arrays correctly (issue gh-2684)
|
| | | | | | |
|
| | | | | | |
|
| |\ \ \ \ \
| | | | | |
| | | | | | |
Fix minor typos in Polynomial class documentation
|
| |/ / / / /
| | | | |
| | | | |
| | | | | |
Signed-Off-By: Kumar Appaiah <a.kumar@alumni.iitm.ac.in>
|
| |\ \ \ \ \
| |_|/ / /
|/| | | | |
DOC: fix typo in linalg.lstsq doc string
|
| | | | | | |
|
| | | | | | |
|
| | | |/ /
| |/| | |
|
| |\ \ \ \
| |_|/ /
|/| | | |
Updated info about sphinx.ext.autosummary.
|
| |/ / / |
|
| |\ \ \
| |/ /
|/| | |
ENH: random: Allow ngood=0 or nbad=0 in mtrand.hypergeometric.
|
| |/ /
| |
| |
| |
| | |
Also edited the 'Parameters' section of the docstring to comply
with the numpy docstring standard.
|
| |\ \
| | |
| | | |
ENH: Optimize array creation by avoiding errors
|
| | |\ \
| | | |
| | | |
| | | | |
enh_np_array0.1
|
| | | | | |
|
| | | |\ \
| | | | |
| | | | |
| | | | | |
enh_np_array0.1
|
| | | | | | |
|
| | | | |\ \ |
|
| | | | | | | |
|
| | | | | | | |
|
| | | | | |\ \ |
|
| | | | | | |\ \
| | | | | | | |
| | | | | | | |
| | | | | | | | |
enh_np_array0.1
|
