* implement rescale using numba
* workaround to pass test_makeARGB.py
* key on (input, output) manually
* remove minimum version check
* signature needs to be a list of signatures
numba considers it a mistake for single-item non-list but works around
it internally
* add numba test to test_makeARGB.py
* add numba as dependency in CI
* handle properly the case where useNumba was already True
* restore useCupy setting after test
* filter numba nan warning
* don't make changes to test_cupy_makeARGB_...
Various places in the library attempt to check if scalars are finite
via numpy methods, which are intended to be used on numpy arrays. Using
the math module equivalent functions on scalars is significantly faster.
In a few places, I also use numpy methods explicitly (np.all vs. all)
Many places in the library were doing radian to degree conversion
via the manual calculation. Doing timeit benchmarks on my system, I
am able to get a 4x speedup by using math.degrees or math.radians
instead
the "offset" argument passed into rescaleData() is typically an element
of an ndarray, i.e. it comes from the lower bound value of the "levels"
ndarray. as such, arithmetic operations on it can overflow.
this triggers a runtime warning in the test suite.
the workaround is to convert it to a Python (integer) scalar.
* implement rescaleData_blocked
clip limits should be int if data is int
* add test for rescaleData_blocked
* dispatch to different versions depending on numpy or cupy
* make rescaleData() the only entry-point
* fix clip_array()
return error for invalid inputs.
use minmax for win32, umath.clip for other platforms
the previous code was penalizing Linux
* force output to be an array
* Initial asv configuration
* makeARGB benchmarks are working
* Fix array type checking and allow making QImage in greyscale mode
* Performance improvements
* benchmark minor update
* Add CLI args to video speed test for easier / automated benchmarking
* udpate asv conf
* use a buffer-qimage so we can avoid allocing so much
this should improve performance under windows
* Add CLI args to video speed test for easier / automated benchmarking
* use a buffer-qimage so we can avoid allocing so much
this should improve performance under windows
* playing with numba
* oh, mins/maxes in the other order
* maybe put the cupy in here and see what happens
* pre-alloc for gpu and cpu
* handle possibility of not having cupy
* no numba in this branch
* organize imports
* name them after their use, not their expected device
* cupy.take does not support clip mode, so do it explicitly
* add CUDA option to the VideoSpeedTest
* rename private attr xp to _xp
* handle resizes at the last moment
* cupy is less accepting of lists as args
* or somehow range isn't allowed? what histogram is this?
* construct the array with python objects
* get the python value right away
* put LUT into cupy if needed
* docstring about cuda toolkit version
* better handling and display of missing cuda lib
* lint
* import need
* handle switching between cupy and numpy in a single ImageItem
* only use xp when necessary
we can now depend on numpy >= 1.17, which means __array_function__-implementing cupy can
seamlessly pass into numpy functions. the remaining uses of xp are for our functions which
need to allocate new data structures, an operation that has to be substrate-specific.
remove empty_cupy; just check if the import succeeded, instead.
* use an option to control use of cupy
* convert cupy.ceil array to int for easier mathing
* RawImageWidget gets to use the getCupy function now, too
* raise error to calm linters; rename for clarity
* Add Generated Template Files
* document things better
* cruft removal
* warnings to communicate when cupy is expected but somehow broken
* Merge branch 'cupy-rebase' into 'l/imageitem-performance'
Abundant conflicts; accept theirs in nearly every case.
* playing with settings to suss out timeout
* playing with more stuff to suss out timeout
* replace with empty list
* skip test_ExampleApp on linux+pyside2 only
* clean out some bits that no longer make sense; linty
* ignore airspeed velocity dir
* lint
* tidy up for merge
* lint
* lint, avoid shadowing
* specific import; run-as-script setup
Co-authored-by: Luke Campagnola <luke.campagnola@gmail.com>
Co-authored-by: Ogi Moore <ognyan.moore@gmail.com>
* Add CLI args to video speed test for easier / automated benchmarking
* use a buffer-qimage so we can avoid allocing so much
this should improve performance under windows
* playing with numba
* oh, mins/maxes in the other order
* maybe put the cupy in here and see what happens
* pre-alloc for gpu and cpu
* handle possibility of not having cupy
* no numba in this branch
* organize imports
* name them after their use, not their expected device
* cupy.take does not support clip mode, so do it explicitly
* add CUDA option to the VideoSpeedTest
* rename private attr xp to _xp
* handle resizes at the last moment
* cupy is less accepting of lists as args
* or somehow range isn't allowed? what histogram is this?
* construct the array with python objects
* get the python value right away
* put LUT into cupy if needed
* docstring about cuda toolkit version
* better handling and display of missing cuda lib
* lint
* import need
* handle switching between cupy and numpy in a single ImageItem
* only use xp when necessary
we can now depend on numpy >= 1.17, which means __array_function__-implementing cupy can
seamlessly pass into numpy functions. the remaining uses of xp are for our functions which
need to allocate new data structures, an operation that has to be substrate-specific.
remove empty_cupy; just check if the import succeeded, instead.
* use an option to control use of cupy
* convert cupy.ceil array to int for easier mathing
* RawImageWidget gets to use the getCupy function now, too
* raise error to calm linters; rename for clarity
* Add Generated Template Files
* document things better
* cruft removal
* warnings to communicate when cupy is expected but somehow broken
* playing with settings to suss out timeout
* playing with more stuff to suss out timeout
* replace with empty list
* skip test_ExampleApp on linux+pyside2 only
Co-authored-by: Luke Campagnola <luke.campagnola@gmail.com>
Co-authored-by: Ogi Moore <ognyan.moore@gmail.com>
The i-th position of that array used to define if points (i) and (i+1)
were connected, but in master it defines whether points (i-1) and (i) are connected.
This commit reverts to (i) and (i+1) interpretation.
* Permit nan, inf, and -inf for float SpinBox
Bounds are enforced against inf and -inf inputs, but not for nan.
* Ensure SpinBox text is updated when out-of-bounds value is entered
* Make inf and nan entry in SpinBox case-insensitive
* Make SpinBox example behave as advertised
* Make non-finite SpinBox values optionally allowed
* Python 2 compatibility
Although the documentation used to say that specifying tipAngle would
override headWidth, headWidth was never used. The new behaviour is that
tipAngle will be used, with a default value of 25, unless headWidth is
specified.
Co-authored-by: Ogi Moore <ognyan.moore@gmail.com>
