pyqtgraph/examples/VideoSpeedTest.py

282 lines
9.0 KiB
Python
Raw Normal View History

# -*- coding: utf-8 -*-
"""
Tests the speed of image updates for an ImageItem and RawImageWidget.
The speed will generally depend on the type of data being shown, whether
it is being scaled and/or converted by lookup table, and whether OpenGL
is used by the view widget
"""
## Add path to library (just for examples; you do not need this)
import initExample
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
import argparse
import sys
import numpy as np
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
import pyqtgraph as pg
import pyqtgraph.ptime as ptime
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
from pyqtgraph.Qt import QtGui, QtCore, QT_LIB
import importlib
ui_template = importlib.import_module(f'VideoTemplate_{QT_LIB.lower()}')
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
try:
import cupy as cp
pg.setConfigOption("useCupy", True)
_has_cupy = True
except ImportError:
cp = None
_has_cupy = False
2021-02-22 08:13:53 +00:00
try:
from pyqtgraph.widgets.RawImageWidget import RawImageGLWidget
except ImportError:
RawImageGLWidget = None
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
parser = argparse.ArgumentParser(description="Benchmark for testing video performance")
parser.add_argument('--cuda', default=False, action='store_true', help="Use CUDA to process on the GPU", dest="cuda")
parser.add_argument('--dtype', default='uint8', choices=['uint8', 'uint16', 'float'], help="Image dtype (uint8, uint16, or float)")
parser.add_argument('--frames', default=3, type=int, help="Number of image frames to generate (default=3)")
parser.add_argument('--image-mode', default='mono', choices=['mono', 'rgb'], help="Image data mode (mono or rgb)", dest='image_mode')
parser.add_argument('--levels', default=None, type=lambda s: tuple([float(x) for x in s.split(',')]), help="min,max levels to scale monochromatic image dynamic range, or rmin,rmax,gmin,gmax,bmin,bmax to scale rgb")
parser.add_argument('--lut', default=False, action='store_true', help="Use color lookup table")
parser.add_argument('--lut-alpha', default=False, action='store_true', help="Use alpha color lookup table", dest='lut_alpha')
parser.add_argument('--size', default='512x512', type=lambda s: tuple([int(x) for x in s.split('x')]), help="WxH image dimensions default='512x512'")
args = parser.parse_args(sys.argv[1:])
2021-02-22 08:13:53 +00:00
if RawImageGLWidget is not None:
# don't limit frame rate to vsync
sfmt = QtGui.QSurfaceFormat()
sfmt.setSwapInterval(0)
QtGui.QSurfaceFormat.setDefaultFormat(sfmt)
app = pg.mkQApp("Video Speed Test Example")
win = QtGui.QMainWindow()
2013-02-25 04:09:03 +00:00
win.setWindowTitle('pyqtgraph example: VideoSpeedTest')
ui = ui_template.Ui_MainWindow()
ui.setupUi(win)
win.show()
2021-02-22 08:13:53 +00:00
if RawImageGLWidget is None:
ui.rawGLRadio.setEnabled(False)
ui.rawGLRadio.setText(ui.rawGLRadio.text() + " (OpenGL not available)")
else:
ui.rawGLImg = RawImageGLWidget()
ui.stack.addWidget(ui.rawGLImg)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
# read in CLI args
ui.cudaCheck.setChecked(args.cuda and _has_cupy)
ui.cudaCheck.setEnabled(_has_cupy)
ui.framesSpin.setValue(args.frames)
ui.widthSpin.setValue(args.size[0])
ui.heightSpin.setValue(args.size[1])
ui.dtypeCombo.setCurrentText(args.dtype)
ui.rgbCheck.setChecked(args.image_mode=='rgb')
ui.maxSpin1.setOpts(value=255, step=1)
ui.minSpin1.setOpts(value=0, step=1)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
levelSpins = [ui.minSpin1, ui.maxSpin1, ui.minSpin2, ui.maxSpin2, ui.minSpin3, ui.maxSpin3]
if args.cuda and _has_cupy:
xp = cp
else:
xp = np
if args.levels is None:
ui.scaleCheck.setChecked(False)
ui.rgbLevelsCheck.setChecked(False)
else:
ui.scaleCheck.setChecked(True)
if len(args.levels) == 2:
ui.rgbLevelsCheck.setChecked(False)
ui.minSpin1.setValue(args.levels[0])
ui.maxSpin1.setValue(args.levels[1])
elif len(args.levels) == 6:
ui.rgbLevelsCheck.setChecked(True)
for spin,val in zip(levelSpins, args.levels):
spin.setValue(val)
else:
raise ValueError("levels argument must be 2 or 6 comma-separated values (got %r)" % (args.levels,))
ui.lutCheck.setChecked(args.lut)
ui.alphaCheck.setChecked(args.lut_alpha)
#ui.graphicsView.useOpenGL() ## buggy, but you can try it if you need extra speed.
vb = pg.ViewBox()
ui.graphicsView.setCentralItem(vb)
vb.setAspectLocked()
img = pg.ImageItem()
vb.addItem(img)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
LUT = None
def updateLUT():
global LUT, ui
dtype = ui.dtypeCombo.currentText()
if dtype == 'uint8':
n = 256
else:
n = 4096
LUT = ui.gradient.getLookupTable(n, alpha=ui.alphaCheck.isChecked())
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
if _has_cupy and xp == cp:
LUT = cp.asarray(LUT)
ui.gradient.sigGradientChanged.connect(updateLUT)
updateLUT()
ui.alphaCheck.toggled.connect(updateLUT)
def updateScale():
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
global ui, levelSpins
if ui.rgbLevelsCheck.isChecked():
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
for s in levelSpins[2:]:
s.setEnabled(True)
else:
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
for s in levelSpins[2:]:
s.setEnabled(False)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
updateScale()
ui.rgbLevelsCheck.toggled.connect(updateScale)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
cache = {}
def mkData():
with pg.BusyCursor():
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
global data, cache, ui, xp
frames = ui.framesSpin.value()
width = ui.widthSpin.value()
height = ui.heightSpin.value()
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
cacheKey = (ui.dtypeCombo.currentText(), ui.rgbCheck.isChecked(), frames, width, height)
if cacheKey not in cache:
if cacheKey[0] == 'uint8':
dt = xp.uint8
loc = 128
scale = 64
mx = 255
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
elif cacheKey[0] == 'uint16':
dt = xp.uint16
loc = 4096
scale = 1024
mx = 2**16
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
elif cacheKey[0] == 'float':
dt = xp.float32
loc = 1.0
scale = 0.1
mx = 1.0
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
else:
raise ValueError(f"unable to handle dtype: {cacheKey[0]}")
if ui.rgbCheck.isChecked():
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
data = xp.random.normal(size=(frames,width,height,3), loc=loc, scale=scale)
data = pg.gaussianFilter(data, (0, 6, 6, 0))
else:
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
data = xp.random.normal(size=(frames,width,height), loc=loc, scale=scale)
data = pg.gaussianFilter(data, (0, 6, 6))
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
if cacheKey[0] != 'float':
data = xp.clip(data, 0, mx)
data = data.astype(dt)
data[:, 10, 10:50] = mx
data[:, 9:12, 48] = mx
data[:, 8:13, 47] = mx
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
cache = {cacheKey: data} # clear to save memory (but keep one to prevent unnecessary regeneration)
data = cache[cacheKey]
updateLUT()
updateSize()
def updateSize():
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
global ui, vb
frames = ui.framesSpin.value()
width = ui.widthSpin.value()
height = ui.heightSpin.value()
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
dtype = xp.dtype(str(ui.dtypeCombo.currentText()))
rgb = 3 if ui.rgbCheck.isChecked() else 1
ui.sizeLabel.setText('%d MB' % (frames * width * height * rgb * dtype.itemsize / 1e6))
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
vb.setRange(QtCore.QRectF(0, 0, width, height))
def noticeCudaCheck():
global xp, cache
cache = {}
if ui.cudaCheck.isChecked():
if _has_cupy:
xp = cp
else:
xp = np
ui.cudaCheck.setChecked(False)
else:
xp = np
mkData()
mkData()
ui.dtypeCombo.currentIndexChanged.connect(mkData)
ui.rgbCheck.toggled.connect(mkData)
ui.widthSpin.editingFinished.connect(mkData)
ui.heightSpin.editingFinished.connect(mkData)
ui.framesSpin.editingFinished.connect(mkData)
ui.widthSpin.valueChanged.connect(updateSize)
ui.heightSpin.valueChanged.connect(updateSize)
ui.framesSpin.valueChanged.connect(updateSize)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
ui.cudaCheck.toggled.connect(noticeCudaCheck)
ptr = 0
lastTime = ptime.time()
fps = None
def update():
global ui, ptr, lastTime, fps, LUT, img
if ui.lutCheck.isChecked():
useLut = LUT
else:
useLut = None
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
downsample = ui.downsampleCheck.isChecked()
if ui.scaleCheck.isChecked():
if ui.rgbLevelsCheck.isChecked():
useScale = [
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
[ui.minSpin1.value(), ui.maxSpin1.value()],
[ui.minSpin2.value(), ui.maxSpin2.value()],
[ui.minSpin3.value(), ui.maxSpin3.value()]]
else:
useScale = [ui.minSpin1.value(), ui.maxSpin1.value()]
else:
useScale = None
if ui.rawRadio.isChecked():
ui.rawImg.setImage(data[ptr%data.shape[0]], lut=useLut, levels=useScale)
ui.stack.setCurrentIndex(1)
elif ui.rawGLRadio.isChecked():
ui.rawGLImg.setImage(data[ptr%data.shape[0]], lut=useLut, levels=useScale)
ui.stack.setCurrentIndex(2)
else:
img.setImage(data[ptr%data.shape[0]], autoLevels=False, levels=useScale, lut=useLut, autoDownsample=downsample)
ui.stack.setCurrentIndex(0)
#img.setImage(data[ptr%data.shape[0]], autoRange=False)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
ptr += 1
now = ptime.time()
dt = now - lastTime
lastTime = now
if fps is None:
fps = 1.0/dt
else:
s = np.clip(dt*3., 0, 1)
fps = fps * (1-s) + (1.0/dt) * s
ui.fpsLabel.setText('%0.2f fps' % fps)
app.processEvents() ## force complete redraw for every plot
timer = QtCore.QTimer()
timer.timeout.connect(update)
timer.start(0)
Performance enhancement: use CUDA in ImageItem (#1466) * 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>
2021-01-20 05:26:24 +00:00
## Start Qt event loop unless running in interactive mode or using pyside.
if __name__ == '__main__':
import sys
if (sys.flags.interactive != 1) or not hasattr(QtCore, 'PYQT_VERSION'):
QtGui.QApplication.instance().exec_()