def setUp(self):
self.h_data = numpy.random.random(self.N).astype("float32")
self.h2_data = numpy.random.random(
(self.N, self.N)).astype("float32").reshape((self.N, self.N))
self.ctx = ocl.create_context(devicetype="GPU")
device = self.ctx.devices[0]
try:
devtype = pyopencl.device_type.to_string(device.type).upper()
except ValueError:
# pocl does not describe itself as a CPU !
devtype = "CPU"
workgroup = device.max_work_group_size
if (devtype == "CPU") and (device.platform.vendor == "Apple"):
logger.info(
"For Apple's OpenCL on CPU: enforce max_work_goup_size=1")
workgroup = 1
self.ws = min(workgroup, self.ws)
self.queue = pyopencl.CommandQueue(
self.ctx,
properties=pyopencl.command_queue_properties.PROFILING_ENABLE)
self.local_mem = pyopencl.LocalMemory(
self.ws * 32) # 2float4 = 2*4*4 bytes per workgroup size
src = pyFAI.utils.read_cl_file("bitonic.cl")
self.prg = pyopencl.Program(self.ctx, src).build()
def init(self, devicetype="GPU", useFp64=True,
platformid=None, deviceid=None):
"""Initial configuration: Choose a device and initiate a
context. Devicetypes can be GPU, gpu, CPU, cpu, DEF, ACC,
ALL. Suggested are GPU,CPU. For each setting to work there
must be such an OpenCL device and properly installed. E.g.: If
Nvidia driver is installed, GPU will succeed but CPU will
fail. The AMD SDK kit (AMD APP) is required for CPU via
OpenCL.
@param devicetype: string in ["cpu","gpu", "all", "acc"]
@param useFp64: boolean specifying if double precision will be used
@param platformid: integer
@param devid: integer
"""
if self._ctx is None:
self._ctx = ocl.create_context(devicetype, useFp64,
platformid, deviceid)
device = self._ctx.devices[0]
self.devicetype = pyopencl.device_type.to_string(device.type)
if (self.devicetype == "CPU")\
and (device.platform.vendor == "Apple"):
logger.warning("This is a workaround for Apple's OpenCL"
" on CPU: enforce BLOCK_SIZE=1")
self.BLOCK_SIZE = 1
if self.nBins:
self.wdim_bins = (self.nBins + self.BLOCK_SIZE - 1) & \
~ (self.BLOCK_SIZE - 1),
if self.nData:
self.wdim_data = (self.nData + self.BLOCK_SIZE - 1) & \
~ (self.BLOCK_SIZE - 1),
self.useFp64 = "fp64" in device.extensions
platforms = pyopencl.get_platforms()
self.platformid = platforms.index(device.platform)
devices = platforms[self.platformid].get_devices()
self.deviceid = devices.index(device)
if self.filename:
self._queue = pyopencl.CommandQueue(
self._ctx,
properties=pyopencl.command_queue_properties.PROFILING_ENABLE)
else:
self._queue = pyopencl.CommandQueue(self._ctx)
else:
logger.warning("Recycling existing context ..."
" if you want to get start from scratch,"
" use clean()")
def setUp(self):
self.h_data = numpy.random.random(self.N).astype("float32")
self.h2_data = numpy.random.random((self.N, self.N)).astype("float32").reshape((self.N, self.N))
self.ctx = ocl.create_context(devicetype="GPU")
device = self.ctx.devices[0]
try:
devtype = pyopencl.device_type.to_string(device.type).upper()
except ValueError:
# pocl does not describe itself as a CPU !
devtype = "CPU"
workgroup = device.max_work_group_size
if (devtype == "CPU") and (device.platform.vendor == "Apple"):
logger.info("For Apple's OpenCL on CPU: enforce max_work_goup_size=1")
workgroup = 1
self.ws = min(workgroup, self.ws)
self.queue = pyopencl.CommandQueue(self.ctx, properties=pyopencl.command_queue_properties.PROFILING_ENABLE)
self.local_mem = pyopencl.LocalMemory(self.ws * 32) # 2float4 = 2*4*4 bytes per workgroup size
src = pyFAI.utils.read_cl_file("bitonic.cl")
self.prg = pyopencl.Program(self.ctx, src).build()
# AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
# LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN
# THE SOFTWARE.
from __future__ import print_function
import utilstest
import numpy, time
import pyFAI, pyFAI.opencl
from pyFAI.opencl import pyopencl, ocl
import pyopencl.array
N = 1024
ws = N // 8
ctx = ocl.create_context("GPU")
queue = pyopencl.CommandQueue(ctx, properties=pyopencl.command_queue_properties.PROFILING_ENABLE)
h_data = numpy.random.random(N).astype("float32")
d_data = pyopencl.array.to_device(queue, h_data)
local_mem = pyopencl.LocalMemory(ws * 32) # 2float4 = 2*4*4 bytes per workgroup size
src = pyFAI.utils.read_cl_file("bsort.cl")
prg = pyopencl.Program(ctx, src).build()
t0 = time.time()
hs_data = numpy.sort(h_data)
t1 = time.time()
time_sort = 1e3 * (t1 - t0)
print(time_sort)