def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, tmax = self.args
fields = gpu.Fields(0, nx, ny, nz, coeff_use, precision_float)
gpu.Core(fields)
fields_ref = naive.Fields(nx,
ny,
nz,
precision_float,
segment_nbytes=64)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_eh_bufs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ce_bufs(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_ch_bufs(*chs)
fields_ref.set_chs(*chs)
tmpf = np.zeros(fields.ns_pitch, dtype=dtype)
# update
if ufunc == 'e':
for tstep in xrange(0, tmax):
fields.update_e()
fields_ref.update_e()
for strf, eh in zip(strf_list, ehs)[:3]:
cuda.memcpy_dtoh(tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(
norm, 0,
'%s, %s, %g, %g' % (self.args, strf, norm, max_diff))
elif ufunc == 'h':
for tstep in xrange(0, tmax):
fields.update_h()
fields_ref.update_h()
for strf, eh in zip(strf_list, ehs)[3:]:
cuda.memcpy_dtoh(tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(
norm, 0,
'%s, %s, %g, %g' % (self.args, strf, norm, max_diff))
fields.context.pop()
def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, use_cpu_core, tmax = self.args
qtask = cpu.QueueTask()
fields = cpu.Fields(qtask, nx, ny, nz, coeff_use, precision_float,
use_cpu_core)
cpu.Core(fields)
fields_ref = naive.Fields(nx,
ny,
nz,
precision_float,
segment_nbytes=16)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_ehs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ces(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_chs(*chs)
fields_ref.set_chs(*chs)
# verify
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
if ufunc == 'e':
for tstep in xrange(0, tmax):
fields.update_e()
fields_ref.update_e()
fields.enqueue_barrier()
for strf, eh in zip(strf_list, ehs)[:3]:
norm = np.linalg.norm(fields.get(strf) - fields_ref.get(strf))
max_diff = np.abs(fields.get(strf) -
fields_ref.get(strf)).max()
self.assertEqual(
norm, 0,
'%s, %s, %g, %g' % (self.args, strf, norm, max_diff))
elif ufunc == 'h':
for tstep in xrange(0, tmax):
fields.update_h()
fields_ref.update_h()
fields.enqueue_barrier()
for strf, eh in zip(strf_list, ehs)[3:]:
norm = np.linalg.norm(fields.get(strf) - fields_ref.get(strf))
max_diff = np.abs(fields.get(strf) -
fields_ref.get(strf)).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % \
(self.args, strf, norm, max_diff) )
def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, tmax = self.args
gpu_devices = common_gpu.gpu_device_list(print_info=False)
context = cl.Context(gpu_devices)
device = gpu_devices[0]
fields = gpu.Fields(context, device, nx, ny, nz, coeff_use,
precision_float)
gpu.Core(fields)
fields_ref = naive.Fields(nx, ny, nz, precision_float)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_eh_bufs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ce_bufs(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_ch_bufs(*chs)
fields_ref.set_chs(*chs)
tmpf = np.zeros(fields.ns, dtype=dtype)
# update
if ufunc == 'e':
for tstep in xrange(0, tmax):
fields.update_e()
fields_ref.update_e()
for strf, eh in zip(strf_list, ehs)[:3]:
cl.enqueue_copy(fields.queue, tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(
norm, 0,
'%s, %s, %g, %g' % (self.args, strf, norm, max_diff))
elif ufunc == 'h':
for tstep in xrange(0, tmax):
fields.update_h()
fields_ref.update_h()
for strf, eh in zip(strf_list, ehs)[3:]:
cl.enqueue_copy(fields.queue, tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(
norm, 0,
'%s, %s, %g, %g' % (self.args, strf, norm, max_diff))
def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, tmax = self.args
gpu_devices = common_gpu.gpu_device_list(print_info=False)
context = cl.Context(gpu_devices)
device = gpu_devices[0]
qtask = cpu.QueueTask()
fields = gpu.Fields(context, device, qtask, nx, ny, nz, coeff_use, precision_float)
gpu.Core(fields)
fields_ref = naive.Fields(nx, ny, nz, precision_float, segment_nbytes=64)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_eh_bufs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ce_bufs(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_ch_bufs(*chs)
fields_ref.set_chs(*chs)
tmpf = np.zeros(fields.ns_pitch, dtype=dtype)
# update
if ufunc == 'e':
for tstep in xrange(0, tmax):
fields.update_e()
fields_ref.update_e()
qtask.enqueue_barrier()
for strf, eh in zip(strf_list, ehs)[:3]:
cl.enqueue_copy(fields.queue, tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % (self.args, strf, norm, max_diff) )
elif ufunc == 'h':
for tstep in xrange(0, tmax):
fields.update_h()
fields_ref.update_h()
for strf, eh in zip(strf_list, ehs)[3:]:
cl.enqueue_copy(fields.queue, tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % (self.args, strf, norm, max_diff) )
def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, use_cpu_core, tmax = self.args
qtask = cpu.QueueTask()
fields = cpu.Fields(qtask, nx, ny, nz, coeff_use, precision_float, use_cpu_core)
cpu.Core(fields)
fields_ref = naive.Fields(nx, ny, nz, precision_float, segment_nbytes=16)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_ehs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ces(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_chs(*chs)
fields_ref.set_chs(*chs)
# verify
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
if ufunc == 'e':
for tstep in xrange(0, tmax):
for instance in fields.instance_list:
instance.update_e('pre')
instance.update_e('post')
fields_ref.update_e()
fields.enqueue_barrier()
for strf, eh in zip(strf_list, ehs)[:3]:
norm = np.linalg.norm(fields.get(strf) - fields_ref.get(strf))
max_diff = np.abs(fields.get(strf) - fields_ref.get(strf)).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % (self.args, strf, norm, max_diff) )
elif ufunc == 'h':
for tstep in xrange(0, tmax):
for instance in fields.instance_list:
instance.update_h('pre')
instance.update_h('post')
fields_ref.update_h()
fields.enqueue_barrier()
for strf, eh in zip(strf_list, ehs)[3:]:
norm = np.linalg.norm(fields.get(strf) - fields_ref.get(strf))
max_diff = np.abs(fields.get(strf) - fields_ref.get(strf)).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % \
(self.args, strf, norm, max_diff) )
def runTest(self):
ufunc, nx, ny, nz, coeff_use, precision_float, tmax = self.args
fields = gpu.Fields(0, nx, ny, nz, coeff_use, precision_float)
gpu.Core(fields)
fields_ref = naive.Fields(nx, ny, nz, precision_float, segment_nbytes=64)
naive.Core(fields_ref)
# allocations
ns = fields.ns
dtype = fields.dtype
strf_list = ['ex', 'ey', 'ez', 'hx', 'hy', 'hz']
ehs = common_random.generate_ehs(nx, ny, nz, dtype, ufunc)
fields.set_eh_bufs(*ehs)
fields_ref.set_ehs(*ehs)
ces, chs = common_random.generate_cs(nx, ny, nz, dtype, coeff_use)
if 'e' in coeff_use:
fields.set_ce_bufs(*ces)
fields_ref.set_ces(*ces)
if 'h' in coeff_use:
fields.set_ch_bufs(*chs)
fields_ref.set_chs(*chs)
tmpf = np.zeros(fields.ns_pitch, dtype=dtype)
# update
if ufunc == 'e':
for tstep in xrange(0, tmax):
fields.update_e()
fields_ref.update_e()
for strf, eh in zip(strf_list, ehs)[:3]:
cuda.memcpy_dtoh(tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % (self.args, strf, norm, max_diff) )
elif ufunc == 'h':
for tstep in xrange(0, tmax):
fields.update_h()
fields_ref.update_h()
for strf, eh in zip(strf_list, ehs)[3:]:
cuda.memcpy_dtoh(tmpf, fields.get_buf(strf))
norm = np.linalg.norm(fields_ref.get(strf) - tmpf)
max_diff = np.abs(fields_ref.get(strf) - tmpf).max()
self.assertEqual(norm, 0, '%s, %s, %g, %g' % (self.args, strf, norm, max_diff) )
fields.context.pop()