1 / 0
knl = lp.split_iname(knl, "e", 16, outer_tag="g.0") #, slabs=(0, 1))
knl = lp.tag_inames(knl, dict(i="l.0", j="l.1"))
print(knl)
#1/0
kernel_gen = lp.generate_loop_schedules(knl)
kernel_gen = lp.check_kernels(kernel_gen, dict(K=1000), kill_level_min=5)
lp.auto_test_vs_ref(
seq_knl,
ctx,
kernel_gen,
op_count=0,
op_label="GFlops",
parameters={"K": K},
print_seq_code=True,
)
if __name__ == "__main__":
import sys
if len(sys.argv) > 1:
exec(sys.argv[1])
else:
from py.test.cmdline import main
main([__file__])
a_gpu = clrand(context, queue, (2000,), dtype=numpy.float32)
a = a_gpu.get().astype(numpy.float64)
a2 = a_gpu.astype(numpy.float64).get()
assert a2.dtype == numpy.float64
assert la.norm(a - a2) == 0, (a, a2)
a_gpu = clrand(context, queue, (2000,), dtype=numpy.float64)
a = a_gpu.get().astype(numpy.float32)
a2 = a_gpu.astype(numpy.float32).get()
assert a2.dtype == numpy.float32
assert la.norm(a - a2)/la.norm(a) < 1e-7
if __name__ == "__main__":
# make sure that import failures get reported, instead of skipping the tests.
import pyopencl as cl
import sys
if len(sys.argv) > 1:
exec(sys.argv[1])
else:
from py.test.cmdline import main
main([__file__])
# simple test runner wrapper (useful for test debugging in IDEs)
if __name__ == "__main__":
from py.test import cmdline
import sys
sys.exit(cmdline.main(['.'] + sys.argv[1:]))