cudaGetErrorString(sts),
n_blocks.x,
n_blocks.y,
n_threads.x,
n_threads.y,
n_threads.z,
n_shared);
%(fail)s;
}
} // END NESTED SCOPE
""" % locals()
def gpu_images2neibs(ten4, neib_shape, neib_step=None, mode='valid'):
return GpuImages2Neibs(mode)(ten4, neib_shape, neib_step)
@local_optimizer([Images2Neibs])
def use_gpu_images2neibs(node):
if (type(node.op) is Images2Neibs and
node.inputs[0].dtype == 'float32' and
node.op.mode in ['valid', 'ignore_borders',
'wrap_centered']):
return [host_from_gpu(gpu_images2neibs(gpu_from_host(node.inputs[0]),
node.inputs[1], node.inputs[2],
mode=node.op.mode))]
if cuda_available:
register_gpu_opt()(use_gpu_images2neibs)
return code
@local_optimizer([CumsumOp])
def use_gpu_cumsum(node):
if type(node.op) is CumsumOp \
and node.inputs[0].dtype == 'float32' \
and node.inputs[0].owner \
and isinstance(node.inputs[0].owner.op, HostFromGpu):
axis = node.op.axis
x = node.inputs[0]
if axis is not None and x.ndim > GpuCumsum.SUPPORTED_NDIMS:
return None
x = gpu_from_host(x)
if axis is None and x.ndim > 1:
x = GpuFlatten()(x)
# ``gpu_cumsum`` assume array has been flattened if needed.
if axis is None:
axis = 0
return [host_from_gpu(GpuCumsum(axis)(x))]
if cuda_available:
register_gpu_opt()(use_gpu_cumsum)
n_threads.x,
n_threads.y,
n_threads.z,
n_shared);
%(fail)s;
}
} // END NESTED SCOPE
""" % locals()
def gpu_images2neibs(ten4, neib_shape, neib_step=None, mode='valid'):
return GpuImages2Neibs(mode)(ten4, neib_shape, neib_step)
@local_optimizer()
def use_gpu_images2neibs(node):
if (type(node.op) is Images2Neibs and node.inputs[0].dtype == 'float32'
and node.op.mode in ['valid', 'ignore_borders', 'wrap_centered']):
return [
host_from_gpu(
gpu_images2neibs(gpu_from_host(node.inputs[0]),
node.inputs[1],
node.inputs[2],
mode=node.op.mode))
]
if cuda_available:
register_gpu_opt()(use_gpu_images2neibs)
return code
@local_optimizer([CumsumOp])
def use_gpu_cumsum(node):
if type(node.op) is CumsumOp \
and node.inputs[0].dtype == 'float32' \
and node.inputs[0].owner \
and isinstance(node.inputs[0].owner.op, HostFromGpu):
axis = node.op.axis
x = node.inputs[0]
if axis is not None and x.ndim > GpuCumsum.SUPPORTED_NDIMS:
return None
x = gpu_from_host(x)
if axis is None and x.ndim > 1:
x = GpuFlatten()(x)
# ``gpu_cumsum`` assume array has been flattened if needed.
if axis is None:
axis = 0
return [host_from_gpu(GpuCumsum(axis)(x))]
if cuda_available:
register_gpu_opt()(use_gpu_cumsum)