def __init__(self, solver_file, snapshot = 0, gpu = 1, sync_freq=1, report=False, do_histogram=False):
self.solver_file = solver_file
self.snapshot = snapshot
self.num_gpu = gpu
self.sync_freq = sync_freq
self.report = report
self.do_histogram=do_histogram
if owl.has_mpi():
self.gpu = []
if gpu == 1:
#self.gpu += [owl.create_gpu_device(i) for i in range(owl.get_gpu_device_count())]
nodes = [owl.get_mpi_device_count(i) for i in range(1,owl.get_mpi_node_count())]
for n in range(len(nodes)):
print "using {} gpu's on node {}\n".format(nodes[n],n+1)
self.gpu += [owl.create_mpi_device(n+1,i+1) for i in range(nodes[n])]
self.num_gpu = len(self.gpu)
else:
self.gpu += [owl.create_cpu_device()]
self.gpu += [owl.create_mpi_device(n,0) for n in range(1,owl.get_mpi_node_count())]
self.num_gpu = len(self.gpu)
print "using {} cpu's over all nodes".format(self.num_gpu)
else:
if gpu == 1:
self.gpu = [owl.create_gpu_device(i) for i in range(self.num_gpu)]
self.num_gpu = len(self.gpu)
print "using {} gpu devices".format(len(self.gpu))
else:
self.gpu = [owl.create_cpu_device()]
self.num_gpu = len(self.gpu)
print "using {} cpus".format(len(self.gpu))
def __init__(self, data_file='mnist_all.mat', num_epochs=100, mb_size=256, eps_w=0.01, eps_b=0.01):
self.cpu = owl.create_cpu_device()
self.gpu = owl.create_gpu_device(0)
self.data_file = data_file
self.num_epochs=num_epochs
self.mb_size=mb_size
self.eps_w=eps_w
self.eps_b=eps_b
# init weight
l1 = 784; l2 = 256; l3 = 10
self.l1 = l1; self.l2 = l2; self.l3 = l3
self.w1 = owl.randn([l2, l1], 0.0, math.sqrt(4.0 / (l1 + l2)))
self.w2 = owl.randn([l3, l2], 0.0, math.sqrt(4.0 / (l2 + l3)))
self.b1 = owl.zeros([l2, 1])
self.b2 = owl.zeros([l3, 1])
def test(self):
# Expected
cpu=owl.create_cpu_device()
owl.set_device(cpu)
img = np.arange(0,32, dtype=np.float32) #/32
img = np.reshape(img,[1,2,4,4])
expected = np.asarray([[[5,7],
[13,15]],
[[21,23],
[29,31]]]) #/32.0
#expected = np.asarray([[[ 110.25, 124.25],
# [ 166.25, 180.25]],
# [[ 278.25, 324.25],
# [ 462.25, 508.25]]])
# test
owlimg = owl.from_numpy(img)
pooler = owl.conv.Pooler(2,2,2,2)
test = pooler.ff(owlimg)
print 'Expected\n',expected
print "Actual\n",test.to_numpy()
print "This test must be run with a fractional bit width of 12"
self.assertTrue(np.allclose(expected, test.to_numpy(), atol= 1.0/(1<<12)*4))
import owl import sys owl.initialize(sys.argv) owl.create_cpu_device() gpu0 = owl.create_gpu_device(0) gpu1 = owl.create_gpu_device(1) owl.set_device(gpu0)
import owl
devices = []
devices.append(owl.create_cpu_device())
if owl.has_mpi():
n = owl.get_mpi_node_count()
for i in range(1,n):
id = owl.create_mpi_device(i,0)
devices.append(id)
owl.set_device(devices[-1])
if args.gpu == 1:
usegpu = True
devs = []
#devs = [owl.create_cpu_device()]
print "enumerating devices"
if usempi:
nodes = owl.get_mpi_node_count()
print "{} mpi nodes found".format(nodes)
if usegpu:
# devs += [owl.create_gpu_device(i) for i in range(owl.get_gpu_device_count())]
# devs = [owl.create_cpu_device()]
devs += [owl.create_mpi_device(i,d+1) for i in range(1,nodes) for d in range(owl.get_mpi_device_count(i))]
print "Using {} MPI GPU's".format(len(devs))
else:
devs = [owl.create_cpu_device()]
devs += [owl.create_mpi_device(i,0) for i in range(1,nodes)]
print "Using {} MPI nodes".format(len(devs))
else:
if usegpu:
print "Using available GPU's"
devs += [owl.create_gpu_device(i) for i in range(owl.get_gpu_device_count())]
else:
print "Using CPU only"
devs = [owl.create_cpu_device()]
owl.set_device(devs[0])
print "Starting model creation"
model = MNISTCNNModel()
print "Starting random initialization"
model.init_random()
print "Starting training"
