def expected():
conv2d_1 = relay.nn.conv2d(
data1,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
device_copy1 = relay.device_copy(conv2d_1, dev2, dev1)
conv2d_2 = relay.nn.conv2d(
data2,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
device_copy2 = relay.device_copy(conv2d_2, dev2, dev1)
add = relay.add(device_copy1, device_copy2)
device_copy3 = relay.device_copy(add, dev1, dev2)
conv2d_3 = relay.nn.conv2d(
device_copy3,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
func = relay.Function([data1, weight, data2], conv2d_3)
return func
def expected():
conv2d_1 = relay.nn.conv2d(
data1,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
device_copy1 = relay.device_copy(conv2d_1, dev2, dev1)
conv2d_2 = relay.nn.conv2d(
data2,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
device_copy2 = relay.device_copy(conv2d_2, dev2, dev1)
add = relay.add(device_copy1, device_copy2)
device_copy3 = relay.device_copy(add, dev1, dev2)
conv2d_3 = relay.nn.conv2d(
device_copy3,
weight,
channels=64,
kernel_size=(3, 3),
padding=(1, 1))
func = relay.Function([data1, data2, weight], conv2d_3)
return func
def expected():
x = relay.var("x", relay.ty.TensorType((3, 3, 4), "float32"))
split = relay.op.split(x, 3)
elem0 = relay.device_copy(split[0], gpu_dev, cpu_dev)
elem1 = relay.device_copy(split[1], gpu_dev, cpu_dev)
sub = elem0 - elem1
func = relay.Function(relay.analysis.free_vars(sub), sub)
return func
def expected():
add = relay.add(x, y)
copy_add_sub1 = relay.device_copy(add, dev2, dev1)
sub1 = relay.subtract(copy_add_sub1, z)
copy_add_sub2 = relay.device_copy(add, dev2, dev1)
sub2 = relay.subtract(copy_add_sub2, z)
func = relay.Function([x, y, z], relay.Tuple([sub1, sub2]))
return func
def expected():
add = relay.add(x, y)
copy_add_sqrt = relay.device_copy(add, cpu_dev, dev_dev)
sqrt = relay.sqrt(copy_add_sqrt)
log = relay.log(add)
copy_sqrt_subtract = relay.device_copy(sqrt, dev_dev, cpu_dev)
subtract = relay.subtract(copy_sqrt_subtract, log)
copy_sub_exp = relay.device_copy(subtract, cpu_dev, dev_dev)
exp = relay.exp(copy_sub_exp)
func = relay.Function([x, y], exp)
return func
def expected():
log = relay.log(x)
_log_left = relay.device_copy(log, dev1, dev2)
_log_right = relay.device_copy(log, dev1, dev2)
log2 = relay.log2(_log_left)
log10 = relay.log10(_log_right)
add = relay.add(log2, log10)
_add = relay.device_copy(add, dev2, dev1)
tan = relay.tan(_add)
func = run_opt_pass(tan, transform.InferType())
return func
def expected():
add = relay.add(x, y)
copy_add_sqrt = relay.device_copy(add, cpu_ctx, dev_ctx)
sqrt = relay.sqrt(copy_add_sqrt)
log = relay.log(add)
copy_sqrt_subtract = relay.device_copy(sqrt, dev_ctx, cpu_ctx)
subtract = relay.subtract(copy_sqrt_subtract, log)
copy_sub_exp = relay.device_copy(subtract, cpu_ctx, dev_ctx)
exp = relay.exp(copy_sub_exp)
func = relay.Function([x, y], exp)
return func
def expected():
add = relay.add(a, b)
mul = relay.multiply(c, d)
copy_mul_sub = relay.device_copy(mul, cpu_dev, dev_dev)
sub = relay.subtract(add, copy_mul_sub)
func = relay.Function([a, b, c, d], sub)
return func
def expected():
add = relay.add(a, b)
mul = relay.multiply(c, d)
copy_mul_sub = relay.device_copy(mul, cpu_ctx, dev_ctx)
sub = relay.subtract(add, copy_mul_sub)
func = relay.Function([a, b, c, d], sub)
return func
def expected():
add = relay.add(x, y)
sqrt = relay.sqrt(add)
log = relay.log(add)
subtract = relay.subtract(sqrt, log)
copy_sub_exp = relay.device_copy(subtract, dev_ctx, cpu_ctx)
exp = relay.exp(copy_sub_exp)
func = relay.Function([x, y], exp)
return func
def expected():
add = relay.add(x, y)
copy_add_sub = relay.device_copy(add, dev1, dev2)
sub = relay.subtract(copy_add_sub, z)
return sub
def expected():
add = relay.add(x, y)
copy_add_sub = relay.device_copy(add, ctx2, ctx1)
sub = relay.subtract(copy_add_sub, z)
func = relay.Function([x, y, z], sub)
return func
def expected():
add = relay.add(x, y)
copy_add_sub = relay.device_copy(add, ctx2, ctx1)
sub = relay.subtract(copy_add_sub, z)
func = relay.Function([x, y, z], sub)
return func