ZihengJiang opened a new issue #6015:
URL: https://github.com/apache/incubator-tvm/issues/6015
We have scalar support in `te.compute` now. But for `te.extern`, tvm will
raise error for scalar input. It would be great that we can have consistent
behavior.
Sample code:
```
import tvm
from tvm import te
import numpy as np
a = np.array(1.0, 'float32')
b = np.array(0.0, 'float32')
@tvm.register_func("tvm.copy")
def mycopy(x, y):
x.copyto(y)
A = te.placeholder(a.shape, name='A')
B = te.extern(a.shape, [A],
lambda ins, outs: tvm.tir.call_packed(
"tvm.copy", ins[0], outs[0]), name="B")
s = te.create_schedule(B.op)
f = tvm.build(s, [A, B], 'llvm')
ta = tvm.nd.array(a)
tb = tvm.nd.array(b)
f(ta, tb)
print(tb)
```
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
