wrongtest commented on pull request #8803:
URL: https://github.com/apache/tvm/pull/8803#issuecomment-903573759
@junrushao1994 Sorry for misleading information. What I tried is like
```python
func = param_in_arith_exprs.specialize({b: tir.decl_buffer([16])})
func = tvm.tir.transform.Simplify()(tvm.ir.IRModule({"main": func}))["main"]
tvm.ir.assert_structural_equal(func, expect)
```
I found the check failure is not becuase 'do not recognize "frontend" ir
nodes"', but at the function's buffer_map field, where there is constant shape
arithmecis can be fold but seems not visisted by the pass. The simplification
to body is well performed.
```
buffers = {A: Buffer(A_1: Pointer(global int32), int32, [floordiv(16, 8),
8], []),
B: Buffer(B_1: Pointer(global int32), int32, [16], [])}
```
--
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.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
