Thanks for the proposal. I agree that it is a valuable problem for dynamic
shape.
Here are two questions from me:
1. Is it necessary to rewrite `(d1*d2)*d0` into `d0*d1*d2`. Can we prove them
equal by `Analyzer` directly?
2. Can we embed the new rule into `tir.Simplify` rather than create a n
We are working on Dyn-shape model compilation with Relay VM, and currently
running into this issue, although this redundant Expr may be eliminated through
compilation with clang/gcc O2 option (@lygztq has located the LLVM::EarlyCSE
pass may help), but we hope to fix it within TIR scope, avoid
Took a quick pass to the updated RFC. I think it's almost ready to merge as
long as the last 3 comments are resolved.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/apache/tvm-rfcs/pull/6#issuecomment-899
okay TQ has an explanation here: https://github.com/apache/tvm/pull/8719
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/apache/tvm/pull/8705#issuecomment-899656015
Meta scheduler (autotir) has not been full merged yet. I believe it will work
on sparse workloads, but I don't know how well it compares to the hand written
kernels.
---
[Visit
Topic](https://discuss.tvm.apache.org/t/sparse-opencl-error-scheduling-sparse-computations-that-use-tir-ir-build
## Background and Motivation
Currently, TVM uses the `tir.Simplify` pass to remove some redundant expression
like nested equivalent if-condition. For example, given a simple softmax
operation like
```
primfn(placeholder_1: handle, T_softmax_norm_1: handle) -> ()
attr = {"from_legacy_te_sche
Merged #8705 into main.
--
You are receiving this because you are subscribed to this thread.
Reply to this email directly or view it on GitHub:
https://github.com/apache/tvm/pull/8705#event-5162861599
