Lunderberg opened a new pull request, #11646: URL: https://github.com/apache/tvm/pull/11646
The GPU schedule for `topi.cuda.schedule_sort` includes expressions that use `tir.ceil`, `tir.log2`, and `tir.shift_left` that remain in the TIR representation through the entire lowering flow. (For example, `@tir.shift_left(2i64, (i_0 + cast(int64, cast(int32, @tir.ceil(@tir.log2(128f64, dtype=float64), dtype=float64)))), dtype=int64)`.) However, these expressions are not currently simplified by `arith::Analyzer`. This PR introduces two changes, such that these expressions can be simplified. * Simplify `ceil(log2(constant))` into a constant integer. * Identify bounds on `x << y`, for use in proving conditionals. -- 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: commits-unsubscr...@tvm.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
