@twmht Hi, if you want to run on CUDA, you would need to do `tune_tir` with the target being `"cuda"`, and also do `compile_tir` with the target being `"cuda"` (or substitute `"cuda"` with a specific target string in the list here https://github.com/apache/tvm/blob/main/src/target/tag.cc#L126-L378 (e.g., `target="nvidia/geforce-rtx-3090-ti"`).
The code snippet you shared is exclusive for LLVM and is not applicable to CUDA. Hope this can help. --- [Visit Topic](https://discuss.tvm.apache.org/t/in-tvm-do-we-have-to-manually-specify-a-schedule/14842/6) to respond. You are receiving this because you enabled mailing list mode. To unsubscribe from these emails, [click here](https://discuss.tvm.apache.org/email/unsubscribe/b0df6d2ff5b4e4248b6c49c3b91e5e9d58f52ac6c1b0f80a8974ab43e4187db7).
