Jason Lowe-Power
Jason Lowe-Power
If you do this with an optional probe, I think it's a great idea. Please open a PR.
Seems feasible to get it in. I'm excited for these new features!
Have you tried setting "allow_hybrid" to false on the KVM object?
Related PRs: - #1127 (revert 1065) - #1138 (some other fix?) - #1065 (enable KVM on hybrid architectures) - #1040 (some fix?) The data we have right now is that...
@Harshil2107 can you run with Matt's new PR (#1138) and see what happens? If all cases fail with #1065, I think we should revert it until @nmosier can resubmit.
Letting @giactra and @tiagormk know that you're working on this.
Thanks for the bug report! You've done a great job explaining the issue!
I'm going to push back on this solution because the spec says that `vset{i}vl{i}` is a control dependency, not a data dependency. See https://github.com/riscv/riscv-v-spec/blob/v1.0/v-spec.adoc#9-vector-memory-consistency-model Correct me if I'm not interpreting...
It's an intriguing option. I also think looking into speculating on `vl` would also be an interesting direction. I think I know the answer to this question, but I'll ask...