-
Notifications
You must be signed in to change notification settings - Fork 12.6k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
aarch64 backend OOM on -O0 when compiling llvm.fptosi.sat.v3i32.v3f32 #94694
Comments
I ran it under valgrind and when i gave up it had used >4GB of ram which is absurd for a function of this size |
@llvm/issue-subscribers-backend-aarch64 Author: Jacob Lifshay (programmerjake)
https://llvm.godbolt.org/z/1Ehsh97nP
```llvm
target datalayout = "e-m:o-i64:64-i128:128-n32:64-S128"
target triple = "arm64-apple-macosx11.0.0"
define internal void @f(ptr sret([12 x i8]) align 4 %_0, ptr align 4 %self) unnamed_addr {
|
valgrind massif heap tree shortly before I killed it:
|
I have been looking at adding fptosi.sat support recently. Currently GISel doesn't handle them, so this will be treated like a generic intrinsic and the real problem might only be tangentially related to them. |
This should be fixed for fptosi.sat, as we legalize those operations correctly after #96297. IIRC There might be other intrinsics that hit the same issue with odd vector types, please let us know and reopen (or create a new issue) if you see any. Thanks. |
https://llvm.godbolt.org/z/1Ehsh97nP
@calebzulawski
originally encountered in Rust's project-portable-simd rust-lang/portable-simd#422 (comment)
The text was updated successfully, but these errors were encountered: