Manuel Drehwald

@jedbrown with the latest updates in enzyme core, the second version (increments sum) now works. The first one still gives an incorrect TA update though.

@jedbrown All combinations now work in dbg and release mode, so closing this.

Iirc pretty hard and only a very limited benefit. We would literally have to hardcode this case (two f32 getting passed as one f64 or i64, I don't remember which)....

Yep, but it's one of the cases where a hack will take time and has it's own downsides, so I currently still lean towards a more proper fix, where I...

Turns out this was just caused by our back then non-existing sret handling, fixed by https://github.com/rust-lang/rust/pull/139465, so closing here. For the bigger ABI issues which still remain I proposed a...

Is this something Enzyme core should understand, or what would be the fix here?

[bug2.ll.txt](https://github.com/user-attachments/files/16434000/bug2.ll.txt) Unfortunately the reduction operator is passed in as pointer. Outside of the function being differentiated, this pointer got created by ``` %210 = load ptr, ptr @RSMPI_SUM, align 8,...

The first one is too far in the Enzyme core weeds, I have no clue how to write that code, so I'll leave it to someone else. If you have...

@wsmoses ``` updating analysis of val: %1 = alloca i128, align 8 current: {} new {[-1]:Pointer} from %1 = alloca i128, align 8 Changed=1 legal=1 updating analysis of val: %2...

bump @wsmoses. Could the assertion be hit due to an Enzyme or rather LLVM Bug?