Sam Elliott

Linker relaxation is not supported in LLD. It is supported by binutils ld, which should be tried. For gp-based addressing, you also need to define `__global_pointer$` in the same output...

> Yes, i actually tried with __global_pointer$ pointing to .data and to .rodata to see where it works better. With binutils ld there is another issue that negative offsets are...

This caught me out too, because the `+y` direction of the unit circle for `arc` and the `+y` direction for coordinates do not match, if you use `vsk.size(…, landscape=True, center=False)`....

why not just use a ? and put the object_class.connection.call(:method) in the array...

Yeah, I'm not sure about the `LoopMicroOpBufferSize` which may just be for OoO cores. There is a way to tell the scheduling model you have forwarding, I just haven't worked...

I haven't got around to updating this, but I see you've spun off a new repo for Hazard5 - should I rename the model to give the core that name...

iirc, the target provided at the command line always overrides the triple in the LLVM IR file.

I think the issue is weirder than that. `%llc_dwarf` expands to `llc`, except when config.target_triple is an msvc triple on windows, where `-mtriple=` is passed with the equivalent gnu triple...

Patch here: https://reviews.llvm.org/D136066

To answer 2, function return bounds are specificed in terms of your parameters. I think that you can do the following for getbuf(…): ``` _Nt_array_ptr getbuf(_Ptr p_str) : count(p_str->len) {...