ldc nvvm GPU intrinsics good news

[Bruce Carneal]

On Friday, 5 March 2021 at 00:16:41 UTC, kinke wrote:
> On Friday, 5 March 2021 at 00:03:26 UTC, Bruce Carneal wrote:
>> After updating the first line to 
>> '@compute(CompileFor.hostAndDevice) module ...' and adding an 
>> 'import ldc.dcompute;' line, the 
>> runtime/import/ldc/gccbuiltins_nvvm.di file from a current LDC 
>> build apparently gives access to all manner of GPU intrinsics.
>
> Sorry, I assumed you had already seen it (plus 
> ldc/gccbuiltins_amdgcn.di); they've been introduced with LDC 
> v1.22. These files are auto-generated from the LLVM .td file 
> using a tiny program [1]. It's currently limited to intrinsics 
> with a GCCBuiltinName, so if you'd like to make more of these 
> available in D, see processRecord().
>
> [1] 
> https://github.com/ldc-developers/ldc/blob/master/utils/gen_gccbuiltins.cpp

So what is the right/clean way to get @compute access to these 
builtins?  Should practitioners do a manual copy edit or is there 
a better way?