ldc/dcompute and shared (programmer managed cache) access

[kinke]

On Thursday, 4 March 2021 at 04:52:52 UTC, Bruce Carneal wrote:
> As a slightly cleaner alternative I'll next look at using 
> ldc/dcompute as a .o and .ptx generator while punting the 
> fatbin/linking stuff to clang or nvcc in a build script.  The 
> current simplicity of single-ptx-file is very nice but 
> foregoing shared memory performance boosts is not nice so ...  
> I'm pretty sure that we'll need to move beyond the single ptx 
> file model if we want to embrace shared cleanly.

Not sure if it's of any help, but using something like 
`-mtriple=nvptx64 -mcpu=sm_50` might suffice: 
https://github.com/ldc-developers/ldc/pull/3411#issuecomment-619385886