`add`, `sub`, `mul`, `fma` getting added in PTX ISA 9.4 for sm_100a (Blackwell) and sm_103a (Blackwell Ultra): https://docs.nvidia.com/cuda/developer-preview/13.4/parallel-thread-execution/index.html#alternate-floating-point-instructions Related: https://github.com/JuliaGPU/cuTile.jl/issues/240
add,sub,mul,fmagetting added in PTX ISA 9.4 for sm_100a (Blackwell) and sm_103a (Blackwell Ultra):https://docs.nvidia.com/cuda/developer-preview/13.4/parallel-thread-execution/index.html#alternate-floating-point-instructions
Related: JuliaGPU/cuTile.jl#240