Is it possible to get the intermediate ttir and ttgir on a non-CUDA machine?

[jamestclin]

Hello,

I have limited CUDA GPU resources. So, I'm wondering if it is possible to let the Triton Compiler to run compilation from Python code and then end at the generation of either ttir or ttgir. If yes, can we do this on the machine without CUDA enabled?

Thanks very much,
James