CUDA: enable Gemma FA for HIP/Pascal#9581

JohannesGaessler merged 1 commit intoggml-org:masterggml-org/llama.cpp:masterfrom

JohannesGaessler:hip-enable-gemma-fattnJohannesGaessler/llama.cpp:hip-enable-gemma-fattnCopy head branch name to clipboard