Set TORCH_CUDA_ARCH_LIST to 8.0 and above

Only compiling on Compute Capability 8.0 and above, see https://developer.nvidia.com/cuda-gpus. I.e. NVIDIA Ampere generation devices or newer.

recipes/flash-attn/meta.yaml

@@ -17,6 +17,7 @@ build:
     - FLASH_ATTENTION_SKIP_CUDA_BUILD=FALSE
     - FLASH_ATTENTION_FORCE_CXX11_ABI=FALSE
     - MAX_JOBS=$CPU_COUNT
+    - TORCH_CUDA_ARCH_LIST="8.0;8.6;8.9;9.0+PTX"
   skip: true           # [cuda_compiler_version in (undefined, "None")]
   skip: true           # [not linux]