Function test_fence

tests/python/tirx/codegen/test_codegen_blackwell.py:97–105

(A: T.Buffer((16, 16), "float16"))

    # fmt: off
    @T.prim_func
    def test_fence(A: T.Buffer((16, 16), "float16")):
        T.device_entry()
        cta_id = T.cta_id([1])
        warp_id = T.warp_id([4])
        lane_id = T.lane_id([32])
        tid = T.thread_id([128])
        T.ptx.tcgen05.fence.before_thread_sync()
        T.ptx.bar.sync(0, 32)
        T.ptx.tcgen05.fence.after_thread_sync()
    # fmt: on

    target = tvm.target.Target("cuda")