tvm_tirx.backend.cuda.op.cta_min

tvm_tirx.backend.cuda.op.cta_min(value, num_warps, scratch)

Convenience wrapper: cta_reduce(value, "min", num_warps, scratch).