I want to know why this is the result of CVX. Why unbounded and then solved?
what 663/663 8 1.277e1 mean?
That is the output os CVX’s Successive Approximation method. Dodn’t worry about anything other than the final status reported by CVX.
That said, for this problem, you are better off using Mosek as solver, or if that is not available to you, following the advice in CVXQUAD: How to use CVXQUAD's Pade Approximant instead of CVX's unreliable Successive Approximation for GP mode, log, exp, entr, rel_entr, kl_div, log_det, det_rootn, exponential cone. CVXQUAD's Quantum (Matrix) Entropy & Matrix Log related functions
OK. Thank you. 
From the results, CVX seems to transform the convex optimization problem into the form of cone programming (i.g.,EXP cone, Poly Cone). Then I am curious about how CVX solves the cone programming problem, is it the interior point method?
CVX calls a solver, which (usually use) conic (interior point) optimization methods (perhaps Simplex or active-set for some (MI)LP and (MI)(QC)QP).
In the case of Successive Approximation method, CVX calls a conic solver multiple times to solve a sequence of problems.
Thank you for your patience in answering my question and I don’t have any doubts now.