How to reduce cplexmiqp computation time for Electrical UC Problem?

I'm trying to solve electrical unit commitment problem (which generator should be on/off and how much power on each generator should generated). On or off status is binary and power in quadratic cost function. The problem is in MIQP and I'm using CPLEX MIQP 12.8 by add CPLEX to the search path:

C:\Program Files\IBM\ILOG\CPLEX_Studio128\cplex\matlab\x64_win64

I already run this two code:

[x,y]=cplexmiqp(F,c,Ai,bi,Ae,be,[],[],[],x_L,x_U,IV,[]); %first code

options = cplexoptimset();
[x,y]=cplexmiqp(F,c,Ai,bi,Ae,be,[],[],[],x_L,x_U,IV,[],options); %second code

The funny thing is they both have different computational time with same x and y result. I thought

options = cplexoptimset()

will give default or auto setting to cplexmiqp, but it seems not(?)

I also has trying to reduce the accuracy by:

options = cplexoptimset('Display','off','TolFun',1e-3);

but, it's still has more computational time than the first code (first code 7s and second code 40s, in 40 units). My question is how to reduce cplexmiqp computation time for Electrical UC Problem (we can sacrifice accuracy)?

Further note: Any tips is accepted whether it is input parameter, matlab setting, or even purchasing matlab toolbox (note that I already own optimization toolbox). The maximum matrices i can run for now is 60 units (20 minutes).

[Quang-Trung Luu]

Perhaps try to set options.MaxTime and options.MaxNodes of cplexoptimset?