Alphasim cluster size & uncorrected in spm

 First thanks you for your great work!! 

I performed a multiple regression analysis in SPM8 with a behavioral score as vector for a resting state task. 
Then I performed the Alphasim test with T treshold at 0.005 rmm at 3 (for 3x3x3) and here is my doubt. 

When I'm performing the t-test in SPM8, what is the best option to put in treshold (T or p value) --> ""the Standart uncorrected value with 0.001 or the uncorrected value 0.005 used in alphasim ???????""

Thanks in advance! 

Juan Gea, 
Msc Neuroscience Student
at University of Balearic Islands.

 Thank you very much Chao-Gan! But I've another question...

Do you know why, if I'm using different masks, in particular masks I can't select the cluster size and automaticaly generates the graphic, with a cluster size = 0?

edited: Ive reallized that you can put the extend threshold voxel when you choose F as statistical value and 0.005 but not with 0.001, there is any possibility to use 0.001 and get the threshold voxel.

Thank you once again!!

Best, 

Juan Gea.