mdx distinct count performance -

i got mdx query below returns average distinct count last 30 days , returns correct result, performance not , in fact table there 2,500,000 rows data everyday.

with member measures.dailyaverageuser     avg ( existing{[date].[date].[date].members},                  [measures].[active tiles employee id distinct count]                 ),format_string="##.000000"              select   measures.dailyaverageuser on 0,             non empty([action targets].[name].children)  on 1             [olap pre]             (                 {[target types].[name].&[tile]},{[employee statuses].[status id].&      [1],[employee statuses].[status id].&[3],[employee statuses].[status id].&[4]},             {[business region].[abbreviation].&[none],[business region].[abbreviation].&[ams],[business region].[abbreviation].&[apj],[business region].[abbreviation].&[emea]},{[employee types].[bits].&[1],[employee types].[bits].&[5],[employee types].[bits].&[9],[employee types].[bits].&[25],[employee types].[bits].&[13],[employee types].[bits].&[29]},{[date].[date key].&[20150428]:null}) 

i changed query this， still same problem， thing 1 cache data between 20150409 ：null， when query 20150410：null， fast.

with member measures.dailyaverageuser              avg ( {[date].[date key].&[20150409]:null},                  [measures].[active tiles employee id distinct count]                 ),format_string="##.000000"              select   measures.dailyaverageuser on 0,             non empty([action targets].[name].children)  on 1             [olap pre]             (             {[target types].[name].&[tile]},{[employee statuses].[status id].&[1],[employee statuses].[status id].&[3],[employee statuses].[status id].&[4]},             {[business region].[abbreviation].&[none],[business region].[abbreviation].&[ams],[business region].[abbreviation].&[apj],[business region].[abbreviation].&[emea]},             {[employee types].[bits].&[1],[employee types].[bits].&[5],[employee types].[bits].&[9],[employee types].[bits].&[25],[employee types].[bits].&[13],             [employee types].[bits].&[29]}) 

need suggestion on performance

do want keep in slicer axis? if not, can push member's definition -

with member measures.dailyaverageuser              avg (                         nonempty(                         (                         {[target types].[name].&[tile]},                         {[employee statuses].[status id].&[1],                          [employee statuses].[status id].&[3],                          [employee statuses].[status id].&[4]},                         {[business region].[abbreviation].&[none],                         [business region].[abbreviation].&[ams],                         [business region].[abbreviation].&[apj],                         [business region].[abbreviation].&[emea]},                         {[employee types].[bits].&[1],                         [employee types].[bits].&[5]                         ,[employee types].[bits].&[9]                         ,[employee types].[bits].&[25]                         ,[employee types].[bits].&[13]                         ,[employee types].[bits].&[29]},                         {[date].[date key].&[20150428]:null}                         ), [measures].[active tiles employee id distinct count]                     )                          ,                  [measures].[active tiles employee id distinct count]                 ),format_string="##.000000"  select   measures.dailyaverageuser on 0, non empty([action targets].[name].children)  on 1 [olap pre]