I have a table with over 97M rows and a primary key:
(px int,
sx int
effectivedate date)

I need to extract rows for each px and sx for the latest date.  We have
a view :
select a.px, a.sx, a.effectivedate, a.cost, a.suggretailprice,
a.cost_pst,
   a.marginpercent, a.quantity
 from mx.productcost a
 where a.effectivedate = (
     select max(b.effectivedate)
       from mx.productcost b
       where a.px = b.px
         and a.sx = b.sx);

Access plan is:
Access Plan

Step Operation                           Object name
  Object type     Cost
---- ----------------------------
-------------------------                      -----------
-------------
1    Table access full                    PRODUCTCOST
   Table     6320322.50
2    Index scan (no table access)   SYSIBM.SQL031013180223470 Index
36.96
3    Group rows
                              36.96
4    Filter data
                                 36.96
5    Nested loop join
                             1183707136.00
6    Data operation complete
                        1183707136.00

I have tried indexes of (px asc, sx asc, effectivedate desc) and (px
asc, sx asc, effectivedate asc) with 'allow reverse scan'  but all give
very poor performance (no results over many hours).

Any suggestions on an efficient way to get at the latest rows of the
key?

thanks a lot in advance.