Help to speed up CALCULATE formula

Question

I am building a reporting system of production profiles for a set of assets. The key data structure is as follows:
Version  Field               Action    Profile          Date          Production     RiskCategory
v1          Field1         Action00    00_Low      31/1/2014        5000         1
v1           Field1         Action00    00_Low     28/2/2014         4980        1
v1           Field1         Action00    00_Low     31/3/2014          4950       1
v1           Field1        Action00      00_Base     31/1/2014        5300       2
v1           Field1        Action00      00_High     31/1/2014         5600      
3
v1           Field1        Action01      01_Low     31/1/2014        8000        
1
 v2           Field1         Action00    00_Low    31/1/2014          2000         1
Ie. the data consists of production profiles. Each [Version] has many [Field]s who in turn have many [Action]s who in turn have a High, Base, Low [Profile]s who in turn have a [Production] for each [Date].
Data size:
Each profile has a monthly entry from 2014 to 2034. There are typically 10 Actions per field, and maybe 4 fields per version. So number of rows per version is: 12months*30years*3profilesperaction*10actions*4fields = 43200. Number of versions per data set
is not decided yet, however as time goes by there will be a steady increase in nr of versions... it all depends on the speed and practicality.
Related tables:A related table (on Version and Profile) adds [Risk Category] to the [Profile] (and other attributes not relevant here. There is also another related table on [Version] [Field] [Date] not relevant here.
Problem Description: I use PowerPivot to tie 3 tables togetether via composite keys (calculated column = [Version]&[Profile] etc). So I dont have to repeat [RiskCategory] for each date in the profile table and makes QCing inputs far
easier. The purpose of the database is to take the production profiles and slice them in different ways to satisfy different input systems.
The most challenging is that I need to calculate the incremental contribution from the Base and High  [Profiles] in each action.
Generate a calculated column that for each Date in each Action calculates:
if [RiskCategory] (thisrow) = 1, then [Production])
If [RiskCategory] (thisrow) = 2, then [Production] (where [RiskCategory]=2)  - 
                                           [Production]
(where[RiskCategory]=1)
If [RiskCategory] (thisrow) = 3, then [Production] (where [RiskCategory]=3)  - 
                                           [Production]
(where[RiskCategory]=2)
My solution below is a Calculated Column that uses Calculate Allexcept and Earlier to remember the row context. However, the calculation is a little slow: 5 sec maybe - which will get worse as I add more versions and possibly more fields with more actions.
and I would like to see if I can speed it up:
=if( Profile[RiskCategory]>1,
Calculate(SUM(Profile[Production]), Allexcept(Profile,Profile[Date],Profile[Version],Profile[Action]), Profile[RiskCategory] = Earlier(Profile[RiskCategory]))-
Calculate(SUM(Profile[Production]), Allexcept(Profile,Profile[Date],Profile[Version],Profile[Action]), Profile[RiskCategory] = Earlier(Profile[RiskCategory])-1)
,Profile[Production])
Any suggestions?
Hope this is clear.
Thanks
Lars

Answer 1

Thanks Avi and Greg,
I am now guessing the sluggishness is coming from the size of the pivot table that I am using:
Columns: Year, Month (ie 12X30 columns), Rows: Version, Field, Action (1x4x10), Values: 4 different measures based on production and delta prodcution. In all it makes for almost 60000 cells.
Data source queries and relationships
The data sits in tables in excel and is organized in a "star" schema to allow the users to input and maintain the data easier. ie without having to copy all the duplicate columns in a flat file. 
I use composite keys and avoid autokeys etc to avoid data entry being tedious and unintelligible. Remember these tables are used by the users to input and store production profiles.
Dax Queries I know nothing about, would they work for my set up? ie without an external sql database etc to query.
Use measure in stead of calculated column
I would like to see if I could make this a measure instead of a column. However, I am struggling to see how as I need multiple references to the current row context (hence why the ability of using Earlier() in the Calculated column was handy. Maybe a recursive
SUMX would do it?
To summarize: The calculation requires the computation between the current [production] and the [production] in the other [RiskCategoriy]s in the same context ie: same [DATE], [ACTION], [FIELD], [VERSION] 
ie: if RiskCateory is 1, just return current [production], if it is 2 or 3 then return Production (2) - Production (1) or Production(3)-Production(2) respectively (belonging to same version/field/action 
Example under of the calculated column*
Version  Field    Action    Profile          Date       Production RiskCategory   Calculation*
v1         Field1  Action00    00_Low   31/1/2014      5000         1                   5000
v1         Field1   Action00    00_Base  31/1/2014      5400        2                    400
v1        Field1    Action00    00_High   31/1/2014      6800       3                    1400
Hope this clarifies a little and I will endeavour to paste my workbook here so you can take a look for yourself.  
Thanks again!
Lars