Performance wise data representation

Question

Hello there,
I would like to start discussion regarding performance wise solution implementation solution,
From performance point of view, which choice would be better to be chosen among the following scenarios,
1. one is to run a process run on every and each record in the a Huge table, for example with million records, for all the records,
2. or to have another table which might have about 800 fields !! which might represents a map crossing the id values in the main Huge table, which must be filled based on the values in the main table!! as a pre-processing
then instead of processing the Huge table, simple query the map table which could be indexed using the needed values,
what is the order of causes of bad performance, a process on many records, or many records that represents a pre-processing
many thanks

Answer 1

Thank you Billy for replying,
Billy  Verreynne  wrote:
Nor is performance something that one looks at after the design is done and the code written. Performance is a primary factor that needs to be considered with the h/w bought, the s/w installed and configured, the design, and every single line of code written.yes, currently I am in the design phase so that I am trying to understand the major performance principles that might effect the software when dealing with huge amount of data , whether pre-processing would be better, and such implementation issues.
Here is the case logically:
The process I have mentioned in the post, corresponds to a procedure that must be applied on data from a table, then return a certain value, calculated on the fly.
some of those processing might not needed to be done , so in order to avoid huge unnecessary operation I need to perform some kind of predicating or indexing based on certain values.
what is the best practice for such scenarios, performance wise!
Thanks