Memory access or calculate address?

[dbelanga]

Dear all,
I'm working on a design with FPGA & image processing. I need to access image data at certain pixels. The pixel data can be accessed using address which use certain equation that might use several multipliers and other components. However, a precalculated number can be used in exchage to the equation and stored to memory. My question is what is the pros & cons of this two approaches? Which one would take longer time, access to memory (external/internal) or calculate the address?


example:
for m=1:10
for n=1:5000
equationA=3*n; % can be more complex
img(equationA)
end
end

or save the value of equationA (3,6,9,12,15....15000)and straight access memory. No need to calculate equationA
for m=1:10
img(3 6 9 12 15....15000)
end

 

[aquakeerthi]

actually inadvertent data's man what what method do you use in case like what sort of sequence are you using then time cycles everything has to define man

 

[TrickyDicky]

It is really going to depend on the equation, and the type of memory being used. Pure random access can mean you have to use srams which are fairly small. if you can cope with a pipelined design and a fairly logical addressing sequence, then large drams can be used.