# Documentation Mapping DIS Stressmark Parameters to WSS
This details exactly how the input parameters of each Stressmark translate to
their dynamic allocations size. With some algebra, we then convert those
equations such that we can determine the needed parameters for a specific
desired dynamic allocation size.

The equations teased out here are implemented in `gen_input.py`.

## Field
1 allocation in main()
f = 1st input param

sizeof(unsigned char) * f

## Matrix
6 allocations in main(), 7 allocations in biConjugateGradient()

*Allocations in main()*
dim = 2nd input param
numberNonzero = 3nd input param

sizeof(double) * (dim^2+3dim+numberNonzero) + sizeof(int) * (2dim+1+numberNonzero)

*Allocations in biConjugateGradient()*
sizeof(double) * 7dim

*Vectors*
sizeof(double) * 8dim
*Matrix*
sizeof(double) * dim * dim
*Helpers*
sizeof(int) * dim+1
sizeof(int) + (nnZR + dim)
sizeof(double) + (nnZR + dim)

## Neighborhood
1 allocation in createImage, 2 allocations in neighborhoodCalculation

*Allocations in createImage()*
dimension = 3rd input param

sizeof(short int) * dimension^2

*Allocations in neighborhoodCalculation()*
bitDepth = 2nd param

sizeof(int) * (2^(bitDepth + 1) - 1)

## Pointer
n = 5th input param
f = 1st input param

sizeof(unsigned int) * 4n + sizeof(int) * f

## Transitive
n = 1st input param

sizeof(unsigned int) * 2n^2

## Update
f = 1st input param

sizeof(int) * f

## Testplan
*Problem!* Larger WSS = more computations
Use testcase #1 for non-specified parameters
Below math computed for x86_64
- Test WSS at powers of 2: 16 KiB, 32, 64, 128, 256, 512, 1MiB, 2, 4, 8, 16, 32
- For each WSS, measure cache allocation of 0, 1, 2, 4, 8, 16

### Field
Just vary first param

f = WSS

### Matrix
0.3 - 16% number nonzero
- Fixed at 8%
Just vary dim (matrix size)

sizeof(double) * (dim^2+10dim+numberNonzero) + sizeof(int) * (2dim+1+numberNonzero) = WSS


### Neighborhood
8 or 15 bit depth
- Fix at 12?
Just vary dim (image size)

### Pointer
10 for n
Just vary f

### Transitive
Just vary n

### Update
Just vary f