Skip to content

automata: Use built-in bitCount #10

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
SuperSamus wants to merge 1 commit into
base: main
Choose a base branch
from
Open

automata: Use built-in bitCount #10

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 21 additions & 32 deletions Assets/Compute Shaders/automata.acompute
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -75,17 +75,6 @@ uint64_t encode_tile_to_uint(ivec2 uv) {
return tile_bits;
}

uint sum_uint_bits(uint64_t i) {
uint sum = 0;

while (i != 0) {
sum += uint(i & 1);
i >>= 1;
}

return sum;
}


[numthreads(8, 8, 1)]
void Automaton() {
Expand Down Expand Up @@ -134,29 +123,29 @@ void Automaton() {
uint64_t lower_left_quadrant_bits = encode_tile_to_uint(uv + ivec2(0, 7));
uint64_t lower_right_quadrant_bits = encode_tile_to_uint(uv + ivec2(7, 7));

uint neighbor_count1 = 0;
neighbor_count1 += sum_uint_bits(upper_left_quadrant_bits & upper_left_mask);
neighbor_count1 += sum_uint_bits(upper_right_quadrant_bits & upper_right_mask);
neighbor_count1 += sum_uint_bits(lower_left_quadrant_bits & lower_left_mask);
neighbor_count1 += sum_uint_bits(lower_right_quadrant_bits & lower_right_mask);
uint64_t neighbor_count1 = 0;
neighbor_count1 += bitCount(upper_left_quadrant_bits & upper_left_mask);
neighbor_count1 += bitCount(upper_right_quadrant_bits & upper_right_mask);
neighbor_count1 += bitCount(lower_left_quadrant_bits & lower_left_mask);
neighbor_count1 += bitCount(lower_right_quadrant_bits & lower_right_mask);

uint neighbor_count2 = 0;
neighbor_count2 += sum_uint_bits(upper_left_quadrant_bits & upper_left_mask2);
neighbor_count2 += sum_uint_bits(upper_right_quadrant_bits & upper_right_mask2);
neighbor_count2 += sum_uint_bits(lower_left_quadrant_bits & lower_left_mask2);
neighbor_count2 += sum_uint_bits(lower_right_quadrant_bits & lower_right_mask2);
uint64_t neighbor_count2 = 0;
neighbor_count2 += bitCount(upper_left_quadrant_bits & upper_left_mask2);
neighbor_count2 += bitCount(upper_right_quadrant_bits & upper_right_mask2);
neighbor_count2 += bitCount(lower_left_quadrant_bits & lower_left_mask2);
neighbor_count2 += bitCount(lower_right_quadrant_bits & lower_right_mask2);

uint neighbor_count3 = 0;
neighbor_count3 += sum_uint_bits(upper_left_quadrant_bits & upper_left_mask3);
neighbor_count3 += sum_uint_bits(upper_right_quadrant_bits & upper_right_mask3);
neighbor_count3 += sum_uint_bits(lower_left_quadrant_bits & lower_left_mask3);
neighbor_count3 += sum_uint_bits(lower_right_quadrant_bits & lower_right_mask3);
uint64_t neighbor_count3 = 0;
neighbor_count3 += bitCount(upper_left_quadrant_bits & upper_left_mask3);
neighbor_count3 += bitCount(upper_right_quadrant_bits & upper_right_mask3);
neighbor_count3 += bitCount(lower_left_quadrant_bits & lower_left_mask3);
neighbor_count3 += bitCount(lower_right_quadrant_bits & lower_right_mask3);

uint neighbor_count4 = 0;
neighbor_count4 += sum_uint_bits(upper_left_quadrant_bits & upper_left_mask4);
neighbor_count4 += sum_uint_bits(upper_right_quadrant_bits & upper_right_mask4);
neighbor_count4 += sum_uint_bits(lower_left_quadrant_bits & lower_left_mask4);
neighbor_count4 += sum_uint_bits(lower_right_quadrant_bits & lower_right_mask4);
int64_t neighbor_count4 = 0;
neighbor_count4 += bitCount(upper_left_quadrant_bits & upper_left_mask4);
neighbor_count4 += bitCount(upper_right_quadrant_bits & upper_right_mask4);
neighbor_count4 += bitCount(lower_left_quadrant_bits & lower_left_mask4);
neighbor_count4 += bitCount(lower_right_quadrant_bits & lower_right_mask4);


// for (int x = -1; x <= 1; ++x) {
Expand Down Expand Up @@ -217,4 +206,4 @@ void Blit() {
float automata = imageLoad(_AutomatonFrom, ((uv + _Offset * _Zoom) / _Zoom) % 1024).r;

imageStore(_RenderTarget, uv, vec4(automata, automata, automata, 1.0));
}
}