Improved performance of boolean comparison (5x-14x)

[jorgecarleitao]

The idea is that we can compare full bytes instead of iterating over individual bits, replacing Fn(bool,bool) -> bool by Fn(u8,u8) -> u8 with corresponding bit operations (e.g. a != b -> a ^ b, a < b -> !a & b).

git checkout 0893100640f22dfc5fec4dede300f63fe095b170
cargo bench --no-default-features --features benchmarks,compute --bench comparison_kernels -- "bool.*2\^20"
git checkout cmp_boolean
cargo bench --no-default-features --features benchmarks,compute --bench comparison_kernels -- "bool.*2\^20"

bool 2^20               time:   [1.0651 ms 1.0719 ms 1.0817 ms]                       
                        change: [-78.884% -78.378% -77.783%] (p = 0.00 < 0.05)

bool scalar 2^20        time:   [212.37 us 213.76 us 215.72 us]                             
                        change: [-93.123% -93.040% -92.965%] (p = 0.00 < 0.05)

[codecov]

Codecov Report

Merging #318 (8ba79ff) into main (8243861) will increase coverage by 0.01%.
The diff coverage is 85.18%.

@@            Coverage Diff             @@
##             main     #318      +/-   ##
==========================================
+ Coverage   80.52%   80.54%   +0.01%     
==========================================
  Files         324      324              
  Lines       21369    21383      +14     
==========================================
+ Hits        17208    17222      +14     
  Misses       4161     4161              

Impacted Files	Coverage Δ
src/compute/comparison/boolean.rs	69.14% <85.18%> (+5.39%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8243861...8ba79ff. Read the comment docs.