feat: cache efficient CPU kernel#23

Merged

ryan-berger merged 27 commits intomainfrom

rberger/cpu-clean

Jan 17, 2026

Commits on Jan 9, 2026

feat: cache efficient CPU kernel
ryan-berger
committed

Commits on Jan 17, 2026

feat: conditionally compile based on architecture
ryan-berger
committed
feat: add inclusive prefix sum code (#25)
ryan-berger
committed
feat: add a method of creating a default unroll
ryan-berger
committed
fix: make main kernel loop range exclusive (#30)

tombh
authored and
ryan-berger
committed
tests: Integrate viewsheds tests into CPU kernel (#29)

tombh
and
ryan-berger
committed
fix: avx512 build issues
ryan-berger
committed
fix: flakey longest line tests, add thread count command line flag
ryan-berger
committed
feat: add better rasterization of valid TVS
ryan-berger
committed
feat: add `highest` carry-through back in removing unnecessary max instruction
ryan-berger
committed
fix: 16-wide float comparison
ryan-berger
committed
refactor: a more robust line of sight trait, couple unrolling and vectorization
ryan-berger
committed
fix: duplicate total surface tallying
ryan-berger
committed
fix: remove asserts, fix refraction
ryan-berger
committed
fix: single thread tests, output areas
ryan-berger
committed
refactor: remove references to `sector_surfaces`

tombh
authored and
ryan-berger
committed
benchmarks: unify cpu and vulkan-cpu viewshed expectation

tombh
authored and
ryan-berger
committed
ci: bump Rust version to 1.92

tombh
authored and
ryan-berger
committed
fix: forward longest lines precedence over backward lines

tombh
authored and
ryan-berger
committed
chore: remove mod.rs, delete unneeded script, fix constant
ryan-berger
committed
chore: update benchmark to use 48-wide Cardiff
ryan-berger
committed
chore: remove cargo toml for portability
ryan-berger
committed
chore: document rusttoolchain file
ryan-berger
committed
chore: add refraction tests
ryan-berger
committed
feat: make unrolled angle calculation for line of sight straightlined
ryan-berger
committed
chore: update unroll factor based on new algorithm
ryan-berger
committed
chore: add option to disable image rendering
ryan-berger
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

feat: cache efficient CPU kernel#23

feat: cache efficient CPU kernel#23
ryan-berger merged 27 commits intomainfrom
rberger/cpu-clean

Commits on Jan 9, 2026

feat: cache efficient CPU kernel

Commits on Jan 17, 2026

feat: conditionally compile based on architecture

feat: add inclusive prefix sum code (#25)

feat: add a method of creating a default unroll

fix: make main kernel loop range exclusive (#30)

tests: Integrate viewsheds tests into CPU kernel (#29)

fix: avx512 build issues

fix: flakey longest line tests, add thread count command line flag

feat: add better rasterization of valid TVS

feat: add `highest` carry-through back in removing unnecessary max instruction

fix: 16-wide float comparison

refactor: a more robust line of sight trait, couple unrolling and vectorization

fix: duplicate total surface tallying

fix: remove asserts, fix refraction

fix: single thread tests, output areas

refactor: remove references to `sector_surfaces`

benchmarks: unify cpu and vulkan-cpu viewshed expectation

ci: bump Rust version to 1.92

fix: forward longest lines precedence over backward lines

chore: remove mod.rs, delete unneeded script, fix constant

chore: update benchmark to use 48-wide Cardiff

chore: remove cargo toml for portability

chore: document rusttoolchain file

chore: add refraction tests

feat: make unrolled angle calculation for line of sight straightlined

chore: update unroll factor based on new algorithm

chore: add option to disable image rendering