Optimization Report — lammps-neighbor¶
Note
The optimized code summarized in this report was generated by the FermiLink AI agent. Review and validate the code changes yourself before using the modified code in scientific or production work. This optimization reporting feature is experimental and is not a final, mature solution.
Primary metric: Weighted median neigh time (s) (lower is better).
Goal¶
Copied source goal for this optimization: goal.md
# Optimization Goal
## Package
lammps
## Language
cpp
## Target
Optimize the neighbor-list construction hot path of the TIP4P long-range water NVE workflow in LAMMPS, with primary focus on rebuild-time binning and pair-list generation for the `neighbor 2.0 bin` configuration used by the benchmark inputs.
In `src/verlet.cpp`, the TIP4P NVE loop calls `neighbor->decide()` every step and `neighbor->build(1)` whenever reneighboring is required. For this input deck, the relevant hot paths are `src/neighbor.cpp` and `src/npair_bin.cpp`: `Neighbor::build()` stores the displacement reference state, bins all local and ghost atoms, builds the perpetual pair lists, and rebuilds topology lists, while `NPairBin::build()` loops over owned atoms and bin stencils, evaluates cutoffs and exclusions, and emits the neighbor lists consumed by the pair style.
Primary optimization interest is reducing neighbor rebuild cost for the fixed-size water systems while preserving the same list completeness, rebuild safety, special-neighbor handling, and stable NVE behavior.
This goal assumes benchmark generation will use the attached input artifacts by filename (`water_216_data.lmp`, `in.tip4p_nve`, and `in.tip4p_nve_long`) and resolve them from the staged goal input root.
## Editable Scope
- src/neighbor.cpp
- src/neighbor.h
- src/npair_bin.cpp
- src/npair_bin.h
## Performance Metric
Minimize weighted median `neigh_seconds` across all benchmark cases.
Benchmark should also record `loop_seconds`, `neigh_seconds`, `pair_seconds`, `kspace_seconds`, `comm_seconds`, `bond_seconds`, and normalized throughput (for example, steps/second or ns/day). Secondary objective should be lower `loop_seconds` without winning by skipping required rebuilds or weakening list safety.
## Correctness Constraints
- Preserve NVE energy behavior: total energy drift per atom per step over the longer runs must stay within benchmark tolerance versus incumbent baseline.
- Preserve sampled thermo observables at matched output steps: `etotal`, `pe`, `ke`, `temp`, `press`, and `density` must stay within benchmark tolerance.
- Preserve sampled force consistency for representative frames: RMS and max absolute force-component deltas must stay within benchmark tolerance.
- Preserve neighbor semantics exactly: same pair-list completeness, same special-neighbor encoding and exclusion behavior, same topology-list correctness for bonds and angles, and no dangerous builds or out-of-range atoms introduced by the optimization.
- Do not change physical model semantics or runtime controls to gain speed: keep `pair_style lj/cut/tip4p/long`, `kspace_style pppm/tip4p 0.0001`, `neighbor 2.0 bin`, default rebuild-safety behavior, `timestep 0.5`, units, and TIP4P geometry assumptions unchanged.
- Do not weaken distance checks, ghost coverage, or rebuild cadence safety to gain speed.
- All benchmark cases must complete successfully with deterministic runner settings.
## Representative Workloads
- train-16r-long: `in.tip4p_nve_long` + `water_216_data.lmp` on 16 MPI ranks, giving a longer run with repeated neighbor rebuild opportunities and lower communication noise.
- train-32r-long: `in.tip4p_nve_long` + `water_216_data.lmp` on 32 MPI ranks to keep the optimization useful across a second domain decomposition.
- train-32r-short: `in.tip4p_nve` + `water_216_data.lmp` on 32 MPI ranks as a shorter-turnaround rebuild case.
- test-16r-short: `in.tip4p_nve` + `water_216_data.lmp` on 16 MPI ranks as a held-out lower-rank case.
- test-64r-short: `in.tip4p_nve` + `water_216_data.lmp` on 64 MPI ranks as a held-out scaling-sensitive case where neighbor and communication interact more strongly.
## Build
```bash
mkdir -p build
cd build
cmake -C ../cmake/presets/most.cmake -C ../cmake/presets/nolib.cmake -D PKG_GPU=off ../cmake
cmake --build . -j4
```
## Notes
- Treat the attached LAMMPS input file(s) as the source of truth for runtime settings and any include-chain files.
- This campaign is intended to find algorithm-level improvements inside `src/neighbor.*` and `src/npair_bin.*`, not generic pair, PPPM, communication, or bonded-kernel tuning.
- Keep benchmark execution deterministic: fixed thread settings, fixed random seeds (if any), and explicit launch command.
- Run LAMMPS with full timer output so the benchmark runner can parse `Neigh`, `Pair`, `Kspace`, `Comm`, `Bond`, and total loop timings from the standard timing table.
- In generated benchmark YAML, include `runtime.pre_commands` derived from the build section so authoritative runs rebuild the edited LAMMPS binary before benchmarking.
- In generated benchmark runtime command, invoke LAMMPS via MPI launcher with the case-specific rank count (16, 32, or 64), not one fixed rank count for every case.
- Set `OMP_NUM_THREADS=1` unless a case explicitly requires hybrid MPI+OpenMP, and keep this setting identical across baseline/candidate runs.
- In generated benchmark YAML, include a split block so worker sees the train cases only:
```yaml
split:
train_case_ids:
- train-16r-long
- train-32r-long
- train-32r-short
```
Summary¶
baseline (e7c0ed95a333):
0.36158best accepted (11db74893452):
0.28487(+21.22% vs baseline)published GitHub branch: fermilink-optimize/lammps-neighbor
iterations: 28 total | 6 accepted | 20 rejected | 0 correctness failure
Optimization Trajectory¶
All iterations¶
iter |
commit |
status |
metric |
summary |
|---|---|---|---|---|
0 |
baseline |
0.36158 |
baseline |
|
1 |
2d1eda35b944 |
rejected |
0.47712 |
Split the orthogonal half-bin/newton self-bin traversal in `npair_bin.cpp` and cache per-atom cut… |
2 |
accepted |
0.33535 |
Specialize the orthogonal half/bin/newton neighbor build in `src/npair_bin.cpp` to remove generic… |
|
3 |
accepted |
0.32842 |
Add a benchmark-specific fast path in `NPairBin<1,1,0,0,0>::build()` for standard molecular syste… |
|
4 |
3bd7cbc7ee03 |
rejected |
0.32287 |
Skip `find_special()` and `minimum_image_check()` for different-molecule pairs inside the incumbe… |
5 |
1a1935b59c00 |
rejected |
0.32384 |
Add a `maxspecial <= 2` molecular/no-exclusion fast path in `NPairBin<1,1,0,0,0>::build()` that b… |
6 |
accepted |
0.31969 |
Refine the incumbent `NPairBin<1,1,0,0,0>::build()` fast path by early-accepting different-molecu… |
|
7 |
d89e6f820474 |
rejected |
0.32426 |
Replace the remaining same-molecule `find_special()` scan in the incumbent `NPairBin<1,1,0,0,0>::… |
8 |
d381c38751ec |
rejected |
0.33329 |
Split the benchmark-hot molecular/no-exclusion neighbor build into encoded-special and generic va… |
9 |
2894990122f9 |
rejected |
0.3187 |
Cache per-neighbor ownership/tag/molecule values and use an encoded-only special lookup inside th… |
10 |
b21ee6319e89 |
rejected |
0.3221 |
Split the incumbent molecular/no-exclusion self-bin walk in `NPairBin<1,1,0,0,0>::build()` into o… |
11 |
accepted |
0.3115 |
Use contiguous `_noalias` aliases for hot coordinate/type/tag/molecule arrays in the incumbent `N… |
|
12 |
e5d6d4548fe4 |
rejected |
0.31003 |
Extend the incumbent `NPairBin<1,1,0,0,0>::build()` hot path with additional neighbor-array alias… |
13 |
7adf3bfc72a8 |
rejected |
0.31711 |
Replace the incumbent molecular/no-exclusion `NPairBin<1,1,0,0,0>::build()` special handling with… |
14 |
e536920bfbe5 |
rejected |
0.30666 |
Add explicit `jnext` traversal and light next-neighbor coordinate prefetching to the incumbent `N… |
15 |
e5eef089d2d0 |
rejected |
0.30618 |
Combine explicit jnext traversal and next-neighbor coordinate prefetching with cached periodic ha… |
16 |
d5b0238dc967 |
rejected |
0.30551 |
Refine the single-path molecular half/bin/newton builder in src/npair_bin.cpp with explicit jnext… |
17 |
4a9dda87cf45 |
rejected |
0.30695 |
Combine iteration-16 style `NPairBin<1,1,0,0,0>` jnext/prefetch/half-box fast-path refinements wi… |
18 |
accepted |
0.29094 |
Use explicit `jnext` linked-list traversal with bin/list `_noalias` aliases and delayed generic-o… |
|
19 |
accepted |
0.28487 |
Optimize Neighbor movement tracking by scanning contiguous x/xhold arrays in check_distance() and… |
|
20 |
11db74893452 |
worker_incomplete |
nan |
optimize iteration 20 |
21 |
e35c58b86e07 |
rejected |
0.28995 |
Widen Neighbor::check_distance() to 8-atom unrolled blocks and compare one blockwise max displace… |
22 |
708f9b2d16b1 |
rejected |
0.28503 |
Specialize the incumbent `NPairBin<1,1,0,0,0>::build()` fast path for all-zero special-bond weigh… |
23 |
a3f1052997d8 |
rejected |
0.28469 |
Add a runtime uniform-cutoff fast path to the specialized `NPairBin<1,1,0,0,0>` molecular/no-excl… |
24 |
84babbbb36e3 |
rejected |
0.49372 |
Add an axis-aligned early reject in the specialized `NPairBin<1,1,0,0,0>` fast path so pairs exce… |
25 |
b78acabc90a5 |
rejected |
0.28592 |
Isolate the iteration-23 Neighbor cadence fast-check by caching the benchmark-common every=1/dela… |
26 |
ba18979d7807 |
rejected |
0.28444 |
Add a benchmark-common `NPairBin<1,1,0,0,0>` fast path for the uniform-cutoff plus all-special-en… |
27 |
321fb756a836 |
rejected |
0.28559 |
Cache j-side neighbor coords/tag/molecule once and inline orthogonal half-box minimum-image check… |
Accepted Commits¶
Accepted candidate detail pages and current manual-review status:
accepted commit |
Human verification |
|---|---|
not verified |
|
not verified |
|
not verified |
|
not verified |
|
not verified |
|
not verified |
Benchmark Contracts¶
Necessary files to reproduce the FermiLink optimization results:
Input files for Benchmarks¶
Copied auxiliary benchmark inputs from .fermilink-optimize/inputs/all/:
Runtime Data¶
FermiLink runtime data for accepted/rejected commits.
Rerun Guide¶
Agent provider codex; model gpt-5.4-xhigh
Use the bundled contract files from this report to recreate the optimization against a fresh upstream checkout.
default upstream clone:
git@github.com:skilled-scipkg/lammps.gitconfirm the upstream default branch before creating the worktree: develop on GitHub
detected package language:
cpp; usefermilink-optimize-cppfor goal-mode rerunsif
goal_inputs.jsonis present, restage the listed auxiliary workload files before rerunningcopied benchmark input files are bundled under
inputs/all/and should be restored into.fermilink-optimize/inputs/all/for deterministic reruns
git clone git@github.com:skilled-scipkg/lammps.git
cd lammps
git worktree add -b fermilink-optimize/lammps-<modified-feature> ../lammps-<modified-feature> develop
Path 1: Rerun from goal.md¶
Rerun from the bundled goal.md.
Note
Tune the copied ## Build section in goal.md before rerunning. Update environment activation, module loads, compiler paths, install prefixes, and other machine-specific setup so FermiLink builds the package correctly.
mkdir -p build
cd build
cmake -C ../cmake/presets/most.cmake -C ../cmake/presets/nolib.cmake -D PKG_GPU=off ../cmake
cmake --build . -j4
Note
The copied ## Representative Workloads section references input files that are also bundled under Input files for Benchmarks. Copy these files into the same directory as the goal.md file used for this rerun before launching FermiLink, so goal mode can capture and stage them:
Run this from the cloned main repo so the launcher can create or reuse the sibling worktree:
fermilink-optimize-cpp \
--project-root "$PWD" \
--goal /path/to/report/contract/goal.md \
--branch fermilink-optimize/lammps-<modified-feature> \
--worktree-root .. \
--worktree-name lammps-<modified-feature>
Path 2: More deterministic rerun from benchmark.yaml¶
Rerun from the copied benchmark.yaml and benchmark_runner.py. These files are generated from goal.md by FermiLink, serving as a deterministic benchmark contract that the agent needs to follow during optimization iterations. FermiLink does not directly rely on goal.md for optimization iterations.
This avoids regenerating the benchmark contract from goal.md before the campaign starts:
Note
Inspect benchmark.yaml before rerunning. Update runtime.pre_commands for machine-specific build/setup steps, and verify that runtime.command paths point at files that exist in the new worktree.
cd ../lammps-<modified-feature>
mkdir -p .fermilink-optimize/autogen .fermilink-optimize/inputs/all
cp /path/to/report/contract/benchmark.yaml .fermilink-optimize/autogen/benchmark.yaml
cp /path/to/report/contract/benchmark_runner.py .fermilink-optimize/autogen/benchmark_runner.py
cp -R /path/to/report/inputs/all/. .fermilink-optimize/inputs/all/
printf '%s\n' '.fermilink-optimize/' >> .git/info/exclude
fermilink optimize lammps "$PWD" \
--benchmark "$PWD/.fermilink-optimize/autogen/benchmark.yaml" \
--skills-source existing
Benchmark Examples¶
Worker iterations run the train-* benchmark cases below while searching for candidate changes:
cases:
- id: train-16r-long
weight: 1.0
description: Longer-run 16-rank training case with repeated rebuilds and lower communication noise.
input_script: in.tip4p_nve_long
data_file: water_216_data.lmp
mpi_ranks: 16
omp_threads: 1
- id: train-32r-long
weight: 1.0
description: Longer-run 32-rank training case for a second domain decomposition.
input_script: in.tip4p_nve_long
data_file: water_216_data.lmp
mpi_ranks: 32
omp_threads: 1
- id: train-32r-short
weight: 1.0
description: Shorter-turnaround 32-rank training case with the same neighbor settings.
input_script: in.tip4p_nve
data_file: water_216_data.lmp
mpi_ranks: 32
omp_threads: 1
Controller reviews run the test-* benchmark cases below to validate accepted candidates:
cases:
- id: test-16r-short
weight: 1.0
description: Held-out lower-rank validation case.
input_script: in.tip4p_nve
data_file: water_216_data.lmp
mpi_ranks: 16
omp_threads: 1
- id: test-64r-short
weight: 1.0
description: Held-out scaling-sensitive validation case.
input_script: in.tip4p_nve
data_file: water_216_data.lmp
mpi_ranks: 64
omp_threads: 1