cbcbox 2.923

Binary distribution of the CBC MILP solver (COIN-OR Branch and Cut)

cbcbox

cbcbox is a high-performance, self-contained Python distribution of the CBC MILP solver (COIN-OR Branch and Cut), built from the latest COIN-OR master branch.

On x86_64 (Linux, macOS, Windows) the wheel ships both a Haswell-optimised binary (AVX2/FMA) for maximum speed and a generic build with runtime CPU dispatch for compatibility with any x86_64 machine — selected automatically. All dynamic dependencies (OpenBLAS, libgfortran, etc.) are bundled; no system libraries or separate installation steps are needed.

Highlights

• Haswell-optimised & generic builds — on x86_64 Linux, macOS, and Windows the wheel ships two complete solver stacks: a Haswell build (OpenBLAS AVX2/FMA kernel) for maximum throughput, and a generic build (DYNAMIC_ARCH runtime dispatch) for compatibility with any x86_64 CPU. The best available variant is selected automatically at import time (see Build variants).

• Parallel branch-and-cut — built with --enable-cbc-parallel. Use -threads=N to distribute the search tree across N threads, giving significant speedups on multi-core machines for hard MIP instances.

• AMD fill-reducing ordering — SuiteSparse AMD is compiled in, enabling the high-quality UniversityOfFlorida Cholesky factorization for Clp's barrier (interior point) solver. AMD reordering produces much less fill-in on large sparse problems than the built-in native Cholesky, making barrier substantially faster. Activate with -cholesky UniversityOfFlorida -barrier (see barrier usage).

Performance (x86_64)

Auto-updated by CI after each successful workflow run. Single-threaded solve time — lower is better.

The AVX2/Haswell build is ~3.2× faster than the generic build on average (geometric mean across 20 instances, 2 x86_64 platforms: Darwin x86_64, Windows AMD64).

Single-threaded solve time across benchmark instances on Linux x86_64. Speedup factor shown above each pair. Lower is better.

See also: Windows AMD64 + macOS x86_64 summary

Build variants

On x86_64 Linux, macOS, and Windows, the wheel ships two complete sets of binaries:

Variant	OpenBLAS kernel	Clp SIMD	Minimum CPU
generic	DYNAMIC_ARCH=1 (runtime dispatch, Nehalem–Zen targets)	standard	any x86_64
avx2	DYNAMIC_ARCH=1 + DYNAMIC_LIST=HASWELL SKYLAKEX	-march=haswell -DCOIN_AVX2=4	Haswell (2013+)

At import time cbcbox automatically selects avx2 when it is available and the running CPU supports AVX2; otherwise it falls back to generic.

You can override this selection with the CBCBOX_BUILD environment variable:

# Force generic (portable) build
CBCBOX_BUILD=generic cbc mymodel.mps -solve -quit

# Force AVX2-optimised build (raises an error if not available)
CBCBOX_BUILD=avx2 cbc mymodel.mps -solve -quit

When CBCBOX_BUILD is set, a short summary of the selected build is printed to stdout on every call — useful for tagging experiment results:

[cbcbox] CBCBOX_BUILD=avx2
[cbcbox]   binary  : .../cbcbox/cbc_dist_avx2/bin/cbc
[cbcbox]   lib dir : .../cbcbox/cbc_dist_avx2/lib
[cbcbox]   libs    : libCbc.so.3, libClp.so.3, libopenblas.so.0

Non-x86_64 platforms (Linux aarch64, macOS arm64) ship the generic build only. CBCBOX_BUILD=avx2 will raise a RuntimeError on those platforms.

Local debug builds

The released wheels are fully optimised and stripped. To debug CBC itself (e.g. with GDB or LLDB), use the scripts in scripts/ to build a local debug-enabled binary. These produce the same full feature set as the release wheels (OpenBLAS, AMD, Nauty, pthreads) but compiled with -O1 -g and, on x86_64, with -march=haswell -DCOIN_AVX2=4 so you can debug AVX2-specific code paths.

Script	Platform	Environment	Output directory
scripts/build_debug.sh	Linux, macOS	native (host compiler)	cbc_dist_debug_avx2/ (x86_64) or cbc_dist_debug/ (ARM64)
scripts/build_debug_manylinux.sh	Linux	Docker — manylinux2014 container (exact CI parity)	same as above
scripts/build_debug_windows.ps1	Windows	MSYS2 / MinGW64	cbc_dist_debug_avx2\

Quick start

Linux / macOS (native build):

# x86_64 → debug + AVX2 → cbc_dist_debug_avx2/bin/cbc
# ARM64  → debug only  → cbc_dist_debug/bin/cbc
./scripts/build_debug.sh

# With AddressSanitizer:
./scripts/build_debug.sh --asan

# With ThreadSanitizer:
./scripts/build_debug.sh --tsan

# Force a clean rebuild from scratch (required when switching sanitizers):
./scripts/build_debug.sh --asan --clean

Linux (manylinux2014 container — matches CI exactly):

# Requires Docker; the script prints install instructions if it is missing.
./scripts/build_debug_manylinux.sh
./scripts/build_debug_manylinux.sh --asan
./scripts/build_debug_manylinux.sh --tsan

Windows (PowerShell):

# Requires MSYS2 at C:\msys64.  Note: sanitizers are not supported on Windows/MinGW.
.\scripts\build_debug_windows.ps1
.\scripts\build_debug_windows.ps1 -Clean   # force full rebuild

Debugging

# GDB (Linux):
gdb cbc_dist_debug_avx2/bin/cbc
(gdb) run mymodel.mps -solve -quit

# LLDB (macOS):
lldb cbc_dist_debug/bin/cbc
(lldb) run mymodel.mps -solve -quit

Sanitizer tips

Sanitizer	Flag	What it catches	Runtime env var
AddressSanitizer	--asan	heap/stack buffer overflows, use-after-free, memory leaks	ASAN_OPTIONS=detect_leaks=0 to suppress system-lib false positives
ThreadSanitizer	--tsan	data races between threads	TSAN_OPTIONS=halt_on_error=0 to log races without aborting

ASan and TSan are mutually exclusive. Neither is available on Windows/MinGW. Always pass --clean when switching from one sanitizer to another to avoid linking mismatched object files.

OpenBLAS is always built without sanitizer flags to avoid false positives from hand-optimised BLAS assembly; only the COIN-OR stack is instrumented.

Note: Debug binaries are not included in the published wheels because of their size. They are intended for local development only.

Supported platforms

Platform	Wheel tag
Linux x86_64	manylinux2014_x86_64
Linux aarch64	manylinux2014_aarch64
macOS arm64 (Apple Silicon)	macosx_11_0_arm64
macOS x86_64	macosx_10_9_x86_64
Windows AMD64	win_amd64

Installation

pip install cbcbox

Usage

Command line

After installation, CBC is available directly as the cbc command (pip installs the entry point into the environment's bin/ on Linux/macOS or Scripts/ on Windows, which is already on PATH):

cbc mymodel.lp -solve -quit
cbc mymodel.mps.gz -solve -quit
cbc mymodel.mps -seconds 60 -timem elapsed -solve -quit
cbc mymodel.mps -dualp pesteep -solve -quit

Alternatively, invoke via the Python module entry point:

python -m cbcbox mymodel.lp -solve -quit

CBC accepts LP, MPS and compressed MPS (.mps.gz) files. Pass -help for the full list of options, or -quit to exit after solving.

Parallel branch-and-cut

This build includes parallel branch-and-cut (--enable-cbc-parallel). Use -threads=N to distribute the search tree across N threads:

cbc mymodel.mps -threads=4 -solve -quit

Barrier (interior-point) solver

Clp's barrier solver can be faster than simplex for large LP relaxations. This build includes SuiteSparse AMD, which enables the high-quality UniversityOfFlorida Cholesky factorization — significantly reducing fill-in compared to the built-in native Cholesky:

# Solve LP relaxation with barrier + AMD Cholesky, then crossover to simplex basis
cbc mymodel.mps -cholesky UniversityOfFlorida -barrier -solve -quit

# Useful as a root-node strategy inside MIP (let CBC use simplex for B&B):
cbc mymodel.mps -cholesky UniversityOfFlorida -barrier -solve -quit

Without AMD, only -cholesky native (less efficient) is available.

Python API

The package exposes helpers to locate the installed files:

import cbcbox
import subprocess

# Path to the cbc binary (cbc.exe on Windows).
cbcbox.cbc_bin_path()
# e.g. '/home/user/.venv/lib/python3.13/site-packages/cbcbox/cbc_dist/bin/cbc'

# Directory containing the shared libraries.
cbcbox.cbc_lib_dir()
# e.g. '.../cbcbox/cbc_dist/lib'

# Directory containing the COIN-OR C/C++ headers.
cbcbox.cbc_include_dir()
# e.g. '.../cbcbox/cbc_dist/include/coin'

# Run CBC programmatically.
result = subprocess.run(
    [cbcbox.cbc_bin_path(), "mymodel.mps", "-solve", "-quit"],
    capture_output=True, text=True,
)
print(result.stdout)

What is built

The build pipeline compiles all components from source inside the CI runner, in the following order:

Component	Version / branch	Purpose
Cbc	master	Branch-and-cut MIP solver
Cgl	master	Cut generation library
Clp	master	Simplex LP solver (used as the MIP node relaxation)
Osi	master	Open Solver Interface
CoinUtils	master	Utility library (shared by all COIN-OR packages)
Nauty	2.8.9	Symmetry detection for MIP presolve
AMD (SuiteSparse v7.12.2)	v7.12.2	Sparse matrix fill-reducing ordering
OpenBLAS	v0.3.31	Optimised BLAS/LAPACK for LP basis factorisation

On x86_64 Linux, macOS, and Windows the entire stack is compiled twice: once for the generic variant (OpenBLAS DYNAMIC_ARCH=1 with a broad set of x86_64 targets for runtime dispatch) and once for the avx2 variant (OpenBLAS DYNAMIC_ARCH=1 restricted to Haswell/Skylake targets via DYNAMIC_LIST, COIN-OR compiled with -march=haswell -DCOIN_AVX2=4). Both variants use NO_CBLAS=1 (COIN-OR only calls the Fortran BLAS interface). AMD and Nauty are built only once (they are pure combinatorial code with no BLAS dependency) and reused by both COIN-OR variants.

All COIN-OR components are built as shared (.so / .dylib / .dll) libraries. The shared libraries are patched with self-relative RPATHs and bundled inside the wheel, making them directly usable via cffi or ctypes without any system installation.

Wheel contents

The wheel installs under cbcbox/ inside the site-packages directory. On x86_64 Linux, macOS, and Windows it contains two dist trees; other platforms contain only cbc_dist/:

cbc_dist/           ← generic build (all platforms)
cbc_dist_avx2/      ← AVX2-optimised build (x86_64 Linux/macOS/Windows)
├── bin/
│   ├── cbc           # CBC MIP solver binary  (cbc.exe on Windows)
│   └── clp           # Clp LP solver binary   (clp.exe on Windows)
├── lib/
│   ├── libCbc.so / libCbc.dylib / libCbc.dll  # CBC solver
│   ├── libCbcSolver.so ...
│   ├── libClp.so ...                          # Clp LP solver
│   ├── libCgl.so ...                          # Cut generation
│   ├── libOsi.so ...                          # Solver interface
│   ├── libOsiClp.so ...                       # Clp OSI binding
│   ├── libOsiCbc.so ...                       # CBC OSI binding (where available)
│   ├── libCoinUtils.so ...
│   ├── libopenblas.so / .dylib / .dll         # OpenBLAS BLAS/LAPACK
│   ├── pkgconfig/                             # .pc files for all libraries
│   └── <bundled runtime shared libs>          # Platform-specific — see below
└── include/
    ├── coin/      # COIN-OR headers (CoinUtils, Osi, Clp, Cgl, Cbc)
    ├── nauty/     # Nauty headers
    └── *.h        # SuiteSparse / AMD headers

Bundled dynamic libraries

Because OpenBLAS links to the Fortran runtime, the following shared libraries are bundled inside the wheel and their paths are rewritten so no system installation is required.

Linux (lib/ directory, RPATH set to $ORIGIN)

Library	Description
libopenblas.so.0	OpenBLAS BLAS/LAPACK
libgfortran.so.5	GNU Fortran runtime
libquadmath.so.0	Quad-precision math (dependency of libgfortran)

macOS (lib/ directory, install names rewritten to @rpath/)

Library	Description
libopenblas.dylib	OpenBLAS BLAS/LAPACK
libgfortran.5.dylib	GNU Fortran runtime
libgcc_s.1.1.dylib	GCC runtime
libquadmath.0.dylib	Quad-precision math

Windows (bin/ directory, DLLs placed next to the executable)

Library	Description
libopenblas.dll	OpenBLAS BLAS/LAPACK
libgfortran-5.dll	GNU Fortran runtime
libgcc_s_seh-1.dll	GCC SEH runtime
libquadmath-0.dll	Quad-precision math
libstdc++-6.dll	C++ standard library (MinGW64)
libwinpthread-1.dll	POSIX thread emulation

CI / build pipeline

Wheels are built and tested automatically via GitHub Actions using cibuildwheel. The workflow (.github/workflows/wheel.yml) runs independent compile jobs in parallel, then packages each platform:

Compile jobs	Runner	Produces
compile-linux-x64-generic + compile-linux-x64-avx2	ubuntu-latest	manylinux2014_x86_64 wheel
compile-linux-arm64-generic	ubuntu-24.04-arm	manylinux2014_aarch64 wheel
compile-macos-arm64-generic	macos-15	macosx_11_0_arm64 wheel
compile-macos-intel-generic + compile-macos-intel-avx2	macos-15-intel	macosx_10_9_x86_64 wheel
compile-windows-generic + compile-windows-avx2	windows-latest	win_amd64 wheel

Each platform's compile jobs run in parallel. Once all compile jobs for a platform finish, the corresponding package-* job assembles the wheel via cibuildwheel and runs the test suite against the installed wheel.

A final combine_reports job collects per-platform performance results and commits the updated README.md to the repository.

Integration tests

The test suite (pytest) solves 21 MIP instances and checks the optimal objective values, in both single-threaded and parallel (3-thread) modes. On x86_64 Linux, macOS, and Windows each test is run twice — once against the generic binary and once against the avx2 binary — and a side-by-side performance comparison is recorded:

Instance	Expected optimal	Time limit
pp08a	7 350	2000 s
sprint_hidden06_j	130	2000 s
air03	340 160	2000 s
air04	56 137	2000 s
air05	26 374	2000 s
nw04	16 862	2000 s
mzzv11	−21 718	2000 s
trd445c	−153 419.078836	2000 s
nursesched-sprint02	58	2000 s
stein45	30	2000 s
neos-810286	2 877	2000 s
neos-1281048	601	2000 s
j3050_8	1	2000 s
qiu	−132.873136947	2000 s
gesa2-o	25 779 856.3717	2000 s
pk1	11	2000 s
mas76	40 005.054142	2000 s
app1-1	−3	2000 s
eil33-2	934.007916	2000 s
fiber	405 935.18	2000 s
neos-2987310-joes	−607 702 988.291	2000 s

Time limits are generous to avoid false failures on slow CI runners.

Performance results

Auto-updated by CI after each successful workflow run.

Summary

Geometric mean solve time (seconds) across all test instances.

1 thread

Platform	generic (s)	avx2 (s)	avx2 speedup
Darwin x86_64	55.74	17.44	3.20×
Darwin arm64	32.54	—	—
Windows AMD64	48.17	15.14	3.18×

3 threads

Platform	generic (s)	avx2 (s)	avx2 speedup
Darwin x86_64	50.58	13.34	3.79×
Darwin arm64	35.03	—	—
Windows AMD64	38.47	14.79	2.60×

Per-instance results

pp08a

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	4.92	8.97	0.55×
Darwin x86_64	generic	9.01	8.58	1.05×
Darwin arm64	generic	9.05	14.72	0.62×
Windows AMD64	avx2	4.91	8.14	0.60×
Windows AMD64	generic	12.52	22.82	0.55×

sprint_hidden06_j

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	52.21	44.13	1.18×
Darwin x86_64	generic	167.56	218.53	0.77×
Darwin arm64	generic	116.90	120.40	0.97×
Windows AMD64	avx2	57.59	54.94	1.05×
Windows AMD64	generic	246.50	210.51	1.17×

air03

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	1.68	1.68	1.00×
Darwin x86_64	generic	6.03	7.80	0.77×
Darwin arm64	generic	3.86	3.86	1.00×
Windows AMD64	avx2	2.32	2.39	0.97×
Windows AMD64	generic	5.90	6.08	0.97×

air04

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	52.16	32.47	1.61×
Darwin x86_64	generic	113.43	99.69	1.14×
Darwin arm64	generic	101.85	76.95	1.32×
Windows AMD64	avx2	33.51	27.09	1.24×
Windows AMD64	generic	154.46	88.84	1.74×

air05

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	24.58	16.39	1.50×
Darwin x86_64	generic	56.03	53.86	1.04×
Darwin arm64	generic	46.16	47.12	0.98×
Windows AMD64	avx2	17.96	23.19	0.77×
Windows AMD64	generic	57.44	48.23	1.19×

nw04

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	13.02	12.41	1.05×
Darwin x86_64	generic	34.71	44.80	0.77×
Darwin arm64	generic	32.80	35.57	0.92×
Windows AMD64	avx2	15.87	15.83	1.00×
Windows AMD64	generic	57.51	54.34	1.06×

mzzv11

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	124.54	78.60	1.58×
Darwin x86_64	generic	501.14	475.71	1.05×
Darwin arm64	generic	241.20	166.05	1.45×
Windows AMD64	avx2	118.42	146.32	0.81×
Windows AMD64	generic	254.40	265.07	0.96×

trd445c

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	107.75	93.87	1.15×
Darwin x86_64	generic	244.45	258.53	0.95×
Darwin arm64	generic	173.44	194.91	0.89×
Windows AMD64	avx2	100.53	109.16	0.92×
Windows AMD64	generic	229.88	228.59	1.01×

nursesched-sprint02

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	39.21	29.25	1.34×
Darwin x86_64	generic	93.62	116.17	0.81×
Darwin arm64	generic	82.99	91.06	0.91×
Windows AMD64	avx2	26.54	26.53	1.00×
Windows AMD64	generic	113.09	84.91	1.33×

stein45

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	10.41	9.72	1.07×
Darwin x86_64	generic	25.82	22.01	1.17×
Darwin arm64	generic	18.54	13.80	1.34×
Windows AMD64	avx2	8.42	7.49	1.12×
Windows AMD64	generic	26.32	17.88	1.47×

neos-810286

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	15.12	11.36	1.33×
Darwin x86_64	generic	43.72	51.83	0.84×
Darwin arm64	generic	24.62	26.27	0.94×
Windows AMD64	avx2	13.01	13.83	0.94×
Windows AMD64	generic	35.83	39.27	0.91×

neos-1281048

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	23.55	7.22	3.26×
Darwin x86_64	generic	118.08	20.31	5.81×
Darwin arm64	generic	38.11	34.21	1.11×
Windows AMD64	avx2	13.61	19.31	0.71×
Windows AMD64	generic	36.23	21.55	1.68×

j3050_8

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	6.00	3.30	1.82×
Darwin x86_64	generic	7.64	9.95	0.77×
Darwin arm64	generic	7.41	6.79	1.09×
Windows AMD64	avx2	2.17	2.33	0.93×
Windows AMD64	generic	8.42	7.00	1.20×

qiu

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	47.14	12.33	3.82×
Darwin x86_64	generic	66.16	33.00	2.00×
Darwin arm64	generic	84.15	27.90	3.02×
Windows AMD64	avx2	23.97	11.98	2.00×
Windows AMD64	generic	80.02	38.81	2.06×

gesa2-o

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	5.17	4.82	1.07×
Darwin x86_64	generic	13.17	12.38	1.06×
Darwin arm64	generic	9.15	9.72	0.94×
Windows AMD64	avx2	3.38	3.24	1.04×
Windows AMD64	generic	10.98	10.72	1.02×

pk1

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	37.42	31.92	1.17×
Darwin x86_64	generic	92.01	83.60	1.10×
Darwin arm64	generic	63.56	52.95	1.20×
Windows AMD64	avx2	33.24	36.52	0.91×
Windows AMD64	generic	102.32	68.82	1.49×

mas76

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	20.28	46.06	0.44×
Darwin x86_64	generic	55.18	96.05	0.57×
Darwin arm64	generic	40.46	53.86	0.75×
Windows AMD64	avx2	19.48	32.70	0.60×
Windows AMD64	generic	53.11	68.40	0.78×

app1-1

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	6.37	7.42	0.86×
Darwin x86_64	generic	728.44	578.89	1.26×
Darwin arm64	generic	13.51	231.96	0.06×
Windows AMD64	avx2	20.44	7.18	2.85×
Windows AMD64	generic	81.69	23.39	3.49×

eil33-2

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	45.60	17.84	2.56×
Darwin x86_64	generic	163.15	72.45	2.25×
Darwin arm64	generic	113.45	75.05	1.51×
Windows AMD64	avx2	30.70	17.46	1.76×
Windows AMD64	generic	168.16	70.78	2.38×

fiber

Platform	Build	1 thread (s)	3 threads (s)	parallel speedup
Darwin x86_64	avx2	1.50	1.10	1.36×
Darwin x86_64	generic	6.79	7.53	0.90×
Darwin arm64	generic	2.07	2.63	0.79×
Windows AMD64	avx2	1.91	2.06	0.93×
Windows AMD64	generic	3.90	4.20	0.93×

NAQ — Never Asked Questions

Why not benchmark on the full MIPLIB 2017 library?

Several practical constraints shape the benchmark set:

1. CI time limits. GitHub Actions enforces a 6-hour wall-clock limit per job. The full MIPLIB 2017 collection contains ~240 instances, many of which take hours even on fast hardware. Including all of them would make every CI run time out before producing any useful measurements.

2. Comparing apples to apples requires instances solved to optimality. If some instances are only solved within a time limit (i.e., a gap > 0 %), a meaningful performance comparison must account for both solve time and solution quality simultaneously. This greatly complicates analysis and makes plots harder to interpret. Restricting to instances that CBC reliably solves to proven optimality keeps the comparison clean: a single elapsed-time number per instance is all that is needed.

3. The instance set is intentionally biased toward set packing / covering / partitioning structure. Most instances in the benchmark (pp08a, sprint_hidden06_j, nw04, mzzv11, nursesched-sprint02, air0x, trd445c) contain large blocks of set packing, covering, or partitioning constraints. This structure arises naturally in applications such as crew scheduling, nurse scheduling, vehicle routing, and cutting stock — exactly the domain where column generation is most valuable. Since the benchmark focuses on this problem class rather than providing a general-purpose solver survey, it is a specially interesting use case.

License

CBC and all COIN-OR components are distributed under the Eclipse Public License 2.0. OpenBLAS is distributed under the BSD 3-Clause licence. SuiteSparse AMD is distributed under the BSD 3-Clause licence. Nauty is distributed under the Apache 2.0 licence.