LinearElasticityKernel

Published configurations

Cell-local named elasticity kernels with explicit quadrature accounting.

Measured operation

Measures one local stiffness matrix without scatter or global setup.

Measurement contract

Warm retains the kernel, output capacity, reference derivatives and mapped geometry. FirstKernel constructs, evaluates and destroys a kernel and its output on every iteration, with mapped geometry already warm. Neither scope represents process-first quadrature lookup or cold geometry. Exactly one selected cell is evaluated per iteration; mesh and space construction, oracle evaluation and guards remain outside timing.

Native and prescribed quadratic geometry use the same unit-box mesh and full-dimensional vector space. Constant real Lamé parameters are fixed at \(\lambda=2\) and \(\mu=3\). The public integral supplies an independent implementation of the same requested quadrature rule, not a claim of exact integration on curved or rational cells. The dimensionless comparison is \(\|A_K-A_K^{\mathrm{integral}}\|_F/ \max(1,\|A_K^{\mathrm{integral}}\|_F)\le10^{-11}\). A perturbed matrix must be rejected before timing starts.

Counters distinguish factor degree, reported element order, requested quadrature order and actual point count. The pair-component work proxy is \(n_q n_{\mathrm{local}}^2 s^2\); it is not an instruction count and excludes reference preparation and geometry costs. No source optimization, quadrature reduction, new cache, collective or CTest timing gate is added.

Recorded results

Each case retains its original inputs, iteration policy and correctness checks. Raw reports preserve counters, repetitions and the Google Benchmark context. CPU, wall and manually supplied times are labeled separately; unlike operations and configurations are not averaged together.

Timing history

RodinLinearElasticityKernelBenchmarks · auto · regular · ranks: 1

Benchmark machine and measurement conditions

Published CI benchmarks run on standard GitHub-hosted ubuntu-24.04 virtual machines for public repositories.

System	CPU	Memory	SSD
Ubuntu 24.04, x64	4 vCPUs	16 GB	14 GB

The benchmark builds use GCC 14 and Release mode. Each job runs one benchmark executable at a time. OpenMP, OpenBLAS and MKL use one thread; MPI ranks share the same runner. The selected configuration identifies the build and rank count.

The CPU model and host load can vary between virtual machines, so small timing differences should be checked across repeated runs. GitHub runner specifications describe the available hardware. The source workflow is BenchmarkGroup.yml.