VIP20

本课程属于：021-LS-DYNA and PAM-CRASH HPC Operations English Technical Courses · 共 20 门相关课程，组合学习更完整。

查看所属VIP全部课程 →

← 返回所属目录咨询 / 购买

HPC Cost and Energy Optimization for Crash Simulation

30 Chapters

View Course Samples Online

Crash simulation workloads on HPC

explicit finite element solvers, model sizes, time-step constraints, and cost drivers.

simulationworkloads

Energy and cost metrics for HPC crash simulation

core-hours, wall-clock time, kWh, PUE, carbon intensity, and total cost of ownership.

Profiling crash simulation jobs

solver telemetry, MPI communication, I/O, memory bandwidth, and hardware counters.

Profilingsimulation

Benchmarking methodology for crash simulation

representative models, scaling studies, repeatability, and fair comparison.

Benchmarkingmethodology

Hardware selection for crash simulation

CPU microarchitectures, memory channels, interconnect, accelerators, and price-performance trade-offs.

Hardwareselection

Parallel scaling of explicit crash solvers

strong scaling, weak scaling, domain decomposition, and communication overhead.

Parallelexplicit

Load balancing strategies for crash simulation

static partitioning, dynamic repartitioning, and contact-dominated imbalance.

balancingstrategies

Solver time-step and mass scaling trade-offs

accuracy, stability, energy use, and simulation cost.

time-steptrade-offs

Contact algorithm cost optimization

penalty vs. kinematic contact, search algorithms, and frequency of contact updates.

algorithmoptimization

Element formulation and integration choices

reduced integration, hourglass control, shell formulations, and their cost impact.

formulationintegration

Material model complexity and cost

constitutive models, state variables, tabulated data, and user subroutines.

Materialcomplexity

I/O optimization for crash simulation

checkpointing, restart files, output frequency, and parallel file systems.

I/Ooptimization

Memory footprint reduction

model condensation, single precision, out-of-core solvers, and memory-aware partitioning.

footprintreduction

Job scheduling and resource allocation

backfill, reservations, topology-aware placement, and power-aware scheduling.

schedulingresource

Power management on HPC clusters

CPU frequency scaling, power capping, GPU power limits, and energy-aware runtime.

Accelerator offload for crash simulation

GPU porting, kernel optimization, data movement, and hybrid CPU-GPU execution.

Acceleratorsimulation

Coupling crash simulation with optimization loops

DOE, surrogate models, and multi-objective cost-energy trade-offs.

Couplingsimulation

Cloud HPC for crash simulation

instance selection, spot pricing, auto-scaling, and cost governance.

Energy-aware model simplification

submodeling, hybrid meshes, and selective refinement for cost reduction.

Energy-awaresimplification

Thermal and cooling considerations for HPC crash simulation

liquid cooling, inlet temperature, and waste heat reuse.

considerationsHPC

Carbon-aware scheduling for crash simulation

temporal and spatial carbon intensity, workload shifting, and reporting.

Carbon-awarescheduling

full-vehicle frontal crash simulation cost and energy optimization on a CPU cluster.

full-vehiclesimulation

side impact simulation with GPU acceleration and power capping.

pedestrian protection simulation using cloud HPC and spot instances.

pedestrianprotection

battery pack crash simulation with coupled electro-thermal-mechanical models.

simulationelectro-thermal-mechanical

occupant safety simulation with high-fidelity dummy models and load balancing.

occupantsimulation

material characterization and validation impact on simulation cost and energy.

materialcharacterization

multi-Objective optimization of crashworthiness and energy consumption using surrogates.

multi-Objectiveoptimization

large-scale crash simulation on heterogeneous HPC with topology-aware scheduling.

large-scalesimulation

Future directions

exascale crash simulation, AI-driven solvers, energy-aware algorithms, and sustainable HPC practices.

directionsexascale