FastBEM AcousticsV9

Verified V8.0 to V9.0 comparison

Highly optimized solvers with significantly improved computational efficiency on multicore CPUs

V9 improves FMM, ACA, and direct BEM solver execution by several folds compared with V8 on multicore CPUs.

Download V9Verification casesRelease notes
>4XFMM solver faster than V8.
4–8XACA solver faster than V8.
≈1.1XDirect solver faster than V8.

Matched sphere benchmarks

Large-model performance summary

Wall-clock times compare V8.0 with V9.0 on the same ThinkPad P1 laptop with an Intel Core i7-13800H CPU and 96 GB memory.

SolverFrequencyMatched rangeMedian speedupMedian reduction
Fast multipole BEMka=297,200–5,880,000 DOF4.20×76.2%
Fast multipole BEMka=2097,200–5,880,000 DOF4.24×76.4%
ACA BEMka=297,200–1,009,200 DOF4.64×78.5%
ACA BEMka=2097,200–1,009,200 DOF7.78×87.2%
Conventional BEMka=2 / 2097,200–145,200 DOF1.15× / 1.07×12.8% / 6.7%
Log-scale V9 wall-clock time versus degrees of freedom for conventional, ACA, and fast multipole BEM at ka 2 and 20