CloudBench
← All Workloads

HPC

High-performance computing benchmarks including LINPACK and CFD simulations.

LINPACK GFLOPS

Higher is better -- sorted by performance

Our Recommendations

H3h3-standard-88

Tightly-coupled HPC and MPI workloads

Purpose-built for HPC with 200 Gbps networking, 2,150 GFLOPS LINPACK performance, and optimized for MPI workloads. The clear winner for tightly-coupled HPC applications.

C3Dc3d-standard-60

Embarrassingly parallel and memory-bound HPC

920 GFLOPS LINPACK with AMD EPYC processors. Up to 360 vCPUs per VM for embarrassingly parallel workloads. Strong memory bandwidth for memory-bound HPC codes.

C3c3-standard-44

Intel-optimized HPC codes with AVX-512

680 GFLOPS with the best per-core performance. Intel Xeon with AVX-512 for vectorized workloads. Good choice for HPC codes optimized for Intel architectures.

Budget PickN2(n2-standard-32)

380 GFLOPS at a significantly lower price than the compute-optimized families. Adequate for development, testing, and smaller-scale HPC workloads.

Price-Performance: LINPACK GFLOPS

VM SeriesMachine TypePerformanceCost/hrPerf/$
H3h3-standard-882,150 GFLOPS$3.726577
C3Dc3d-standard-60920 GFLOPS$2.880319.4
C3c3-standard-44680 GFLOPS$2.300295.7
N2n2-standard-32380 GFLOPS$1.553244.7

All Benchmark Data

VM SeriesMachine TypeMetricResultNotes
C3c3-standard-44LINPACK GFLOPS680 GFLOPSHPL benchmark, 44 vCPUs, Intel MKL optimized
C3Dc3d-standard-60LINPACK GFLOPS920 GFLOPSHPL benchmark, 60 vCPUs, AMD BLIS optimized
H3h3-standard-88LINPACK GFLOPS2,150 GFLOPSHPL benchmark, 88 vCPUs, Intel MKL, 200 Gbps network
N2n2-standard-32LINPACK GFLOPS380 GFLOPSHPL benchmark, 32 vCPUs, Intel MKL
H3h3-standard-88OpenFOAM Iterations/sec14.2 iter/smotorbike tutorial, 88 cores, OpenMPI
C3c3-standard-44OpenFOAM Iterations/sec6.8 iter/smotorbike tutorial, 44 cores, OpenMPI
C3Dc3d-standard-60OpenFOAM Iterations/sec8.4 iter/smotorbike tutorial, 60 cores, OpenMPI