Elsevier / Physica A
Sole-author, peer-reviewed paper
A Monte Carlo study on reducing bias and variance in Detrended Fluctuation Analysis (DFA).
GRUENCY / Scientific computing
Workstations for scientific computing
Simulation and numerical research involve long unattended runs, heavy memory use and results that must be correct. GRUENCY builds workstations for this work. Its founder, Sebastian Michalski, runs Monte Carlo studies of his own.
Matched to the solver
- CPU and GPU balance chosen for the codes you run
- ECC memory and a platform suited to results that must be correct
- Memory bandwidth and core count weighed for each solver
- Storage layout for datasets, checkpoints and results
- Stability for long unattended runs
- Noise targets suitable for an office or laboratory
Testing for long runs
Each machine goes through a sustained full-load burn-in, thermal profiling and a noise check. Remote commissioning into your environment is included, worldwide.
Research background
NBP, National Bank of Poland
First-degree award
Cash-circulation optimisation and economic modelling for Poland's central bank.
Prototype, 2017
11 water-cooled GPUs in one chassis
A dual-CPU server platform in a single CAD-designed case, built in 2017.
Questions about scientific computing
CPU-heavy or GPU-heavy for simulation?
It depends on the solver. Memory bandwidth, core count and GPU acceleration are assessed against your codes in the first conversation.
Is ECC memory worth it?
For long runs and research that must be correct, usually yes. It is a platform decision, and the specification explains it.
Do you understand research workflows?
Sebastian runs Monte Carlo studies himself and has published a sole-author methodology paper in Physica A. He uses machines of this kind in his own work.
Request a consultation
Describe the codes, the datasets and where the machine will be used.
Or email info@gruency.com