What’s an RPU?
Quside’s new Randomness Processing Unit, based on high-speed, high-quality QRNGs and hardware acceleration, allows customers to offload their randomness generation and processing tasks from the CPU, thereby accelerating and optimizing their randomized workloads. This fact simultaneously improves their effective computational capacity and the quality of their simulation, optimization, and prediction needs.
Uncover the benefits for your randomized workloads
PHARMA/SCIENCE
Stochastic Simulations
Computational Chemistry
Drug Discovery
FINANCE/INSURANCE
Speed up and optimize Monte Carlo simulations
Portfolio Optimization
Risk Assessment
LOGISTICS/OPERATIONS RESEARCH
Scheduling and dispatching simulations
Inventory Optimization
Stochastic strategic and tactical Planning (forecasting)
What makes RPUs better for randomized workloads?
Pseudorandom number generation subroutines may involve even up as 95% of the total simulation resources for stochastic workloads.
RPU One
Quside’s first-generation RPUs combine quantum-based randomness generation and reprogrammable hardware acceleration in a convenient PCIe form factor. Provable performance improvements of up to 10X and energy efficiencies above 20X have been demonstrated in customer workloads. Easily deployed in C++ and available also through Virtual Machine images for on-premise deployments.
- 10 Gb/s uniform, normal and logarithmic distributions (IEEE-754)
- Repeatability and reproducibility of simulations
- Avoid artifacts and co-dependency risk
- Minimal code migration cost
- Available for on-prem and via Quside Cloud Partners
- Native cloud version deployable from cloud marketplace
Request an assessment of how to boost your workloads using an RPU One
*Subject to availability
RPU stands for “Randomness Processing Unit”, Quside’s new acceleration device based on high-speed, high-quality, quantum-based random number generation and fine-tuned hardware acceleration. Quside’s RPUs allow customers to offload their randomness generation and processing tasks from the CPU, thereby accelerating and optimizing their randomized workloads. This fact simultaneously improves their effective computational capacity and the quality of their simulation, optimization, and prediction needs.
Customers that rely on Quside’s RPU have shown faster simulation speeds and better convergence of their stochastic simulations. Besides, the introduction of the device into their pipelines has enabled the deployment of new, advanced algorithms for their workloads, for which they didn’t have enough computing capabilities before.
Random number generation subroutines may involve up to 95% of the total simulation resources for stochastic workloads. Quside’s RPU technology accelerates randomized workloads, thus improving your effective computational capacity and the quality of your simulation, optimization, and prediction needs. Thanks to these improvements you can obtain competitive advantages with the possibility of making better decision in a shorter time and reduce your infrastructure costs in terms of CAPEX and OPEX.
Quside’s RPU utilizes ultrafast quantum RNGs and offloads RN processing, improving efficiency and reducing hidden patterns in PRNGs.
Quside, a Barcelona-based company that believes technological advances create new capabilities that can transform the way billions of people communicate and compute. At Quside, our mission is to deliver the highest performance quantum random number generators (QRNGs) to empower the transition to safer connectivity and more efficient computation for everyone, everywhere.
Any randomness-intensive workload can hugely take benefit from the enhanced randomness generation and acceleration capabilities of Quside’s RPUs. Different industries use a variety of randomized workloads. Monte Carlo based calculations are extensively used in the Financial industry for risk or asset valuation calculations, Engineering industries uses it for the simulation of material or fluxes behaviour, Logistic industries uses optimization algorithms to calculate the best routes, etc.
PHARMA/SCIENCE: Stochastic Simulations, Computational Chemistry, Drug Discovery
FINANCE/INSURANCE: Speed up and optimize Monte Carlo simulations, Portfolio Optimization, Risk Assessment
LOGISTICS/OPERATIONS RESEARCH: Scheduling and dispatching simulations, Inventory Optimization, Stochastic strategic and tactical Planning (forecasting)
Having the end-user in mind, the RPU is probably one of the easiest hardware accelerators to deploy in any kind of pipeline. Either if you have an on-prem cluster -either baremetal or virtualized- or you prefer to run your workloads on a cloud environment, the RPU is easy to deploy and integrate into your current stochastic runs.
Plenty: as long as you have a randomness-intensive workload, the RPU can be used to offload it all. Out-of-the-box, though, we provide these main functionalities:
- Provable performance improvements of up to 10X and energy efficiencies of up to 20x, compared with CPU computations.
- Easily deployed in C++, with wrappers in other languages (C, Python)
- Virtual Machine images for on-premise deployments.
- 10 Gb/s uniform, normal, and logarithmic arbitrary-sized floating-point distributions
- Repeatability and reproducibility of simulations
- Avoid artifacts and co-dependency risk
- Minimal code migration cost
- Available for on-prem and via Quside Cloud Partners
