Benchmarking
Radiation Transport Modeling
Understanding propagation of charged particles through a material is vital for industries requiring shielding for ionizing radiation. Hospitals, high-energy physics laboratories, and nuclear energy facilities need to provide effective shielding of radiation to ensure regulatory compliance and to minimize radiation exposure of humans on site. The thickness of shielding required depends on the choice of material as well as the energy and type of ionizing radiation to be shielded against. Because of the health and safety implications of radiation shielding and the cost of implementing it, organizations use radiation transport simulations to estimate the amount of shielding required for a given site’s layout. One of the key components in these simulations uses the Monte Carlo method to estimate the distribution (and mean value) of radiation absorbed by a given material.
This application kernel models a high-energy proton propagating through a concrete block and calculates the energy absorbed by the concrete from the proton through electron scattering (using the Bethe-Bloch formula) and any additional inelastic collisions which may occur.
When running on the Signaloid Cloud Compute Engine (SCCE), the kernels implementing the calculation can replace the usual approach of sampling followed by application of the kernel to these sample inputs, replacing it with a direct computation on a representation of the probability distributions. Thus, in a single computation, the code kernel running on SCCE can compute the same type of output distribution that would take a Monte Carlo simulation thousands or hundreds of thousands of iterations.
The ionizing radiation absorption model running on the Signaloid Cloud Compute Engine with a single-threaded Signaloid C0Pro-S+ core achieves runtimes 56x faster than an optimized C-language Monte-Carlo-based implementation of the same kernel running on an Amazon EC2 R7iz instance. With 95% confidence, an 80k-iteration Monte Carlo implementation will have the same accuracy as the Signaloid UxHw®-based version, yet the Signaloid UxHw-based version gives the speedup quoted above; if requiring higher confidence in the accuracy of the Monte Carlo compared to UxHw, the speedups of UxHw are even greater.
The plots here show the total energy absorbed by a concrete block 0.1m thick from a proton with an initial energy of 800 MeV. The example is set up with the probability of an inelastic collision at 5% for each 1mm of concrete. Following the standard approach to such radiation transport models, the energy absorbed is equal to E(1 - U), where E is the proton’s current energy and U is a uniform distribution.
Plot of output distribution when running on a Signaloid C0Pro-S+ core that provides the 56.4x speedup.
Plot of the output of an 81.2K-iteration Monte Carlo for this use case. This Monte Carlo iteration count provides the same or better Wasserstein distance to ground truth (81.2K-iteration) Monte Carlo as execution on a Signaloid C0Pro-S+ core (which is 56.4x faster).
Plot of ground truth (20M-iteration) Monte Carlo
Benchmarking Methodology
Monte Carlo simulations work by statistical sampling and therefore each multi-iteration Monte Carlo run will result in a slightly different output distribution. By contrast, Signaloid's platform is deterministic and each run produces the same distribution for a given Signaloid C0 core type.
The performance improvements are calculated by comparing Signaloid's platform with a Monte Carlo simulation of a similar quality of distribution. First, we run a large Monte Carlo simulation (about 50M iterations) on an AWS r7iz high-performance AWS instance: We use this distribution result as a baseline or ground truth reference of distribution quality. Then we calculate the performance of Signaloid's technology, and compare it with the performance of a Monte Carlo iteration count where the output distribution's Wasserstein distance (to the output distribution of the ground truth reference) is as accurate as the Signaloid-core-executed algorithm's output distribution, with 95% confidence level.
Performance data based on Fall 2025 release of Signaloid's technology.