Tonatiuh++ Benchmark: Target-Aligned Heliostat Field of Pentagonal Heliostats Reference Dataset

onatiuh++ Benchmark is the official reproducible benchmark dataset for Tonatiuh++, the open-source Monte Carlo ray tracer for solar concentrating systems. The benchmark models a large-scale solar central receiver system composed of 1,971 pentagonal heliostats with target-aligned tracking reflecting sunlight onto a tilted flat receiver located approximately 80 metres above ground. The benchmark methodology remains benchmark_v1. The reference dataset included in this release is benchmark reference v2, generated with the deterministic per-chunk std::mt19937_64 random-number subsystem introduced in Tonatiuh++ v0.1.8.26. The scene, ray-tracing physics, benchmark calculations, and acceptance criteria are unchanged; only the deterministic random-number streams used to generate the official reference results have changed. The dataset contains: benchmark_heliostat_field_v1.tnhpp: frozen Tonatiuh++ scene used for the benchmark. benchmark_config_500M_v2.json: benchmark configuration for the 500-million-ray reference simulation. benchmark_reference_500M_v2.json: official benchmark reference values and metadata. benchmark_reference_flux_grid_500M_v2.csv: reference flux grid in CSV format. benchmark_reference_flux_grid_500M_v2.bin: reference flux grid in binary format. README.md: documentation describing the benchmark dataset and file formats. The official reference simulation was generated using: 500,000,000 rays; seed 123456789; chunk size 10,000 rays; a 100 × 100 target flux grid. The worker count is intentionally not part of the scientific reference, as the deterministic per-chunk random-number streams produce identical benchmark results independently of worker scheduling. The benchmark is intended for: reproducibility validation; regression testing; performance benchmarking; cross-platform verification of Tonatiuh++ installations. The binary reference flux grid uses: row-major ordering; IEEE 754 binary64 floating-point values; little-endian byte order; no header or metadata. The benchmark can be executed using the Tonatiuh++ headless benchmark mode, available since Tonatiuh++ v0.1.8.20: tonatiuhpp.exe --headless benchmark benchmark_config_500M_v2.json The generated benchmark results are automatically compared against the official reference dataset. A successful execution verifies the total collected power, the maximum flux, and the SHA-256 hash of the generated binary flux grid against the published reference values. This benchmark reference supersedes the previous benchmark reference v1 while preserving the same benchmark definition, scene, and validation methodology. Only the deterministic random-number subsystem has changed, enabling reproducible benchmark results across platforms and independent of worker scheduling.