Allpix 2 (read: Allpix Squared) is a generic, open-source software framework for the simulation of silicon pixel detectors. Its goal is to ease the implementation of detailed simulations for both single detectors and more complex setups such as beam telescopes from incident radiation to the digitised detector response. Predefined detector types can be automatically constructed from simple model files describing the detector parameters.The simulation chain is arranged with the help of intuitive configuration files and an extensible system of modules, which implement separate simulation steps such as realistic charge carrier deposition with the Geant4 toolkit or propagation of charge carriers in silicon using a drift-diffusion model. Detailed electric field maps imported from TCAD simulations can be used to precisely model the drift behaviour of charge carriers within the silicon, bringing a new level of realism to Monte Carlo based simulations of particle detectors.This paper provides an overview of the framework and a selection of different simulation modules, and presents a comparison of simulation results with test beam data recorded with hybrid pixel detectors. Emphasis is placed on the performance of the framework itself, using a first-principles simulation of the detectors without addressing secondary ASIC-specific effects.

Heterogeneous computers combine a general-purpose host processor with domain-specific programmable many-core accelerators, uniting high versatility with high performance and energy efficiency. While the host manages ever-more application memory, accelerators are designed to work mainly on their local memory. This difference in addressed memory leads to a discrepancy between the optimal address width of the host and the accelerator. Today 64-bit host processors are commonplace, but few accelerators exceed 32-bit addressable local memory, a difference expected to increase with 128-bit hosts in the exascale era. Managing this discrepancy requires support for multiple data models in heterogeneous compilers. So far, compiler support for multiple data models has not been explored, which hampers the programmability of such systems and inhibits their adoption.In this work, we perform the first exploration of the feasibility and performance of implementing a mixed-data-model heterogeneous system. To support this, we present and evaluate the first mixed-data-model compiler, supporting arbitrary address widths on host and accelerator. To hide the inherent complexity and to enable high programmer productivity, we implement transparent offloading on top of OpenMP. The proposed compiler techniques are implemented in LLVM

No abstract

No abstract

No abstract