Search

Particle transport simulations are a cornerstone of high-energy physics (HEP), constituting almost half of the entire computing workload performed in HEP. To boost the simulation throughput and energy efficiency, GPUs as accelerators have been explored in recent years, further driven by the increasing use of GPUs on HPCs. The Accelerated demonstrator of electromagnetic Particle Transport (AdePT) is an advanced prototype for offloading the simulation of electromagnetic showers in Geant4 to GPUs, and still undergoes continuous development and optimization. Improving memory layout and data access is vital to use modern, massively parallel GPU hardware efficiently, contributing to the challenge of migrating traditional CPU based data structures to GPUs in AdePT. The low-level abstraction of memory access (LLAMA) is a C++ library that provides a zero-runtime-overhead data structure abstraction layer, focusing on multidimensional arrays of nested, structured data. It provides a framework for defining and switching custom memory mappings at compile time to define data layouts and instrument data access, making LLAMA an ideal tool to tackle the memory-related optimization challenges in AdePT. Our contribution shares insights gained with LLAMA when instrumenting data access inside AdePT, complementing traditional GPU profiler outputs. We demonstrate traces of read/write counts to data structure elements as well as memory heatmaps. The acquired knowledge allowed for subsequent data layout optimizations.

The performance gap between CPU and memory widens continuously. Choosing the best memory layout for each hardware architecture is increasingly important as more and more programs become memory bound. For portable codes that run across heterogeneous hardware architectures, the choice of the memory layout for data structures is ideally decoupled from the rest of a program. This can be accomplished via a zero-runtime-overhead abstraction layer, underneath which memory layouts can be freely exchanged. We present the Low-Level Abstraction of Memory Access (LLAMA), a C++ library that provides such a data structure abstraction layer with example implementations for multidimensional arrays of nested, structured data. LLAMA provides fully C++ compliant methods for defining and switching custom memory layouts for user-defined data types. The library is extensible with third-party allocators. Providing two close-to-life examples, we show that the LLAMA-generated AoS (Array of Structs) and SoA (Struct of Arrays) layouts produce identical code with the same performance characteristics as manually written data structures. Integrations into the SPEC CPU® lbm benchmark and the particle-in-cell simulation PIConGPU demonstrate LLAMA's abilities in real-world applications. LLAMA's layout-aware copy routines can significantly speed up transfer and reshuffling of data between layouts compared with naive element-wise copying. LLAMA provides a novel tool for the development of high-performance C++ applications in a heterogeneous environment.

The performance gap between CPU and memory widens continuously. Choosing the best memory layout for each hardware architecture is increasingly important as more and more programs become memory bound. For portable codes that run across heterogeneous hardware architectures, the choice of the memory layout for data structures is ideally decoupled from the rest of a program. This can be accomplished via a zero-runtime-overhead abstraction layer, underneath which memory layouts can be freely exchanged. We present the Low-Level Abstraction of Memory Access (LLAMA), a C++ library that provides such a data structure abstraction layer with example implementations for multidimensional arrays of nested, structured data. LLAMA provides fully C++ compliant methods for defining and switching custom memory layouts for user-defined data types. The library is extensible with third-party allocators. Providing two close-to-life examples, we show that the LLAMA-generated AoS (Array of Structs) and SoA (Struct of Arrays) layouts produce identical code with the same performance characteristics as manually written data structures. Integrations into the SPEC CPU® lbm benchmark and the particle-in-cell simulation PIConGPU demonstrate LLAMA's abilities in real-world applications. LLAMA's layout-aware copy routines can significantly speed up transfer and reshuffling of data between layouts compared with naive element-wise copying. LLAMA provides a novel tool for the development of high-performance C++ applications in a heterogeneous environment.

Particle transport simulations are a cornerstone of high-energy physics (HEP), constituting almost half of the entire computing workload performed in HEP. To boost the simulation throughput and energy efficiency, GPUs as accelerators have been explored in recent years, further driven by the increasing use of GPUs on HPCs. The Accelerated demonstrator of electromagnetic Particle Transport (AdePT) is an advanced prototype for offloading the simulation of electromagnetic showers in Geant4 to GPUs, and still undergoes continuous development and optimization. Improving memory layout and data access is vital to use modern, massively parallel GPU hardware efficiently, contributing to the challenge of migrating traditional CPU based data structures to GPUs in AdePT. The low-level abstraction of memory access (LLAMA) is a C++ library that provides a zero-runtime-overhead data structure abstraction layer, focusing on multidimensional arrays of nested, structured data. It provides a framework for defining and switching custom memory mappings at compile time to define data layouts and instrument data access, making LLAMA an ideal tool to tackle the memory-related optimization challenges in AdePT. Our contribution shares insights gained with LLAMA when instrumenting data access inside AdePT, complementing traditional GPU profiler outputs. We demonstrate traces of read/write counts to data structure elements as well as memory heatmaps. The acquired knowledge allowed for subsequent data layout optimizations.

The performance gap between CPU and memory widens continuously. Choosing the best memory layout for each hardware architecture is increasingly important as more and more programs become memory bound. For portable codes that run across heterogeneous hardware architectures, the choice of the memory layout for data structures is ideally decoupled from the rest of a program. This can be accomplished via a zero-runtime-overhead abstraction layer, underneath which memory layouts can be freely exchanged. We present the Low-Level Abstraction of Memory Access (LLAMA), a C++ library that provides such a data structure abstraction layer with example implementations for multidimensional arrays of nested, structured data. LLAMA provides fully C++ compliant methods for defining and switching custom memory layouts for user-defined data types. The library is extensible with third-party allocators. Providing two close-to-life examples, we show that the LLAMA-generated AoS (Array of Structs) and SoA (Struct of Arrays) layouts produce identical code with the same performance characteristics as manually written data structures. Integrations into the SPEC CPU® lbm benchmark and the particle-in-cell simulation PIConGPU demonstrate LLAMA's abilities in real-world applications. LLAMA's layout-aware copy routines can significantly speed up transfer and reshuffling of data between layouts compared with naive element-wise copying. LLAMA provides a novel tool for the development of high-performance C++ applications in a heterogeneous environment.

The performance gap between CPU and memory widens continuously. Choosing the best memory layout for each hardware architecture is increasingly important as more and more programs become memory bound. For portable codes that run across heterogeneous hardware architectures, the choice of the memory layout for data structures is ideally decoupled from the rest of a program. This can be accomplished via a zero-runtime-overhead abstraction layer, underneath which memory layouts can be freely exchanged. We present the Low-Level Abstraction of Memory Access (LLAMA), a C++ library that provides such a data structure abstraction layer with example implementations for multidimensional arrays of nested, structured data. LLAMA provides fully C++ compliant methods for defining and switching custom memory layouts for user-defined data types. The library is extensible with third-party allocators. Providing two close-to-life examples, we show that the LLAMA-generated AoS (Array of Structs) and SoA (Struct of Arrays) layouts produce identical code with the same performance characteristics as manually written data structures. Integrations into the SPEC CPU® lbm benchmark and the particle-in-cell simulation PIConGPU demonstrate LLAMA's abilities in real-world applications. LLAMA's layout-aware copy routines can significantly speed up transfer and reshuffling of data between layouts compared with naive element-wise copying. LLAMA provides a novel tool for the development of high-performance C++ applications in a heterogeneous environment.

Objective: Bevacizumab maintenance following platinum-based chemotherapy is an effective treatment for epithelial ovarian cancer (EOC), both in primary and recurrent disease. Our aim was to identify criteria to select elderly patients who can safely benefit from bevacizumab addition. Methods: This is a case-control study on patients with primary or recurrent EOC who received platinum-based chemotherapy plus bevacizumab, between January 2015 and December 2016. Patient characteristics, treatment details and adverse events were reviewed and analyzed in 2 settings: younger (<65 years, group 1) and elderly (≥65 years, group 2). A binary logistic model was applied to correlate clinical variables and severe (grade ≥3) toxicity risk. Results: Overall, 283 patients with EOC were included, with 72 (25.4%) older patients compared with 211 (74.6%) younger women. Bevacizumab had been administered to 234 patients (82.7%) as first-line treatment and in 49 (17.3%) with recurrent disease. At diagnosis, elderly patients presented with at least one comorbidity and were taking at least 1 medication in 84.7% and 80.6% of the cases respectively, compared with correspondingly 47.4% and 37.4% in group 1 (p<0.001). Nonetheless, the occurrence of serious (grade ≥3) adverse events did not increase among the older group. Creatinine serum levels >1.1 g/dL, estimated glomerular filtration rate (eGFR) ≤60 mL/min, ≥3 comorbidities were independently associated with a higher severe toxicity. Conclusions: Elderly patients with EOC can safely be treated with bevacizumab; factors other than age, as higher creatinine serum levels, eGFR and number of comorbidities should be considered to better estimate bevacizumab-related toxicity risk.

Borderline ovarian tumors (BOT) are uncommon but not rare epithelial ovarian neoplasms, intermediate between benign and malignant categories. Emerging knowledge supports the notion that subtypes of borderline ovarian tumors comprise distinct biologic, pathogenetic, and molecular entities, precluding a single unifying concept for BOT. The identification of valuable markers for the diagnosis and classification of these tumors is in need. Among the molecular candidates, the Retinoblastoma (Rb) family members Rb/p105 and Rb2/p130 seem to play a pivotal role in ovarian cancer. In particular, Rb/p105, when in the unphosphorylated form, acts as a growth suppressor controlling cell cycle and tumor progression; whereas, the phosphorylated form activates gene transcription and cellular proliferation. While Rb/p105 is ubiquitously confined to the nuclei of cycling and quiescent cells, Rb2/p130 activity is also regulated by intracellular localization. According to this, Rb family members could represent a novel marker in diagnosis and classification risk for patients with BOT. In this study, we evaluated the expression and subcellular localization of proteins of the retinoblastoma (Rb) gene family in 65 ovarian borderline tumors. Statistically significant differences were found in nuclear and cytoplasmic expressions of Rb/p105 and Rb2/p130 according to different examined histotypes. In detail, the nuclear expression of Rb/p105 and Rb2/p130 was more frequently detected in serous (84.6%) than sero-mucinous (42.1%) and mucinous (50%) types. Conversely, the cytoplasmic expression of Rb2/p130 was not detected in serous tumors and frequently observed in mucinous subtypes (80%). Our findings suggest that Rb proteins do not play a key role in the tumor progression of serous borderline tumors since any cases showed cytoplasmic localization. By contrast, the observed higher cytoplasmic expression of Rb2/p130 in intestinal mucinous BOTs is indicative of Rb protein family involvement in th

Particle physics has an ambitious and broad experimental programme for the coming decades. This programme requires large investments in detector hardware, either to build new facilities and experiments, or to upgrade existing ones. Similarly, it requires commensurate investment in the R&D of software to acquire, manage, process, and analyse the shear amounts of data to be recorded. In planning for the HL-LHC in particular, it is critical that all of the collaborating stakeholders agree on the software goals and priorities, and that the efforts complement each other. In this spirit, this white paper describes the R\&D activities required to prepare for this software upgrade.