Search

GASNet-EX is a portable, open-source, high-performance communication library designed to efficiently support the networking requirements of PGAS runtime systems and other alternative models in emerging exascale systems. It provides network-independent, high-performance communication primitives including Remote Memory Access (RMA) and Active Messages (AM). GASNet-EX is an evolution of the popular GASNet communication system, building upon over 20 years of lessons learned, and the primary goals are high performance, interface portability, and expressiveness. The library has been used to implement parallel programming models and libraries such as UPC, UPC++, Fortran coarrays, Legion, Chapel, and many others.This anthology collects together the four separate volumes that currently comprise the GASNet-EX specification, as of the 2024.5.0 release of GASNet-EX.

Tomographic imaging has benefited from advances in X-ray sources, detectors and optics to enable novel observations in science, engineering and medicine. These advances have come with a dramatic increase of input data in the form of faster frame rates, larger fields of view or higher resolution, so high performance solutions are currently widely used for analysis. Tomographic instruments can vary significantly from one to another, including the hardware employed for reconstruction: from single CPU workstations to large scale hybrid CPU/GPU supercomputers. Flexibility on the software interfaces and reconstruction engines are also highly valued to allow for easy development and prototyping. This paper presents a novel software framework for tomographic analysis that tackles all aforementioned requirements. The proposed solution capitalizes on the increased performance of sparse matrix-vector multiplication and exploits multi-CPU and GPU reconstruction over MPI. The solution is implemented in Python and relies on CuPy for fast GPU operators and CUDA kernel integration, and on SciPy for CPU sparse matrix computation. As opposed to previous tomography solutions that are tailor-made for specific use cases or hardware, the proposed software is designed to provide flexible, portable and high-performance operators that can be used for continuous integration at different production environments, but also for prototyping new experimental settings or for algorithmic development. The experimental results demonstrate how our implementation can even outperform state-of-the-art software packages used at advanced X-ray sources worldwide.

Summary: Los Alamos National Laboratory has deployed (over the last year and a half) a pair of Cray Shasta machines—a development testbed named Guaje and and production machine named Chicoma, which will soon comprise the bulk of LANL's open science research computing portfolio. In the process, we have encountered a number of problems and challenges in several realms—authentication and authorization, cluster health management, image management, and configuration management. Both independently and in collaboration with Cray/HPE, we have found solutions and brought the system into stable production. The presentation will discuss the solutions and how they came about, and issues we are working to resolve in the near future. [ABSTRACT FROM AUTHOR]

In this study, hydroxypropyl cellulose (HPC) was utilized as the raw material, with the addition of β-cyclodextrin (β-CD), citric acid (CA) as the crosslinking agent, and sodium hypophosphite (SHP) as the catalyst to produce hydroxypropyl cellulose/β-CD composite films. The inclusion of β-CD resulted in an increase in the tensile strength of the film, with the maximum value of 13.5 MPa for the 1 % β-CD composite membrane. Additionally, after degradation in soil for 28 days, the degradation ability was significantly enhanced, with the 1.0 % β-CD composite film exhibiting the highest degradation rate of 27.21 %. Furthermore, the water permeability of the composite membrane was improved with the addition of β-CD. Specifically, when the β-CD content was 1.0 %, the water vapor transmission reached its lowest point at 2,445 g* ( m 2 * 24 d ) − 1 ${({m}^{2}\ast 24d)}^{-1}$ . The findings demonstrated that the 1 % β-cyclodextrin/hydroxypropyl cellulose composite film effectively preserved the freshness of strawberries, reducing the weight loss rate by 1.65 % compared to the control group. In conclusion, this research highlights the potential for preparing composite membranes using HPC and β-CD crosslinking, thereby expanding the application of hydroxypropyl cellulose and β-CD in food preservation. [ABSTRACT FROM AUTHOR]

The Monte Carlo Computational Summit was held on the campus of the University of Notre Dame in South Bend, Indiana, USA on 25–26 October 2023. The goals of the summit were to discuss algorithmic and software alterations required for successfully porting respective code bases to exascale-class computing hardware, compare software engineering techniques used by various code teams, and consider the adoption of industry-standard benchmark problems to better facilitate code-to-code performance comparisons. Participants reported that identifying and implementing suitable Monte Carlo algorithms for GPUs continues to be a sticking point. They also report significant difficulty porting existing algorithms between GPU APIs (specifically Nvidia CUDA to AMD ROCm). To better compare code-to-code performance, participants decided to design a C5G7-like benchmark problem with a defined figure of merit, with the expectation of adding more benchmarks in the future. The participants also identified the need to explore the intermediate and long-term future of the Monte Carlo neutron transport community and how best to modernize and contextualize Monte Carlo as a useful tool in modern industry. Overall the summit was considered to be a success by the organizers and participants, and the group shared a strong desire for future, potentially larger, Monte Carlo summits. [ABSTRACT FROM AUTHOR]

The solution of reactive computational fluid dynamics (CFD) simulations is accelerated by the implementation of a hybrid central processing unit/graphics processing units (CPU/GPU) Finite Volume solver based on the operator‐splitting strategy, where the chemistry integration is treated independently of the flow solution. The integration of ordinary differential equations (ODEs) describing the finite‐rate chemical kinetics is solved by an adaptive multi‐block explicit solver on GPUs, while the load of the fluid solution is distributed on a multicore CPU algorithm. The resulting speed‐up for reactive CFD simulations is up to 10×$$ \times $$; the performance gain increases with the size of the mechanism. The proposed implementation is general and can be applied to any CFD problem where the governing equations for the fluid transport are coupled with an ODE system. Code validation is performed against reference solutions on a selection of test cases involving reacting flows. [ABSTRACT FROM AUTHOR]

The harsh geological conditions in the northwest region of China, characterized by widespread saline-alkali soil rich in alkali ions, pose a high risk of Alkali-Silica reaction (ASR) in concrete, particularly due to the presence of ASR-active natural river sands. To address ASR hazards, locally applied concrete often employs High-Performance concrete (HPC) prepared with high proportions of mineral admixtures. In this paper, the alkali content is controlled by adding mixed water with NaOH to the initial configuration of concrete, and three different alkali content states are set up. A 1 mol/L NaOH solution was used to simulate alkaline conditions, and HPC specimens were immersed for an extended period to investigate the effects of equivalent alkali content, immersion time, concrete strength, and admixture on the flexural mechanical properties of HPC under the condition of long-term alkali immersion. Results indicate that, the strength grade was positively correlated with the flexural strength of HPC, but the alkali content was negatively. Initial immersion significantly enhances strength, followed by a gradual decline after long-term immersion. Among three types of admixture addition methods, the impact on flexural strength of HPC immersed in alkaline solution for 10 years follows the order: Double doped air entraining agent and rust inhibitor is greater than single doped air entraining agent is greater than single doped rust inhibitor. In the process of macroscopic test, it is difficult to observe the variation rule of stress and strain in detail, only the final aggregate failure mode can be analyzed. In order to analyze the strain change of the specimen and the failure process of the aggregates more accurately, a three-dimensional random aggregate concrete mesoscopic model was established, and equations relating microhardness to the mechanical properties of concrete components were derived from statistical analysis, providing a basis for parameter selection in the model. Results demonstrate that with increasing strength, the occurrence time of initial cracks is delayed, and the ratio of cracks bypassing aggregates (cracks develop along the ITZ between aggregate and mortar until complete failure) decreases, and the ratio of cracks penetrating aggregates (cracks develop directly through aggregates in an almost vertical direction) increases. [ABSTRACT FROM AUTHOR]

The ability to accurately predict when a job on a supercomputer will leave the queue and start to run is not only beneficial for providing insights to users, but can also help enable non‐traditional HPC workloads that are not necessarily suited to the batch queue style‐approach that is ubiquitous on production HPC machines. However there are numerous challenges in achieving such a prediction with high accuracy, not least because the queue's state can change rapidly and depend upon many factors. In this work, we explore a novel machine learning approach for predicting queue wait times, hypothesising that such a model can capture the complex behavior resulting from the queue policy and other interactions to generate accurate job start times. For ARCHER2 (HPE Cray EX), Cirrus (HPE 8600), and 4‐cabinet (HPE Cray EX) we explore how different machine learning approaches and techniques improve the accuracy of our predictions, comparing against the estimation generated by Slurm. By combining categorization and regression models, we demonstrate that our approach delivers the most accurate predictions across our machines of interest, with the result of this work being the ability to predict job start times within 1 min of the actual start time for around 65% of jobs on ARCHER2 and 4‐cabinet, and 76% of jobs on Cirrus. When compared against what Slurm can deliver, via the backfill plugin, this represents around 3.8 times better accuracy on ARCHER2 and 18 times better for Cirrus. Furthermore our approach can accurately predicting the start time for three quarters of all job within 10 min of the actual start time on ARCHER2 and 4‐cabinet, and for 90% of jobs on Cirrus. Whilst the initial driver of this work was to better facilitate non‐traditional, interactive and urgent, workloads on HPC machines, the insights gained can also be used to provide wider benefits to users, enrich existing batch queue systems, and inform supercomputing center policy also. [ABSTRACT FROM AUTHOR]

Filesystems are essential components in contemporary computer systems that organize and manage data. Their performance is crucial in various applications, from web servers to data storage systems. This paper helps to pick the suitable filesystem by comparing btrfs with ZFS by considering multiple situations and applications, ranging from sequential and random performance in the most common use cases to extreme use cases like high-performance computing (HPC). It showcases each option's benefits and drawbacks, considering different usage scenarios. The performance of btrfs and ZFS will be evaluated through rigorous testing. They will assess their capabilities in handling huge files, managing numerous small files, and the speed of data read and write across varied usage levels. The analysis indicates no definitive answer; the selection of the optimal filesystem is contingent upon individual data-access requirements. [ABSTRACT FROM AUTHOR]

In regions characterized by the challenging combination of brine corrosion in the salt lakes and river sand with alkali silica reaction (ASR) activity in areas of the Northwest, high-performance concrete (HPC) formulated with high-volume composite mineral admixtures as ASR suppression measures has been preferred for civil engineering structures in the region. This study investigates the splitting tensile strength, corrosion products, microscopic structure characteristics, and mesoscopic mechanical mechanisms of splitting failure of such HPC under 10-year corrosion from salt lake brine. The relationship between mechanical properties and corrosion damage, as well as the characteristics of internal crack propagation paths and failure mechanisms of HPC under splitting load, are explored. The findings reveal that as the alkali content within HPC rises, corrosion damage intensifies, resulting in a reduction in splitting tensile strength. Moreover, a linear association between mechanical properties and corrosion damage is observed. Microscopic structural analysis and numerical simulation of the splitting failure process of HPC elucidate that while the substantial presence of mineral admixtures effectively suppresses the ASR risk associated with alkali-reactive aggregates in concrete, uneven ASR gel products persist. These discontinuous micro-fine interface cracks induced by the gel products and the cracks induced by the gel products around the selective alkali-active aggregate particles distributed in the local area are the initiation sources of mortar cracks in HPC splitting failure. In terms of the overall failure state observed during the concrete splitting process, mortar cracks manifest two distinct extension paths: along the coarse aggregate interface and directly through the aggregates themselves. Notably, a greater proportion of coarse aggregates are directly penetrated by mortar cracks, as opposed to the number of interface failures bypassing coarse aggregates. More importantly, the above work establishes a theoretical reference in three dimensions: macroscopic, mesoscopic, and microscopic, for studying concrete corrosion damage in complex environments such as salt lake brine corrosion and ASR inhibition. [ABSTRACT FROM AUTHOR]

Virtual screening is an early stage in the drug discovery process that selects the most promising candidates. In the urgent computing scenario, finding a solution in the shortest time frame is critical. Any improvement in the performance of a virtual screening application translates into an increase in the number of candidates evaluated, thereby raising the probability of finding a drug. In this paper, we show how we can improve application throughput using Out-of-kernel optimizations. They use input features, kernel requirements, and architectural features to rearrange the kernel inputs, executing them out of order, to improve the computation efficiency. These optimizations' implementations are designed on an extreme-scale virtual screening application, named LiGen, that can hinge on CUDA and SYCL kernels to carry out the computation on modern supercomputer nodes. Even if they are tailored to a single application, they might also be of interest for applications that share a similar design pattern. The experimental results show how these optimizations can increase kernel performance by 2 × , respectively, up to 2.2 × in CUDA and up to 1.9 × , in SYCL. Moreover, the reported speedup can be achieved with the best-proposed parameterization, as shown by the data we collected and reported in this manuscript. [ABSTRACT FROM AUTHOR]

Developing and executing large-scale data analysis applications in parallel and distributed environments can be a complex and time-consuming task. Developers often find themselves diverted from their application logic to handle technical details about the underlying runtime and related issues. To simplify this process, ParSoDA, a Java library, has been proposed to facilitate the development of parallel data mining applications executed on HPC systems. It simplifies the process by providing built-in scalability mechanisms relying on the Hadoop and Spark frameworks. This paper presents ParSoDA-Py, the Python version of the ParSoDA library, which allows for further support of commonly used runtimes and libraries for big data analysis. After a complete library redesign, ParSoDA can be now easily integrated with other Python-based distributed runtimes for HPC systems, such as COMPSs and Apache Spark, and with the large ecosystem of Python-based data processing libraries. The paper discusses the adaptation process, which takes into consideration the new technical requirements, and evaluates both usability and scalability through some case study applications. [ABSTRACT FROM AUTHOR]

In the present study, the compressive strength of binary and ternary blended high-performance concrete mixtures was modeled using different machine learning tools, such as artificial neural network (ANN), gradient tree boosting (GTB), and multivariate adaptive regression splines (MARS). The compressive strength of binary and ternary blended concrete at 28 days of curing period was modeled using seven quantitative input parameters such as cement, blast furnace slag, fly ash, superplasticizer, water, coarse aggregate, and fine aggregate based on three different training and testing (Tr–Te) scenarios. The performance of the models developed was analyzed based on several statistical evaluation metrics, of error and efficiency. During the testing phase, the compressive strength estimates obtained via modeling using ANN, GTB, and MARS had Kling–Gupta efficiency (KGE) values of 0.8389, 0.8602, and 0.8423, respectively, for the first Tr–Te scenario; similarly, the KGE values were 0.8025/0.8830, 0.8541/0.8901, and 0.8434/0.8582, respectively, for the second/third Tr–Te scenarios. The estimation accuracy of the GTB model was relatively superior to that of the ANN and MARS models, taking into consideration both the error and efficiency indices. All three models perform relatively well for the first I/O combination compared to the other two combinations. [ABSTRACT FROM AUTHOR]

Background: Standard flow cytometry protocols for CD34+ cell enumeration designed for fresh samples are not appropriate for cryopreserved products. Special protocols have been developed to remove the cryoprotectant by quickly washing a freshly thawed sample. Exposing cells to a large volume of hypotonic solution and subsequent washing process was hypothesized to cause lab‐induced cell death. Moreover, standard gating strategies must be altered to avoid reporting falsely high viabilities. Study Design and Methods: We developed a novel method whereby thawed samples were diluted step‐wise to 1:2 by 3 additions of 1/3 sample volume using 1% Human Albumin in Dextran 40 (10% Low Molecular Weight Dextran in 0.9% NaCl) separated by 5 min between each addition. An additional 1:10 dilution was required to obtain a desired cell concentration for flow cytometry testing resulting in a 1:20 dilution. Results: Twenty samples were tested simultaneously in a method comparison; the new method demonstrated significant increases in mean cell viabilities for white blood cells, hematopoietic progenitor cells, and T cells as well as reduced standard deviations for each parameter. Discussion: Slow, step‐wise dilutions of freshly thawed samples of cryopreserved apheresis products to 1:20 yielded higher and more precise viability measurements compared to quickly washing samples to remove DMSO. [ABSTRACT FROM AUTHOR]

In recent years, significant strides have been made to enable Quantum ESPRESSO (QE) to operate on heterogeneous HPC architectures, effectively harnessing the computational capabilities of GPUs. Initially, emphasis has been placed on NVIDIA-based hardware and software infrastructure. Nonetheless, the HPC landscape is multifaceted and intricate, prompting the project to prioritize achieving peak performance across various GPU models. To this end, efforts are being directed towards empowering QE to leverage AMD and Intel GPUs through OpenMP offload directives. In this paper, coding approach and benchmark tests of the OpenMP porting of Plane-Wave Self-Consistent Field (PWSCF) code is presented and discussed. [ABSTRACT FROM AUTHOR]

Drug discovery is a long and costly process. Recent studies demonstrated how the introduction of an in-silico stage, named virtual screening, that suggests which molecule to test in-vitro, increases the drug discovery success probability. In the context of urgent computing, where it is important to find a therapeutic solution in a short time frame, the number of candidates that we can virtual screen is limited only by the available computation power. In this paper, we focus on LiGen, the virtual screening application of the EXSCALATE platform. In particular, we address two challenges of performing an extreme-scale virtual screening on a modern HPC system. The first one is posed by hardware heterogeneity, where GPUs of different vendors account for a large fraction of their performance. The second challenge concerns the operational difficulties of running the campaign since it requires significant effort and technical skills that are not common among domain experts. We show how hinging on SYCL and the LEXIS Platform, is the solution that the EXSCALATE Platform uses to address these challenges. [ABSTRACT FROM AUTHOR]

The application of Artificial Intelligence (AI) and Machine Learning (ML) models is increasingly leveraged to automate and optimize Data Centre (DC) operations. However, the interpretability and transparency of these complex models pose critical challenges. Hence, this paper explores the Shapley Additive exPlanations (SHAP) values model explainability method for addressing and enhancing the critical interpretability and transparency challenges of predictive maintenance models. This method computes and assigns Shapley values for each feature, then quantifies and assesses their impact on the model's output. By quantifying the contribution of each feature, SHAP values can assist DC operators in understanding the underlying reasoning behind the model's output in order to make proactive decisions. As DC operations are dynamically changing, we additionally investigate how SHAP can capture the temporal behaviors of feature importance in the dynamic DC environment over time. We validate our approach with selected predictive models using an actual dataset from a High-Performance Computing (HPC) DC sourced from the Enea CRESCO6 cluster in Italy. The experimental analyses are formalized using summary, waterfall, force, and dependency explanations. We delve into temporal feature importance analysis to capture the features' impact on model output over time. The results demonstrate that model explainability can improve model transparency and facilitate collaboration between DC operators and AI systems, which can enhance the operational efficiency and reliability of DCs by providing a quantitative assessment of each feature's impact on the model's output. [ABSTRACT FROM AUTHOR]

Influence of heat treatment on the bond creep of high‐strength concrete The bond between concrete and reinforcing steel facilitates the transfer of forces between the two building materials. The mechanical interlock of the ribs with the surrounding concrete results in high compressive stresses on the concrete corbels. Under permanent load, this leads to creeping and thus deformation (bond creep). The relative displacement (slip) between the reinforcing bar and the concrete increases steadily, which be detrimental for crack widths and durability. In this article, the reduction of bond creep by an initial, rapid heat treatment at 80 °C with 60 % relative humidity and durations of 4, 16 or 24 h is investigated experimentally. This involves subjecting specimens made of high‐strength concrete – after the appropriate heat treatment – to a constant permanent load for 300 days to pull‐out and recording the gradual increase in relative displacements. Non‐tempered specimens serve for reference. The unloaded samples are used to separate creep deformations from shrinkage effects. Moreover, undesired bending effects from the inevitable curved reinforcing steel are identified and eliminated by calculation. The results show that the heat treatment reduces the bond creep by around 40 %. [ABSTRACT FROM AUTHOR]

Cloud computing platforms have been continuously evolving. Features such as the Elastic Fabric Adapter (EFA) in the Amazon Web Services (AWS) platform have brought yet another revolution in the High Performance Computing (HPC) world, further accelerating the convergence of HPC and cloud computing. Other public clouds also support similar features further fueling this change. In this paper, we show how and why the performance of a large-scale computational fluid dynamics (CFD) HPC application on AWS competes very closely with the one on Beskow—a Cray XC40 supercomputer at the PDC Center for High-Performance Computing - in terms of cost-efficiency with strong scaling up to 2304 processes. We perform an extensive set of micro and macro benchmarks in both environments and conduct a comparative analysis. Until as recently as 2020 these benchmarks have notoriously yielded unsatisfactory results for the cloud platforms compared with on-premise infrastructures. Our aim is to access the HPC capabilities of the cloud, and in general to demonstrate how researchers can scale and evaluate the performance of their application in the cloud. [ABSTRACT FROM AUTHOR]

The development of better, more reliable and more efficient susceptibility assessments for shallow landslides is becoming increasingly important. Physically based models are well‐suited for this, due to their high predictive capability. However, their demands for large, high‐resolution and detailed input datasets make them very time‐consuming and costly methods. This study investigates if a spatially transferable model calibration can be created with the use of parameter ensembles and with this alleviate the time‐consuming calibration process of these methods. To investigate this, the study compares the calibration of the model TRIGRS in two different study areas. The first study area was taken from a previous study where the dynamic physically based model TRIGRS was calibrated for the Laternser valley in Vorarlberg, Austria. The calibrated parameter ensemble and its performance from this previous study are compared with a calibrated parameter ensemble of the model TRIGRS for the Passeier valley in South Tyrol, Italy. The comparison showed very similar model performance and large similarities in the calibrated geotechnical parameter values of the best model runs in both study areas. There is a subset of calibrated geotechnical parameter values that can be used successfully in both study areas and potentially other study areas with similar lithological characteristics. For the hydraulic parameters, the study did not find a transferable parameter subset. These parameters seem to be more sensitive to different soil types. Additionally, the results of the study also showed the importance of the inclusion of detailed information on the timing of landslide initiation in the calibration of the model. [ABSTRACT FROM AUTHOR]

Probabilistic domain decomposition (PDD) is an alternative paradigm for solving boundary value problems (BVPs) in parallel with excellent scalability properties, thanks to its reliance on stochastic representations of the BVP. However, there are cases when the latter is less numerically convenient, or unknown. Semilinear elliptic BVPs and the Helmholtz equation are prominent examples of either class. In this paper, we overcome this issue by designing suitable iterative schemes for either problem. These schemes not only retain the desirable properties of PDD but also are optimally suited for pathwise variance reduction, resulting in a systematic, nearly cost-free reduction of the statistical error through the iterations. Numerical tests carried out on the supercomputer Marconi100 are presented. [ABSTRACT FROM AUTHOR]

The Italian territory is located at the heart of one of the global hot spots of climate change, where the implementation of climate‐smart land management practices is imperative to guarantee the present and future maintenance of ecosystem functions as well as the sustainability of human socioeconomic activities. The project HIGHLANDER (HIGH performance computing to support smart LAND sERvices) led by Cineca aims at building a comprehensive and multi‐sector framework for land‐management decision‐making in Italy. The project relies on high quality information on different components of the landscape, with a focus on climate‐driven processes, and state‐of‐the‐art computing infrastructures. The HIGHLANDER Data Portal maximizes the impact of HIGHLANDER results by providing access to data products and services. In this article, we describe the architectural features of the platform, as well as the available HIGHLANDER datasets and downstream applications. [ABSTRACT FROM AUTHOR]

BackgroundMCA bifurcation aneurysms pose treatment challenges because of the complex hemodynamics at the bifurcation and the risk of rupture. FDS implantation has been controversial and there are only limited reports. Therefore, the aim of this study was to assess the efficacy and safety of this treatment strategy using p64 MW HPC and p48 MW HPC FDSs for MCA bifurcation aneurysms, compared with the p64 classic FDS.Materials and methodsWe retrospectively analyzed our institutional database and identified all patients with saccular, non-ruptured MCA bifurcation aneurysms treated with p64 MW HPC, p48 MW HPC, or p64 classic FDS implantation alone. Aneurysms with implantation of additional devices in the same session, previous treatments, and acutely ruptured and fusiform aneurysms were excluded.ResultsA total of 79 aneurysms met the inclusion criteria: 23 receiving a p64 MW HPC, 34 receiving a p48 MW HPC, and 22 receiving a p64 classic FDS. The occlusion rate was highest for the p48 MW HPC 2 mm FDS, at 88.9% at FU2, compared with 72.2% for the p64 MW HPC and 70.6% for the p64 classic. The time to aneurysm occlusion was shortest with the p64 MW HPC, at 178.31 days. The highest retreatment rate was observed with the p48 MW HPC 3 mm.ConclusionTreatment of MCA bifurcation aneurysms with a p48 MW HPC 2 mm or p64 MW HPC FDS is a safe and reliable strategy achieving high aneurysm occlusion rates - attributable to their lower porosity in relation to the parent vessel diameter as compared to the p48 MW HPC 3 mm FDS-, with reasonable morbidity and mortality.

Masking is an effective countermeasure against side-channel attacks. It replaces every logic gate in a computation by a gadget that performs the operation over secret sharings of the circuit’s variables. When masking is implemented in hardware, care should be taken to protect against leakage from glitches, which could otherwise undermine the security of masking. This is generally done by adding registers, which stop the propagation of glitches, but introduce additional latency and area cost. In masked pipeline circuits, a high latency further increases the area overheads of masking, due to the need for additional registers that synchronize signals between pipeline stages. In this work, we propose a technique to minimize the number of such pipeline registers, which relies on optimizing the scheduling of the computations across the pipeline stages. We release an implementation of this technique as an open-source tool, Compress. Further, we introduce other optimizations to deduplicate logic between gadgets, perform an optimal selection of masked gadgets, and introduce new gadgets with smaller area. Overall, our optimizations lead to circuits that improve the state-of-the art in area and achieve state-of-the-art latency. For example, a masked AES based on an S-box generated by Compress reduces latency by 19% and area by 27% over a state-of-the-art implementation, or, for the same latency, reduces area by 45%.