Proxy Application Research Articles

Diatom silicon isotope ratios in Quaternary research: Where do we stand?

Silicon stable isotope ratios (expressed as δ30Si) in biogenic silica have been widely used as a proxy for past and present biogeochemical cycling in both marine and lacustrine settings, in particular for nutrient utilization reconstructions. Yet an analysis of publication trends suggests a significant decline in the application of δ30Si to Quaternary science questions in the last five years. At the same time as δ30Si proxy applications have decreased, we are learning more about its complexities: an expanding body of work is highlighting biases, caveats or complications involved in the application of δ30Si-based approaches to the sediment record. These include the demonstration of species-specific silicon isotope fractionation factors (i.e. ‘vital effects’) or the potential for Fe or other trace metals to influence silicon isotope fractionation. Others have inferred the potential of biogenic silica dissolution to alter an initial δ30Si value, or questioned the preservation of the initial δ30Si through early diagenetic processes more generally. Another challenge receiving more attention is centered around deconvolving a δ30Si-value into a signal reflecting biological productivity and a signal reflecting changes in the δ30Si of dissolved silicon driven by whole-system and/or circulation changes. Finally, a number of studies focus on analytical difficulties, especially during sample preparation related to achieving and demonstrating a contaminant free biogenic silica. These challenges lead us to posit that the Quaternary science community is moving away from silicon isotope proxies because they are losing confidence in their reliability and usefulness. Here, focusing on the diatoms – the dominant biosilicifiers in both lakes and the ocean – we synthesize progress in understanding nuances and caveats of δ30Si-based proxies in order to answer whether the fall-off in δ30Si-based Quaternary research is warranted. We suggest that with some simple steps that can be readily implemented, and with the closing of key knowledge gaps, there is no reason to believe silicon isotopes do not have a promising future in the Quaternary sciences.

An automated OpenMP mutation testing framework for performance optimization

Performance optimization continues to be a challenge in modern HPC software. Existing performance optimization techniques, including profiling-based and auto-tuning techniques, fail to indicate program modifications at the source level thus preventing their portability across compilers. This paper describes Muppet, a new approach that identifies program modifications called mutations aimed at improving program performance. Muppet’s mutations help developers reason about performance defects and missed opportunities to improve performance at the source code level. In contrast to compiler techniques that optimize code at intermediate representations (IR), Muppet uses the idea of source-level mutation testing to relax correctness constraints and automatically discover optimization opportunities that otherwise are not feasible using the IR. We demonstrate the Muppet’s concept in the OpenMP programming model. Muppet generates a list of OpenMP mutations that alter the program parallelism in various ways, and is capable of running a variety of optimization algorithms such as delta debugging, Bayesian Optimization and decision tree optimization to find a subset of mutations which, when applied to the original program, cause the most speedup while maintaining program correctness. When Muppet is evaluated against a diverse set of benchmark programs and proxy applications, it is capable of finding sets of mutations that induce speedup in 75.9% of the evaluated programs.

Runtime support for CPU-GPU high-performance computing on distributed memory platforms

IntroductionHardware heterogeneity is here to stay for high-performance computing. Large-scale systems are currently equipped with multiple GPU accelerators per compute node and are expected to incorporate more specialized hardware. This shift in the computing ecosystem offers many opportunities for performance improvement; however, it also increases the complexity of programming for such architectures.MethodsThis work introduces a runtime framework that enables effortless programming for heterogeneous systems while efficiently utilizing hardware resources. The framework is integrated within a distributed and scalable runtime system to facilitate performance portability across heterogeneous nodes. Along with the design, this paper describes the implementation and optimizations performed, achieving up to 300% improvement on a single device and linear scalability on a node equipped with four GPUs.ResultsThe framework in a distributed memory environment offers portable abstractions that enable efficient inter-node communication among devices with varying capabilities. It delivers superior performance compared to MPI+CUDA by up to 20% for large messages while keeping the overheads for small messages within 10%. Furthermore, the results of our performance evaluation in a distributed Jacobi proxy application demonstrate that our software imposes minimal overhead and achieves a performance improvement of up to 40%.DiscussionThis is accomplished by the optimizations at the library level and by creating opportunities to leverage application-specific optimizations like over-decomposition.

Penerapan Metode Vulnerability Assessment untuk Identifikasi Keamanan Website berdasarkan OWASP ID Tahun 2021

Universities, as educational institutions, are potential targets of cyber attacks. This is inevitable problem, one of which the University of Papua (UNIPA). The purpose this research is to find the security gaps the UNIPA website based on OWASP ID in 2021 and implement mitigation. Type of research is quantitative research with Vulnerability Assessment and Penetration Testing Life Cycle (VAPT) method. The VAPT method in research goes through five stages, namely scope, information gathering, vulnerability assessment, risk assessment, and reporting. The object of research is UNIPA website. Data collection uses primary data, the results of scanning the Zed Attack Proxy (ZAP) application. Data obtained from alerts ID, alerts, risk, and OWASP ID as information on vulnerability of UNIPA website. Research data analysis using OWASP ID. The results our findings, the vulnerability of UNIPA website is influenced by two factors, website security weaknesses and user negligence. Vulnerabilities with alerts ID A1, A2, A3, A4 A5, and A6 are a group website security weaknesses. The solution, vulnerabilities need utilize special systems such as anti-CSRF, CSP, CDN, Strict-Transport-Security Header, and timestamp checking so that the website is proportional. Meanwhile, the vulnerability with alerts ID A7 is a classification of user negligence. The solution is users must use the latest version of the browser. Browsers with latest version have X-Content-Type-Options: nosniff security mechanism to prevent sniffing attacks.

SOMA: Observability, monitoring, and in situ analytics for exascale applications

SummaryWith the rise of exascale systems and large, data‐centric workflows, the need to observe and analyze high performance computing (HPC) applications during their execution is becoming increasingly important. HPC applications are typically not designed with online monitoring in mind, therefore, the observability challenge lies in being able to access and analyze interesting events with low overhead while seamlessly integrating such capabilities into existing and new applications. We explore how our service‐based observation, monitoring, and analytics (SOMA) approach to collecting and aggregating both application‐specific diagnostic data and performance data addresses these needs. We present our SOMA framework and demonstrate its viability with LULESH, a hydrodynamics proxy application. Then we focus on Astaroth, a multi‐GPU library for stencil computations, highlighting the integration of the TAU and APEX performance tools and SOMA for application and performance data monitoring.

Applied Machine Learning Predictive Analytics to SQL Injection Attack Detection and Prevention

Abstract: The back-end database is pivotal to the storage of the massive size of big data Internet exchanges stemming from cloud-hosted web applications to Internet of Things (IoT) smart devices. Structured Query Language (SQL) Injection Attack (SQLIA) remains an intruder’s exploit of choice on vulnerable web applications to pilfer confidential data from the database with potentially damaging consequences. The existing solutions of mostly signature approaches were all before the recent challenges of big data mining and at such lacks the functionality and ability to cope with new signatures concealed in web requests. An alternative Machine Learning (ML) predictive analytics provides a functional and scalable mining to big data in detection and prevention of SQLIA. Unfortunately, lack of availability of readymade robust corpus or data set with patterns and historical data items to train a classifier are issues well known in SQLIA research. In this project, we explore the generation of data set containing extraction from known attack patterns including SQL tokens and symbols present at injection points. Also, as a test case, we build a web application that expects dictionary word list as vector variables to demonstrate massive quantities of learning data. The data set is pre-processed, labelled and feature hashing for supervised learning. The trained classifier to be deployed as a web service that is consumed in a custom dot NET application implementing a web proxy Application Programming Interface (API) to intercept and accurately predict SQLIA in web requests thereby preventing malicious web requests from reaching the protected back-end database. This project demonstrates a full proof of concept implementation of an ML predictive analytics and deployment of resultant web service that accurately predicts and prevents SQLIA with empirical evaluations presented in Confusion Matrix (CM) and Receiver Operating Curve (ROC).

Impact of Seawater Inorganic Carbon Chemistry on Element Incorporation in Foraminiferal Shell Carbonate

AbstractReconstruction of the marine inorganic carbon system relies on proxy signal carriers, such as element/calcium (El/Ca) ratios in foraminiferal shells. Concentrations of boron, lithium, strontium, and sulfur have been shown to vary with carbonate system parameters, but when comparing individual proxy reconstructions based on these elements, they are rarely in complete agreement. This is likely caused by the simultaneous effects of multiple environmental factors on element incorporation. Culture experiments with benthic foraminifera have revealed that the shell's S/Ca reflects the carbon chemistry and can potentially be used as a proxy for seawater []. Aiming to investigate the application potential of sulfur incorporation for carbonate speciation reconstruction, we present S/Ca ratios in five planktonic foraminiferal species, namely Globigerina bulloides, Globigerinoides ruber albus, Globigerinoides ruber ruber, Trilobatus sacculifer, and Neogloboquadrina incompta from core‐top sediments in regions with contrasting [], [], temperature, and salinity. Analyses of B/Ca and Mg/Ca ratios are included here since these elements have been shown to depend to a certain degree on carbon system parameters (e.g., calcite saturation state and pH, respectively) as well. Moreover, foraminiferal Mg/Ca values covary with S/Ca values and thereby might compromise its proxy application. In contrast to previously published results, this new data set shows a positive correlation between the incorporation of sulfur in the foraminifer's shell and seawater []. As the incorporation of sulfur and magnesium are positively correlated, S/Mg values of the same foraminifera may be used to improve inorganic carbon system reconstructions.

A Dynamic Model for the Computation of Gesture Types for Image-based Software Agents

<p>Computer vision technology allows the computer to interact with a human operator to quickly complete the interpretation of events and improve the operational workflow of processing events. As computer vision technology continues to evolve, the cost of the equipment continues to increase. Therefore, the stability of the system can be ensured through the design of the middleware and the calculation of the auxiliary functions of the software agents. At present, the recognition of characters for image processing is based on the technology of image recognition, which can provide a more flexible user experience. However, the dilemma of contactless design lies in the processing and calculation of images, which should reduce the inconvenience caused by delays.This article uses a Raspberry Pi as an example of a computing proxy application. After the visual inspection, the verification operation of the software is carried out. Our system is the detection of hand position and movement, and the detection of hand mark position in reconnaissance. In addition, we simultaneously developed management and remote control events and connected to the remote edge computer. After that, we successfully completed the automatic control and serial application of two different edge computing recognition jobs, and verified the image vision computing based on the Raspberry Pi software agent, which can be used for image vision analysis and control applications.</p> <p> </p>

Barium in seawater: dissolved distribution, relationship to silicon, and barite saturation state determined using machine learning

Abstract. Barium is widely used as a proxy for dissolved silicon and particulate organic carbon fluxes in seawater. However, these proxy applications are limited by insufficient knowledge of the dissolved distribution of Ba ([Ba]). For example, there is significant spatial variability in the barium–silicon relationship, and ocean chemistry may influence sedimentary Ba preservation. To help address these issues, we developed 4095 models for predicting [Ba] using Gaussian process regression machine learning. These models were trained to predict [Ba] from standard oceanographic observations using GEOTRACES data from the Arctic, Atlantic, Pacific, and Southern oceans. Trained models were then validated by comparing predictions against withheld [Ba] data from the Indian Ocean. We find that a model trained using depth, temperature, and salinity, as well as dissolved dioxygen, phosphate, nitrate, and silicate, can accurately predict [Ba] in the Indian Ocean with a mean absolute percentage deviation of 6.0 %. We use this model to simulate [Ba] on a global basis using these same seven predictors in the World Ocean Atlas. The resulting [Ba] distribution constrains the Ba budget of the ocean to 122(±7) × 1012 mol and reveals oceanographically consistent variability in the barium–silicon relationship. We then calculate the saturation state of seawater with respect to barite. This calculation reveals systematic spatial and vertical variations in marine barite saturation and shows that the ocean below 1000 m is at equilibrium with respect to barite. We describe a number of possible applications for our model outputs, ranging from use in mechanistic biogeochemical models to paleoproxy calibration. Our approach demonstrates the utility of machine learning in accurately simulating the distributions of tracers in the sea and provides a framework that could be extended to other trace elements. Our model, the data used in training and validation, and global outputs are available in Horner and Mete (2023, https://doi.org/10.26008/1912/bco-dmo.885506.2).

Making applications faster by asynchronous execution: Slowing down processes or relaxing MPI collectives

Comprehending the performance bottlenecks at the core of the intricate hardware–software interactions exhibited by highly parallel programs on HPC clusters is crucial. This paper sheds light on the issue of automatically asynchronous MPI communication in memory-bound parallel programs on multicore clusters and how it can be facilitated. For instance, slowing down MPI processes by deliberate injection of delays can improve performance if certain conditions are met. This leads to the counter-intuitive conclusion that noise, independent of its source, is not always detrimental but can be leveraged for performance improvements. We employ phase-space graphs as a new tool to visualize parallel program dynamics. They are useful in spotting certain patterns in parallel execution that will easily go unnoticed with traditional tracing tools. We investigate five different microbenchmarks and applications on different supercomputer platforms: an MPI-augmented STREAM Triad, two implementations of Lattice-Boltzmann fluid solvers (D3Q19 and SPEChpc D2Q37), the LULESH and HPCG proxy applications.

HipBone: A performance-portable graphics processing unit-accelerated C++ version of the NekBone benchmark

We present hipBone, an open-source performance-portable proxy application for the Nek5000 (and NekRS) computational fluid dynamics applications. HipBone is a fully GPU-accelerated C++ implementation of the original NekBone CPU proxy application with several novel algorithmic and implementation improvements which optimize its performance on modern fine-grain parallel GPU accelerators. Our optimizations include a conversion to store the degrees of freedom of the problem in assembled form in order to reduce the amount of data moved during the main iteration and a portable implementation of the main Poisson operator kernel. We demonstrate near-roofline performance of the operator kernel on three different modern GPU accelerators from two different vendors. We present a novel algorithm for splitting the application of the Poisson operator on GPUs which aggressively hides MPI communication required for both halo exchange and assembly. Our implementation of nearest-neighbor MPI communication then leverages several different routing algorithms and GPU-Direct RDMA capabilities, when available, which improves scalability of the benchmark. We demonstrate the performance of hipBone on three different clusters housed at Oak Ridge National Laboratory, namely, the Summit supercomputer and the Frontier early-access clusters, Spock and Crusher. Our tests demonstrate both portability across different clusters and very good scaling efficiency, especially on large problems.

H2M: Exploiting Heterogeneous Shared Memory Architectures

Over the past decades, the performance gap between the memory subsystem and compute capabilities continued to spread. However, scientific applications and simulations show increasing demand for both memory speed and capacity. To tackle these demands, new technologies such as high-bandwidth memory (HBM) or non-volatile memory (NVM) emerged, which are usually combined with classical DRAM. The resulting architecture is a heterogeneous memory system in which no single memory is “best”. HBM is smaller but offers higher bandwidth than DRAM, whereas NVM provides larger capacity than DRAM at a reasonable cost and less energy consumption. Despite that, in several cases, DRAM still offers the best latency out of all three technologies.In order to use different kinds of memory, applications typically have to be modified to a great extent. Consequently, vendor-agnostic solutions are desirable. First, they should offer the functionality to identify kinds of memory, and second, to allocate data on it. In addition, because memory capacities may be limited, decisions about data placement regarding the different memory kinds have to be made. Finally, in making these decisions, changes over time in data that is accessed, and the actual access pattern, should be considered for initial data placement and be respected in data migration at run-time.In this paper, we introduce a new methodology that aims to provide portable tools and methods for managing data placement in systems with heterogeneous memory. Our approach allows programmers to provide traits (hints) for allocations that describe how data is used and accessed. Combined with characteristics of the platforms’ memory subsystem, these traits are exploited by heuristics to decide where to place data items. We also discuss methodologies for analyzing and identifying memory access characteristics of existing applications, and for recommending allocation traits.In our evaluation, we conduct experiments with several kernels and two proxy applications on Intel Knights Landing (HBM + DRAM) and Intel Ice Lake with Intel Optane DC Persistent Memory (DRAM + NVM) systems. We demonstrate that our methodology can bridge the performance gap between slow and fast memory by applying heuristics for initial data placement.

Effects of redox variability and early diagenesis on marine sedimentary Hg records

Volcanism is a dominant natural source of mercury (Hg) to the atmosphere, biosphere, ocean and sediments. In recent years, sedimentary Hg contents have emerged as a tool to reconstruct volcanic activity, and particularly activity of (subaerially emplaced) large igneous provinces in geological deep time. More specifically, Hg has shown potential as a useful proxy to illuminate the previously elusive impact of such large-scale volcanism on marine and terrestrial paleo-environments. While Hg is now widely applied as volcanism tracer, non-volcanic factors controlling sedimentary Hg content are generally not well constrained. Part of this uncertainty stems from our inability to directly observe a natural unperturbed “steady-state” environment as a baseline, as the modern Hg cycle is heavily influenced by anthropogenic activity. Here we focus on the effects of ambient redox conditions in the water column and shallow sediments (early diagenesis), quantify their influence on the geological Hg record and thereby contribute to constraining their potential impact on the use of Hg as a proxy for deep-time volcanic activity. Constraining these factors is of critical importance for the application of Hg as such a proxy. Many periods in the geological past for which records have been generated, such as the Mesozoic Oceanic Anoxic Events, are marked by a variety of high-amplitude environmental perturbations, including widespread deoxygenation and deposition of organic-rich sediments. We estimate the impact of redox changes and early diagenesis on the geological Hg record using a suite of (sub)recent–Pleistocene and Upper Cretaceous sediments representing oxic to euxinic marine conditions. Our sample set includes a transect through an oxygen minimum zone and cores that record transient shifts in oxygenation state, as well as post-depositional effects – all unrelated to volcanism, to the best of our knowledge. We find substantial alterations to the Hg record and the total organic carbon and total sulfur content, which are typically assumed to be the most common carrier phases of Hg in marine sediments. Moreover, these biases can lead to signal alteration on a par with those interpreted to result from volcanic activity. Geochemical modifications are ubiquitous and their potential magnitude implies that the factors leading to biases in the geological record warrant careful consideration before interpretation. Factors of particular concern to proxy application are (1) the disproportionate loss of organic carbon and sulfur relative to Hg during oxidation that strongly modulates normalized Hg records, (2) the evasion of Hg in anoxic and mildly euxinic sediments and (3) sharp focusing of Hg during post-depositional oxidation of organic matter. We suggest that paired analyses of additional redox-sensitive trace elements such as molybdenum, and organic-matter characteristics, particularly the type of organic matter, could provide first-order constraints on the role that redox and diagenetic changes played in shaping the Hg record as part of checking the attribution of enrichments to volcanic activity.

Design of a portable implementation of partitioned point‐to‐point communication primitives

AbstractThe Message Passing Interface (MPI) has been the dominant message passing solution for scientific computing for decades. MPI point‐to‐point communications are highly efficient mechanisms for process‐to‐process communication. However, MPI performance when processes utilize multiple threads is slowed by concurrency protections in the MPI library. MPI's current thread level interface imposes these overheads throughout the library when thread safety is needed. While much work has been done to reduce multithreading overheads in MPI, a solution is needed that reduces the number of messages exchanged in a threaded environment. Partitioned communication is included in the MPI 4.0 standard as an alternative that addresses the challenges of multithreaded communication in MPI today. Partitioned communication reduces overall message volume by creating a buffer‐sharing mechanism between threads such that they can indicate when portions of a communication buffer are available to be sent. Separation of the control and data planes in MPI is enabled by allowing persistent initialization and single occurrence message buffer matching from the indication that the data is ready to be sent. This enables the usage of underlying hardware primitives like triggered operations, where commands (destination, size, etc.) can be set up prior to data buffer readiness and readiness triggered with a simple doorbell/counter later. This approach is useful for future development of MPI operations in environments where traditional networking commands can have performance challenges, like accelerators (GPUs, FPGAs). In this paper, we detail the design and implementation of a layered library (built on top of MPI‐3.1) and an integrated Open MPI solution that supports the new, MPI‐4.0 partitioned communication feature set. The library will enable applications to use currently released MPI implementations and older legacy libraries to provide partitioned communication support while also enabling further exploration of this new communication model in new applications and use cases. We will compare the designs of the library and native Open MPI support, provide performance results and comparisons between the two approaches, and lessons learned from the implementation of partitioned communication in both library and native forms. We find that the native implementation and library have similar performance with a percentage difference under 0.94% in microbenchmarks and performance within 5% for a partitioned communication enabled proxy application.

DNPC: A Dynamic Node-Level Power Capping Library for Scientific Applications

As the race to exa-scale computing accerlerates, power consumption continues to be a critical challenge. While several technologies are available for power management, balancing energy efficiency and application performance during execution remains an important topic of research. In this study, we develop an open-source library called DNPC for dynamically controlling an application's package-level poewr consumption during execution. Given a specific sperformance degradation threshold, DNPC aims to minimize power consumption by following the application's power profile and adjusting the power cap accordingly. Further, DNPC is able to estimate the online performance of an application under capping, and predict its estimed performance degradation relative to an uncapped run. In this paper, we present an overview of the library, followed by a case study to illustrate the use of DNPC with an Exascale Computing Project proxy application on a production supercomputer at Argonne National Laboratory.

How Partial Anonymity May Reduce Students’ Anxiety During Remote Active Learning─A Case Study Using Clubhouse

Active learning, a common practice in higher education, has been shown to promote higher order thinking and skills. Class discussions have been chosen to be the medium to incorporate active learning in schools‘ curriculum. However, the rate of class participation could be low for certain courses. Literature has shown that the fear of negative evaluation from peers is the most common reason as to why students choose not to partake in class discussions. Anonymity via clickers or applications such as Kahoot! has shown to be useful in reducing students’ anxiety and increasing class participations. However, this is not a viable method to employ if vocal discussion is required for the course. Here, partial anonymity (voice only), Speak Your Mind, was applied into an environmental chemistry course with 20 students in the National University of Singapore (NUS) to study its correlation with students‘ anxiety and class participation. Participants survey results suggested that a reason for not participating in class discussions was the fear of being judged by their peers. Remote learning was conducted due to COVID-19 and partial anonymity was obtained by a proxy application: Clubhouse. This application allowed students to partake in a podium discussion while maintaining psychological safety via partial anonymity. Participants survey responses indicated that partial anonymity reduced their anxiety (Cohen’s d = 0.58) and slightly increased their self-reported class participation rate (Cohen’s d = 0.21); it was noted that partial anonymity did not have much effect on their fears of being judged if they provided the wrong answer (Cohen’s d = 0.11).

Near-Lossless MPI Tracing and Proxy Application Autogeneration

Traces of MPI communications are used by many performance analysis and visualization tools. Storing exhaustive traces of large-scale MPI applications is infeasible, however, because of their large volume. Aggregated or lossy MPI traces are smaller but provide much less information. In this paper we present Pilgrim, a near-lossless MPI tracing tool that, by using sophisticated compression techniques, generates small trace files at large scales and incurs only moderate overheads. We perform comprehensive studies of various compression techniques used for storing timestamps associated with each call. This timing information is essential for analysis purposes such as skews study. To demonstrate the usefulness of the detailed information stored by Pilgrim, we present a proxy application generator that can generate proxy apps that preserve original communication patterns from the Pilgrim traces.

Accelerating communication for parallel programming models on GPU systems

As an increasing number of leadership-class systems embrace GPU accelerators in the race towards exascale, efficient communication of GPU data is becoming one of the most critical components of high-performance computing. For developers of parallel programming models, implementing support for GPU-aware communication using native APIs for GPUs such as CUDA can be a daunting task as it requires considerable effort with little guarantee of performance. In this work, we demonstrate the capability of the Unified Communication X (UCX) framework to compose a GPU-aware communication layer that serves multiple parallel programming models of the Charm++ ecosystem: Charm++, Adaptive MPI (AMPI), and Charm4py. We demonstrate the performance impact of our designs with microbenchmarks adapted from the OSU benchmark suite, obtaining improvements in latency of up to 10.1x in Charm++, 11.7x in AMPI, and 17.4x in Charm4py. We also observe increases in bandwidth of up to 10.1x in Charm++, 10x in AMPI, and 10.5x in Charm4py. We show the potential impact of our designs on real-world applications by evaluating a proxy application for the Jacobi iterative method, improving the communication performance by up to 12.4x in Charm++, 12.8x in AMPI, and 19.7x in Charm4py.

Physiological control on carbon isotope fractionation in marine phytoplankton

Abstract. One of the great challenges in biogeochemical research over the past half a century has been to quantify and understand the mechanisms underlying stable carbon isotope fractionation (εp) in phytoplankton in response to changing CO2 concentrations. This interest is partly grounded in the use of fossil photosynthetic organism remains as a proxy for past atmospheric CO2 levels. Phytoplankton organic carbon is depleted in 13C compared to its source because of kinetic fractionation by the enzyme RubisCO during photosynthetic carbon fixation, as well as through physiological pathways upstream of RubisCO. Moreover, other factors such as nutrient limitation, variations in light regime as well as phytoplankton culturing systems and inorganic carbon manipulation approaches may confound the influence of aquatic CO2 concentrations [CO2] on εp. Here, based on experimental data compiled from the literature, we assess which underlying physiological processes cause the observed differences in εp for various phytoplankton groups in response to C-demand/C-supply, i.e., particulate organic carbon (POC) production / [CO2]) and test potential confounding factors. Culturing approaches and methods of carbonate chemistry manipulation were found to best explain the differences in εp between studies, although day length was an important predictor for εp in haptophytes. Extrapolating results from culturing experiments to natural environments and for proxy applications therefore require caution, and it should be carefully considered whether culture methods and experimental conditions are representative of natural environments.

Preparing for the Future—Rethinking Proxy Applications

A considerable amount of research and engineering went into designing proxy applications, which represent common high-performance computing (HPC) workloads, to co-design and evaluate the current generation of supercomputers, e.g., RIKEN’s supercomputer Fugaku, ANL’s Aurora, or ORNL’s Frontier. This process was necessary to standardize the procurement while avoiding duplicated effort at each HPC center to develop their own benchmarks. Unfortunately, proxy applications force HPC centers and providers (vendors) into an undesirable state of rigidity, in contrast to the fast-moving trends of current technology and future heterogeneity. To accommodate an extremely heterogeneous future, we have to reconsider how to co-design supercomputers during the next decade, and avoid repeating past mistakes.