Compile Time Research Articles

An Efficient Algorithm for Astrochemical Systems Using Stoichiometry Matrices

Astrochemical simulations are a powerful tool for revealing chemical evolution in the interstellar medium. Astrochemical calculations require efficient processing of large matrices for the chemical networks. The large chemical reaction networks often present bottlenecks for computation because of time derivatives of chemical abundances. We propose an efficient algorithm using a stoichiometry matrix approach in which this time-consuming part is expressed as a loop, unlike the algorithm used in previous studies. Since stoichiometry matrices are sparse in general, the performances of simulations with our algorithm depend on which sparse-matrix storage format is used. We conducted a performance comparison experiment using the common storage formats, including the coordinate format, the compressed column storage format, the compressed row storage (CRS) format, and the sliced ELLPACK format. Experimental results showed that the simulations with the CRS format are the most suitable for astrochemical simulations and about a factor of 2 faster than those with the algorithm used in previous studies. In addition, our algorithm significantly reduces not only the computation time but also the compilation time. We also explore the beneficial effects of parallelization and sparse-matrix reordering in these algorithms.

Statistical analysis of lexemes generated in ‘C’ programming using fuzzy automation

The present work is an effort to support the typographical errors of keywords that are not supported by existing compilers and integrated development environment(IDE) in ’C’ language. The fuzzy automata modelling approximate string matching is proposed for error handling during lexical analysis. By introducing fuzziness to lexemes the typographical errors can be rectified at the time of compilation and flexibility of lexical analyser can be greatly improved. The recognition of fuzzy tokens during lexical analysis is described in order to correct errors caused by sticking key, deletion, typing and swapping key in keywords during C programming. Algorithms and pseudo code are being developed to measure the degree of membership of crisp and fuzzy lexemes. Accuracy is tested and examined once the fuzzy lexemes are trained using a neural network. The proposed method is an add on feature that can be incorporated in existing compilers and IDEs to increase their flexibility.

IMPLEMENTASI METODE SERVQUAL DALAM MENGUKUR TINGKAT KEPUASAN PASIEN TERHADAP KUALITAS PELAYANAN RAWAT JALAN Studi Kasus: Puskesmas Maubesi

The service quality of a Community Health Center does have a great impact on the patients' satisfaction. The Maubesi Community Health Center is one of the health services centers in Maubesi Village which serves a number of six surrounding villages. Given that many patients are making use of this health center, the service quality of the medical staff is of course quite essential. Patients' satisfaction is the primary factor which becomes the image benchmark of a community health center. Up to now, patients' satisfaction is measured in the conventional way, namely by writing on papers. The use òf paper certainly makes it hectic when the time for data compilation and analysis processing is due. To overcome this problem, the researcher developed a website-based information system (SISKESAN) to measure patients' satisfaction level at Maubesi Community Health Center. SISKESAN was developed based on website, to make it easy whenever and wherever the data is to be accessed. The system employs Service Quality Method (Servqual) to measure patients' satisfaction level using hope and fact values. To set up this system the researcher made use of Waterfall Method known as measurable gradual stages. By implementing SISKESAN at Maubesi Community Health Center, the patients' satisfaction level becomes easier in measurement. Patients can easily access the website and fill in the questionnaires prepared in the system. The result can be straightly seen by the staff in charge and it will become the benchmark for future policy making.

A Design Flow for Scheduling Spiking Deep Convolutional Neural Networks on Heterogeneous Neuromorphic System-on-Chip

Neuromorphic systems-on-chip (NSoCs) integrate CPU cores and neuromorphic hardware accelerators on the same chip. These platforms can execute spiking deep convolutional neural networks (SDCNNs) with a low energy footprint. Modern NSoCs are heterogeneous in terms of their computing, communication, and storage resources. This makes scheduling SDCNN operations a combinatorial problem of exploring an exponentially-large state space in determining mapping, ordering, and timing of operations to achieve a target hardware performance, e.g., throughput. We propose a systematic design flow to schedule SDCNNs on an NSoC. Our scheduler, called SMART ( S DCNN MA pping, Orde R ing, and T iming), branches the combinatorial optimization problem into computationally-relaxed sub-problems that generate fast solutions without significantly compromising the solution quality. SMART improves performance by efficiently incorporating the heterogeneity in computing, communication, and storage resources. SMART operates in four steps. First, it creates a self-timed execution schedule to map operations to compute resources, maximizing throughput. Second, it uses an optimization strategy to distribute activation and synaptic weights to storage resources, minimizing data communication-related overhead. Third, it constructs an inter-processor communication (IPC) graph with a transaction order for its communication actors. This transaction order is created using a transaction partial order algorithm, which minimizes contention on the shared communication resources. Finally, it schedules this IPC graph to hardware by overlapping communication with the computation, and leveraging operation, pipeline, and batch parallelism. We evaluate SMART using 10 representative image, object, and language-based SDCNNs. Results show that SMART increases throughput by an average 23%, compared to a state-of-the-art scheduler. SMART is implemented entirely in software as a compiler extension. It doesn’t require any change in a neuromorphic hardware or its interface to CPUs. It improves throughput with only a marginal increase in the compilation time. SMART is released under the open-source MIT licensing at https://github.com/drexel-DISCO/SMART to foster future research.

HAOTuner: A Hardware Adaptive Operator Auto-Tuner for Dynamic Shape Tensor Compilers

Deep learning compilers with auto-tuners have the ability to generate high-performance programs, particularly tensor programs on accelerators. However, the performance of these tensor programs is shape-sensitive and hardware resource-sensitive. When the tensor shape is only known at runtime instead of compile time, auto-tuners must tune the tensor programs for every possible shape, leading to significant time and cost overhead. Additionally, if a tensor program tuned for one device is deployed on a different device, the performance may not be as optimal as before. To address these challenges, we propose HAOTuner, a hardware-adaptive deep learning operator auto-tuner specifically designed for dynamic shape tensors. We leverage the concept of micro-kernels as the unit of task allocation and have observed that the size of the micro-kernel greatly impacts performance. In HAOTuner, we determine the size of micro-kernels based not only on the tensor shapes but also on the available hardware resources. Specifically, we present an algorithm to select hardware-friendly micro-kernels as candidates, reducing the tuning time. We also design a cost model that is sensitive to hardware resources to support various hardware architectures. Furthermore, we provide a model transfer solution to enable fast deployment of the cost model on different hardware platforms. We evaluate HAOTuner on six different types of GPUs. The experiments demonstrate that HAOTuner surpasses the state-of-the-art dynamic shape tensor auto-tuner in terms of running time by an average of 26% and tuning time by 25%. Moreover, HAOTuner outperforms the state-of-the-art compiler with padding in terms of running time by an average of 39% and tuning time by 6×.

Compiling Structured Tensor Algebra

Tensor algebra is essential for data-intensive workloads in various computational domains. Computational scientists face a trade-off between the specialization degree provided by dense tensor algebra and the algorithmic efficiency that leverages the structure provided by sparse tensors. This paper presents StructTensor, a framework that symbolically computes structure at compilation time. This is enabled by Structured Tensor Unified Representation (STUR), an intermediate language that can capture tensor computations as well as their sparsity and redundancy structures. Through a mathematical view of lossless tensor computations, we show that our symbolic structure computation and the related optimizations are sound. Finally, for different tensor computation workloads and structures, we experimentally show how capturing the symbolic structure can result in outperforming state-of-the-art frameworks for both dense and sparse tensor algebra.

Prevalence of sports supplements consumption and its association with food choices among female elite football players

ObjectivesThe present study aimed to 1) investigate the consumption of Sports Supplements (SSs) among female elite football players, 2) evaluate the influence of age on SS consumption, and 3) determine the relationship between the consumption of SSs and dietary choices among elite football players. MethodsA total of 126 female football players of Primera Iberdrola and Reto Iberdrola who participated in this descriptive, observational, and cross-sectional study completed a self-administered questionnaire on SSs and the Athletes’ Food Choices Questionnaire. ResultsOverall, 84.1% of participants consumed supplements, mainly for improved sports performance (68.3%) and health (34.1%). The main sources of purchase were the Internet (34.9%) and specialized shops (23.8%), and players were commonly advised by a dietitian-nutritionist to use SSs (56.3%). The SSs most often consumed included whey protein (30.2%), sports drinks (28.6%), creatine monohydrate (28.6%), sports bars (27.8%), and caffeine (27.8%). Older players consumed more supplements at the time of data compilation. Players predominantly acquired these supplements by using the Internet and reported benefits from their use (all P ≤ 0.036). Additionally, players who consumed SSs conveyed more concern about their food choices. ConclusionsA high prevalence of female football players consumed SSs, particularly SSs supported by robust scientific evidence. Older players had higher supplement consumption rates. The use of SSs was related to food choices through nutritional characteristics of foods, knowledge about health and nutrition, weight control with the help of food, and the performance benefits players could acquire.

Design of efficient Programming Language with Lexer using ’$’-prefixed identifier

An identifier which starts with ‘$‘is known as ‘$‘-prefixed identifier and this type of identifiers are used in our research paper to improve the lexical analysis phase. This paper talks about a new programming language with ‘$‘-prefixed identifier that features a novel approach for optimizing the lexer for efficient lexical analysis which can be applied to any existing language. This approach is used to classify identifiers and keywords using ‘$‘-prefixed variables, which significantly reduces the time taken and number of iterations required during the tokenization process, leading to improved overall performance. This type of language structure allows for fast lookup and matching of tokens. We conducted a series of experiments to evaluate the performance of our lexer and compared it with a regular lexer. Our results show that our approach leads to significant improvements in time complexity and number of iterations for identifying whether the token is an identifier or a keyword, resulting in faster compilation times and improved overall performance. Our language has reduced the amount of time taken by 7-10% and 45-50% in terms of iterations. Our language and lexer represent a significant step forward in the design and implementation of high-performance programming languages by reducing the number of iterations and time taken to identify whether a token is a keyword or an identifier.

RapidStream 2.0: Automated Parallel Implementation of Latency–Insensitive FPGA Designs Through Partial Reconfiguration

Field-programmable gate arrays (FPGAs) require a much longer compilation cycle than conventional computing platforms such as CPUs. In this article, we shorten the overall compilation time by co-optimizing the HLS compilation (C-to-RTL) and the back-end physical implementation (RTL-to-bitstream). We propose a split compilation approach based on the pipelining flexibility at the HLS level, which allows us to partition designs for parallel placement and routing. We outline a number of technical challenges and address them by breaking the conventional boundaries between different stages of the traditional FPGA tool flow and reorganizing them to achieve a fast end-to-end compilation. Our research produces RapidStream, a parallelized and physical-integrated compilation framework that takes in a latency-insensitive program in C/C++ and generates a fully placed and routed implementation. We present two approaches. The first approach (RapidStream 1.0) resolves inter-partition routing conflicts at the end when separate partitions are stitched together. When tested on the Xilinx U250 FPGA with a set of realistic HLS designs, RapidStream achieves a 5 to 7× reduction in compile time and up to 1.3× increase in frequency when compared with a commercial off-the-shelf toolchain. In addition, we provide preliminary results using a customized open-source router to reduce the compile time up to an order of magnitude in cases with lower performance requirements. The second approach (RapidStream 2.0) prevents routing conflicts using virtual pins. Testing on Xilinx U280 FPGA, we observed 5 to 7× compile time reduction and 1.3× frequency increase.

Parallelising Control Flow in Dynamic-scheduling High-level Synthesis

Recently, there is a trend to use high-level synthesis (HLS) tools to generate dynamically scheduled hardware. The generated hardware is made up of components connected using handshake signals. These handshake signals schedule the components at runtime when inputs become available. Such approaches promise superior performance on “irregular” source programs, such as those whose control flow depends on input data. This is at the cost of additional area. Current dynamic scheduling techniques are well able to exploit parallelism among instructions within each basic block (BB) of the source program, but parallelism between BBs is under-explored, due to the complexity in runtime control flows and memory dependencies. Existing tools allow some of the operations of different BBs to overlap, but to simplify the analysis required at compile time they require the BBs to start in strict program order, thus limiting the achievable parallelism and overall performance. We formulate a general dependency model suitable for comparing the ability of different dynamic scheduling approaches to extract maximal parallelism at runtime. Using this model, we explore a variety of mechanisms for runtime scheduling, incorporating and generalising existing approaches. In particular, we precisely identify the restrictions in existing scheduling implementation and define possible optimisation solutions. We identify two particularly promising examples where the compile-time overhead is small and the area overhead is minimal and yet we are able to significantly speed up execution time: (1) parallelising consecutive independent loops; and (2) parallelising independent inner-loop instances in a nested loop as individual threads. Using benchmark sets from related works, we compare our proposed toolflow against a state-of-the-art dynamic-scheduling HLS tool called Dynamatic. Our results show that, on average, our toolflow yields a 4× speedup from (1) and a 2.9× speedup from (2), with a negligible area overhead. This increases to a 14.3× average speedup when combining (1) and (2).

Machine learning models for chronic kidney disease diagnosis and prediction

Background and objectiveChronic kidney disease is a severe health problem that affects people all over the world, particularly in South Asia. Therefore, proper diagnosis and treatment are required as early as possible. The main goal of this study is to detect the presence or absence of CKD in the human body utilizing a variety of features grasped from a few medical tests. MethodsThis paper has focused on eight ensemble learning methods for diagnosing CKD on the UCI machine learning datasets. The datasets have been fixed by imputing the missing values using the MICE imputation method and handling the imbalance properties using the borderline SVMSMOTE method to improve the performance of classifiers. Moreover, recursive feature elimination and the boruta method have been used to find the most significant features and reduce the compilation time, while the hyperparameter tuning technique was used to raise the performance of classifiers and get optimal solutions. ResultsTaking the most significant feature into consideration, RFE outperformed boruta and selected only 50% of the total features. Moreover, various performance matrices are used to find the best competent classifiers for detecting CKD. LightGBM outperformed state-of-the-art and other ensemble methods with the lowest compilation time and highest accuracy. Based on experimental findings, the proposed method achieved the highest average of 99.75% accuracy, 99.40% precision, 99.41% recall, 99.61% F-measure and 99.57% AUC-ROC. Moreover, our proposed method rises the average detection rate by 5.64%, 1%, 2.04%, 8.63%,1.99%, 2.84%, 2.42% and 4.76%, respectively, in comparison with different approaches performing on the same dataset. ConclusionExperiments show that our suggested method can identify CKD more precisely than the most recent methods.

An efficient method for predicting PV modules performance based on the two-diode model and adaptable to the single-diode model

This paper introduces a simplified and accurate modeling approach to model photovoltaic modules. The current method uses a hybrid technique combining numerical and analytical approaches to predict the current-voltage characteristics of PV generators by calculating the parameters of the double-diode model. First, to reduce the complexity of the current equation and minimize the calculation time, some simplifications reducing the research space from seven to only five unknowns are taken into consideration. Second, three of these five parameters are extracted using simple analytical equations derived from the output current's equation based on the available values of the typical points on the manufacturer's datasheet. Then, the remaining two parameters are extracted using a combination of numerical approach and slope adjustment technique at the short-circuit point. The technique is adapted to the single-diode model in order to study the effect of the second diode on the simulating accuracy. The effectiveness is tested over six PV modules matching different PV technologies, and the results are compared with a great number of modeling methods based on different approaches, and the adjustment technique has shown the highest accuracy levels (RMSE values less than 0.045A) at the least computational times (compilation time less than 0.084 s).

A portable C++ library for memory and compute abstraction on multi‐core CPUs and GPUs

AbstractWe present a C++ library for transparent memory and compute abstraction across CPU and GPU architectures. Our library combines generic data structures like vectors, multi‐dimensional arrays, maps, graphs, and sparse grids with basic generic algorithms like arbitrary‐dimensional convolutions, copying, merging, sorting, prefix sum, reductions, neighbor search, and filtering. The memory layout of the data structures is adapted at compile time using C++ tuples with optional memory double‐mapping between host and device and the capability of using memory managed by external libraries with no data copying. We combine this transparent memory layout with generic thread‐parallel algorithms under two alternative common interfaces: a CUDA‐like kernel interface and a lambda‐function interface. We quantify the memory and compute performance and portability of our implementation using micro‐benchmarks, showing that the abstractions introduce negligible performance overhead, and we compare performance against the current state of the art in a real‐world scientific application from computational fluid mechanics.

Underdetermined direction of arrival estimation for multiple input and multiple outputs sparse channel based on Bayesian learning framework

Direction of arrival (DOA) estimation for a sparse channel has attracted serious attention recently. Better signal analysis and denoising achieve accuracy in DOA determination. This paper proposes an underdetermined DOA estimation for multiple input and multiple outputs (MIMO) sparse channels. A novel multi-kernel-based non-negative sparse Bayesian learning (MK NNSBL) framework is implemented using the multiplied form of basis vector within the manifold matrix for a defined grid. Meanwhile, virtual antenna locations are reconfigured by exploiting the conventional cuckoo search algorithm (CCSA) for the fine reception of incoming signals on a nonuniform linear array (NULA). The simulated results reveal that the novel approach outperforms in its optimal root mean square error (RMSE) for various signal-to-noise ratio (SNR) limits and the compilation time. The convergence comparative graph indicates the improved performance in the proposed framework over existing algorithms.

TECHNOLOGY FOR IDENTIFYING AND FORMING POSSIBLE RELATIONSHIPS BETWEEN USE CASES IN THE PROCESS OF THE INFORMATION SYSTEM DESIGN

Context. Use cases are widely used as a means of formulating requirements in the development of information systems. All subsequent design stages depend on the quality of their presentation. Structuring use cases can significantly increase their understanding and maintenance in the face of changing requirements..
 Objective. Flexible technologies involve working in small teams. The existing communication between teams is not sufficient to highlight sub use cases at the project level. There is a need for automated analysis of the corpus of all use cases.
 Method. A mathematical model of a use case which makes it possible to define the criteria for comparing scenarios and eliminate the redundancy of descriptions is proposed. A four-step method for restructuring use cases has been developed. At the first stage, use cases are presented in a formalized form. At the second, they are stored in the repository, which ensures their quick search and placement. At the third stage, procedures of scenario comparison are performed. Scenario similarity criteria are proposed. At the fourth stage, the formation of subordinate use cases is carried out, their texts are coordinated with all interested teams, and the use cases that cause subordinate use cases are corrected.
 Results. Experiments providing the formalized compilation of use cases by several development teams followed by automated restructuring were carried out to test the proposed solutions. As a result, new subordinate use cases were correctly identified and the scope of use of previously formed ones was expanded. There was a significant reduction in the time for restructuring.
 Conclusions. The proposed method of restructuring use cases improves the clarity and consistency of requirements, the possibility of their adjustment and maintenance, and reduces the compilation time. The method can be used in the design of any information system, where the requirements are presented in the form of use cases.

魏志倭人傳과 古代日本資料에서의 表記字에 관한 小考

The purpose of this paper is to more clearly understand the phonology of Wajinden through comparison with ancient Korean and Japanese materials. The subjects of the study were 蘇 and 奴, which have different phonetic values from ancient Japanese materials among the phonetic transcriptions shown in the Wajinden. As for the research method, the phonetic values of 蘇 and 奴 shown in the Wajinden were examined through the estimated Old Chinese phonology, the Fanqie(反切) in a Rhyme Dictionary(韻書), the phonetic values by scholars, and the examples of ancient Korean and Japanese materials.
 As a conclusion of the study, it was found that 蘇 and 奴 shown in the Wajinden are the markers for the syllables of [sa] and [na]. In Wajinden, 蘇 is a notation for the syllable of [sa さ], and all of them were used as national names. In ancient Japanese sources, 蘇 represents the sound of [so ソ].
 As a result of the study of 蘇, first, the phonetic value of the estimated Old Chinese phonology was divided into [sa] and [so], and the phonetic value estimated through the Fanqie(反切) in a Rhyme Dictionary(韻書) was [so]. However, since the sound values shown in a Rhyme Dictionary(韻書) is in Middle-Old Chinese phonology, the authenticity is questionable. second, through the place names and names of people in Goguryeo and Baekje, the corresponding relationship of 蘇：斯：史：泗：所 is established, and it is read as [sa]. Considering these points, it can be said that 蘇 represents the sound value of [sa].
 In addition, there are 16 examples of the use of 奴, which can be found in Wajinden, and all refer to proper nouns such as the name of an official position or the name of a country. Among them, only one usage shows the sound value of [nu ぬ], and the remaining 15 usage examples show the sound value of [na な]. In ancient Japanese sources, the sound values of [nuヌ, noノ, doド] are indicated.
 As a result of consideration, first, as with 蘇, the estimated upper ancient sound of 奴 is divided into [nu] and [na], and the phonetic value estimated through the Fanqie(反切) in a Rhyme Dictionary(韻書) and Ōno's indication was [no], but the time of compilation of a Rhyme Dictionary(韻書) is Middle-Old Chinese phonology. second, the corresponding relationship between 奴=那=弩=內=儒 and 奴=那 can be confirmed through the geographical names and five tribes in the SamgukSagi. third, it was confirmed that 奴 represents the sound value of [na] in the name notation of Nihonsyoki. Through this fact, the sound value of 奴 can be read as [na].

The Onset of Musculoskeletal Pain in the COVID-19 Era: A Survey of Physiotherapy Students in Sicily.

Online teaching has resulted in university students adopting a sedentary lifestyle. Prolonged sitting and reduced physical activity due to pandemic restrictions have led to musculoskeletal pain in various body areas, significantly impacting students' quality of life. This study aims to investigate the effects of remote learning on Sicilian physiotherapy students during the COVID-19 pandemic, specifically focusing on the occurrence of musculoskeletal pain. An observational study was conducted using an online survey administered through Google Forms. The survey consisted of 26 multiple-choice questions and was distributed to students enrolled in physiotherapy programs at the universities of Catania, Messina, and Palermo. Participants were contacted via social channels or email, and data collection spanned 5 weeks. The collected data were analyzed using R software. A total of 128 questionnaires were collected. At the time of compilation, most respondents (n = 103/201, 51.2%) were enrolled in the third year of the course of study in physiotherapy at the universities of Catania, Messina, and Palermo. Their ages ranged between 22 and 25 years (43.3%), and most were female (n = 104/201, 51.7%). More than half of the students (51.6%) reported dedicating 15-22 h per week to distance learning for a duration of 6-12 months (50%). Regarding study location, most students preferred studying at a desk (82.8%), and slightly over half (57.8%) adopted a backrest while studying remotely. Analysis of the students' posture during study hours revealed common positions, including tilting the head forward by more than 20 degrees (47.8%), leaning the trunk forward by more than 20 degrees (71.9%), hunching both shoulders forward (57.0%), wrists positioned above the level of the elbows (46.1%), thighs pointing upwards (41.4%), and one or both feet in a downward or dorsiflexed position (69.5%). In conclusion the questionnaire responses indicate that the lifestyle of university students, influenced by online teaching, has deteriorated, leading to musculoskeletal pain, including myofascial pain. These results are primarily influenced by the adopted posture and the duration of time spent in these positions. Additionally, research is needed to identify the most effective therapeutic approaches for managing musculoskeletal pain.

Side-channel Elimination via Partial Control-flow Linearization

Partial control-flow linearization is a code transformation conceived to maximize work performed in vectorized programs. In this article, we find a new service for it. We show that partial control-flow linearization protects programs against timing attacks. This transformation is sound: Given an instance of its public inputs, the partially linearized program always runs the same sequence of instructions, regardless of secret inputs. Incidentally, if the original program is publicly safe, then accesses to the data cache will be data oblivious in the transformed code. The transformation is optimal: Every branch that depends on some secret data is linearized; no branch that depends on only public data is linearized. Therefore, the transformation preserves loops that depend exclusively on public information. If every branch that leaves a loop depends on secret data, then the transformed program will not terminate. Our transformation extends previous work in non-trivial ways. It handles C constructs such as “goto,” “break,” “switch,” and “continue,” which are absent in the FaCT domain-specific language (2018). Like Constantine (2021), our transformation ensures operation invariance but without requiring profiling information. Additionally, in contrast to SC-Eliminator (2018) and Lif (2021), it handles programs containing loops whose trip count is not known at compilation time.

Rust vs C++, a Battle of Speed and Efficiency

This study compares the performance of two excellent options for system-level development, the programming languages C++ and Rust. Through a series of tests and experiments using socket servers and various algorithms, this experiment analyses the speed and efficiency of code written in each language by looking at variables like memory management, and compilation times. The findings reveal that Rust has a number of advantages over C++, including quicker compilation times, improved memory safety, and in many situations equivalent or better performance. C++ still performs exceptionally well in several fields, nevertheless, such as low-level hardware programming and backward compatibility. Overall, the results indicate that Rust is a strong candidate for systems programming jobs, especially for new projects or those requiring a high level of performance and security

A VM-Agnostic and Backwards Compatible Protected Modifier for Dynamically-Typed Languages

In object-oriented languages, method visibility modifiers hold a key role in separating internal methods from the public API. Protected visibility modifiers offer a way to hide methods from external objects while authorizing internal use and overriding in subclasses. While present in main statically-typed languages, visibility modifiers are not as common or mature in dynamically-typed languages. In this article, we present ProtDyn, a self-send-based visibility model calculated at compile time for dynamically-typed languages relying on name-mangling and syntactic differentiation of self vs non self sends. We present #Pharo, a ProtDyn implementation of this model that is backwards compatible with existing programs, and its port to Python. Using these implementations we study the performance impact of ProtDyn on the method lookup, in the presence of global lookup caches and polymorphic inline caches. We show that our name mangling and double method registration technique has a very low impact on performance and keeps the benefits from the global lookup cache and polymorphic inline cache. We also show that the memory overhead on a real use case is between 2% and 13% in the worst-case scenario. Protected modifier semantics enforces encapsulation such as private but allow developers to still extend the class in subclasses. ProtDyn offers a VM-agnostic and backwards-compatible design to introduce protected semantics in dynamically-typed languages.