Motivation: Program visualizations are widely used to support novice programmers, yet students often ignore or resist well-designed visual scaffolds. Research on multiple external representations (MERs) suggests cognitive design principles for coordinating views, but says little about what determines whether learners actually engage with the representations available to them.

Spatial skills have been shown to correspond with performance in computer science courses, and individuals further along in their computing journey tend to score higher on spatial skills tests. There is also evidence that these skills can be trained, consequently mitigating gaps in computing performance across demographic groups. However, it is unclear if training spatial skills has a lasting effect throughout a student's computing degree program. To fill this gap, we present longitudinal analyses from two separate institutions that implemented spatial skills training in their entry-level computing course. We perform a survival analysis on the retention data after four to five years and demonstrate that students with higher spatial skills after training are retained at higher rates throughout a degree program than students with lower spatial skills, even when the students initially started with lower spatial skills. However, the patterns of retention of students with high initial spatial skills and those who improve their spatial skills after training differ in our two cases, indicating the potential impact of other factors, such as self-regulation, which are critical for academic success. These patterns underscore that spatial-skill gains, while important, are not a panacea for student success in computing. Complementary supports that bolster self-regulation and related competencies may be required to sustain long-term retention.

As computing education becomes increasingly globalized, many students learn computer science through a second language. Prior work has documented cognitive and performance challenges faced by non-native English-speaking (NNES) students, yet less is known about how language background shapes their motivational experiences and intentions to persist. Drawing on Expectancy-Value Theory, we analyzed matched pre- and post-term survey data from 374 students (198 NNES, 176 NES) in a CS1 course at a large North American university. We measured programming self-efficacy, implicit theories of intelligence, need for cognition, sense of belonging, motivation and learning strategies, and intentions to major in computing. NNES students began the course believing intelligence is fixed and had lower self-efficacy on language-dependent tasks, gaps that persisted throughout the term. However, despite lower confidence, NNES students were more willing to choose challenging assignments and consistently used more strategic learning approaches. While NNES and native English-speaking (NES) students reported similar overall belonging, NNES students experienced greater belonging uncertainty, specifically when encountering difficulties, and were more likely to want to fade into the background within the CS community. These findings reveal language background as a persistent motivational cost in introductory computing and underscore the need for instructional designs that explicitly support belonging, self-efficacy, and adaptive strategy use for linguistically diverse learners.

Motivation: Database courses focus on SQL querying (DQL) while treating schema definition (DDL) and data manipulation (DML) as side topics, even though real-world database work begins with understanding schema design and data updates. This misalignment leaves students underprepared for authentic data management practice. Method: We integrated scaffolded DDL and DML exercises as a core concept in a third-year data course across three offerings (2023-2025). Students completed structured weekly tasks in an LMS that provides immediate feedback and unlimited attempts, encouraging low-stakes, iterative practice. We analyzed student interaction data (number of attempts and performance) to examine learning patterns across DDL/DML and DQL. We analyzed 9,071 total exercise submissions from 669 students, examining both the number of LMS exercise attempts and assignment performance across DDL/DML and DQL. Results: Students required fewer attempts on DDL/DML tasks than on traditional DQL tasks, indicating strong receptiveness when these topics were properly scaffolded. Early performance on schema-definition tasks was moderately correlated with later SQL performance, suggesting that schema competence supports subsequent query learning. Implications: We encourage database educators to teach schema design and data manipulation as core topics to strengthen students' conceptual foundations, as our results suggest these skills are learnable with scaffolding and may support subsequent query learning.

Previous work on prerequisites in computing found that GPAs correlated more strongly with course grades than direct measures of prerequisite knowledge. This suggests that course grades encompass more than subject knowledge. This work was conducted in an advanced data structures (ADS) course at a single institution. To investigate the generalizability of these findings, we replicate the study in an upper-year machine learning course. We find that the pattern holds. These results underscore the need to distinguish course performance from demonstrated skill.

The Impostor Phenomenon (IP) and the Confidence Gap describe gendered differences in self-assessment and perceived competence, with well-documented implications for learning, retention, and performance. While each has been studied independently, little research has investigated how these phenomena interact -- particularly within computing education contexts. This study addresses the gap by examining the relationship between IP and confidence among university students in web software development courses in Northern Europe. Drawing on 392 survey responses, we analyze how IP and self-reported confidence vary with gender, degree level, and prior programming experience, and expanding prior work, how these constructs relate to one another. Our findings confirm previously reported patterns -- women report higher IP and lower confidence than men -- and provide new insights: confidence partially mediates the relationship between gender and IP, indicating that lower confidence may partially help explain gender disparities in IP. Furthermore, while programming experience predicts confidence, its relationship to IP is more nuanced. Experience has a small direct positive effect on IP that is largely offset by an indirect confidence-driven reduction. These findings suggest that impostor feelings are shaped more by self-perception than skill, and they highlight the need for interventions that focus on both skill development and psychological support.

This pilot study investigated students' perceptions of visual progress bar and on-demand hints in a CS1 course. Students valued both feedback systems, with more confident students preferring on-demand hints for direct problem-solving support. Progress bars were consistently perceived as beneficial across varying confidence levels.

Machine learning (ML) has become increasingly important for students, yet university-level ML courses are often perceived as challenging and time-intensive. This study explores the perceived challenges and motivations of students in a university ML course to inform curricular and teaching strategies. Through 5 surveys conducted in two instances of a 12-week introductory ML course, we examined students' engagement with both theoretical and practical aspects of ML. Results indicate that while students initially express strong interest in applying ML concepts, their reported interests can shift toward theoretical foundations. Challenges in both theory and practice are reported, including difficulties in mathematical notation and vectorization of gradient components, as well as model implementation. Students also discuss the time commitment required in a course with both theoretical and practical content. We recommend aligning course content with student motivations, providing targeted support for mathematical notation and vectorization, and balancing theoretical depth with practical application.

Self-explanation promotes active learning by having students articulate their conceptual understanding in their own words. This study investigates whether the timing of self-explanations (before vs. after solving an exercise) relates to performance in a flipped, second-year computer organization course. Although students who self-explained before the exercises achieved higher course marks, the difference was not significant. Still, these findings suggest that early self-explanation may better prepare students for problem-solving when learning new concepts.

Self-explanation deepens understanding by giving learners an opportunity to reflect on what they are learning in a structured way. However, many students struggle to engage in it effectively. We investigate whether large language models (LLMs) can scaffold self-explanations in a flipped computer organization course. In an A/B test, one group used a fixed prompt to compare their explanations with an expert's, while another engaged in an interactive dialogue with an LLM to identify gaps. Although the overall quality of the explanation did not differ significantly between conditions, some students (non-native English speakers and women) reported greater comfort and perceived value when using the LLM.

Creating a supportive community in introductory programming courses is vital to student success, yet forums meant to facilitate this can cause stress due to social comparison. According to social identity theory, students are more likely to engage and feel a sense of belonging when they perceive connections with their peers. This study investigates whether peer-modeled posts that simulate students exhibiting desirable engagement behavior can reduce feelings of isolation and foster social connection among students. We introduced curated posts modeling expected student behavior -- covering content, providing emotional support, and offering study tips -- into Q&A forums for two introductory computing courses. These posts were inserted using different student accounts. Surveys and forum data were analyzed to measure the impact on students' feelings of isolation. Students responded positively to the seeded posts, reporting a significant reduction in feelings of isolation. Notably, women reported feeling less isolated after seeing the posts more than men, and many students reported feeling relieved that other students had the same worries and concerns as them. Seeding peer-modeled posts can significantly reduce student isolation and foster a greater sense of belonging in competitive academic contexts. However, future work may explore alternative delivery mechanisms, such as instructor posts framed as ''questions from last year,'' to determine if they can achieve similar effects.

This research evaluates voice-based self-explanations as a pedagogical tool in preparation for lectures, assesses user preferences between voice and text, and derives design insights. We report two studies: Study 1, a quasi-experimental field study, with 247 participants divided into voice-based (N = 83), text-based (N = 81), and choice (N = 83) conditions. Study 2 uses semi-structured interviews (N = 16) to explore perceptions of the interaction paradigms in-depth. Results from the first study revealed a general preference for text, though voice users produced longer responses and more topic-related keywords. Over time, the preference for voice increased among students, from 10% to 46%, when given a choice. Study 2 suggested that factors like social presence contribute to hesitance toward voice-based explanations, with a cognitive load, self-confidence, and performance anxiety also influencing medium preferences. Our findings highlight design recommendations and demonstrate the potential of voice-based self-explanations in educational settings, indicating that mixed interfaces might better meet diverse needs.

Motivation: Q&A forums are a critical resource for supporting students in large educational environments, yet students often perceive these forums as stressful and report discomfort in participating visibly, especially in classes that are large and have students with varying levels of prior programming experience (PE). Method: We divided students in a CS1 Q&A forum into smaller, homogenous groups based on their PE. We use a mixed-methods approach to compare data from this experience to data from a setting where all students shared a single, large Q&A forum (a “mixed” setting). We quantitatively analyze measures of student engagement and use an open-ended qualitative approach to examine responses about student experience on the forums. This approach helps us identify the motivation behind student decisions to participate in visible or non-visible ways and to evaluate their alignment with theoretical frameworks. Results: In the mixed setting, students frequently use anonymity, with students without PE using anonymity more than students with PE and women using anonymity more than men. In contrast, in the homogenous groups, novices used anonymity less than novices in the mixed setting, while the students in higher-experience groups tended to use it more. We also observe a reduced anonymity usage among women in the homogenous experience groups, suggesting that PE plays a critical role in the observed gender disparities in forum participation. The qualitative analysis provides additional evidence that social status issues and confidence may explain these behavioral patterns. Conclusion: This study highlights the potential benefits and consequences of grouping students by experience. Homogenous PE groups foster increased student comfort and engagement within the Q&A forum for students with less experience, but students with more experience are exposed to more perceived status threats. We discuss how these results align with the theories we used to design the homogenous group setting. This exploration contributes to a deeper understanding of the underlying dynamics shaping student behavior in online learning communities. Educators and platform designers can use these lessons to more effectively create inclusive environments that accommodate diverse student needs and preferences.

In computing courses, instructor involvement and social comfort are vital for resilience and belonging. We examine engagement with instructor emails aimed at strengthening the connection with students. We sent weekly emails from instructors to first- and upper-year computing students. These emails included reminders for the assignments due each week. Half of the students received reminders embedded in an informal message that contained approachable wording and relevant current course events, while the rest received a list of precise deadlines. This text had no emotional engagement from the instructor. We collected and analyzed email access and link click rates, along with student survey responses about email preferences and engagement. We found that first-year students had lower email access and link click rates than upper-year students. While we did not find differences in first-year engagement based on the type of email, upper-year students appeared to be more engaged when receiving the intentionally informal version of the email. Understanding the message preferences of computing students can enhance instructor messaging and improve engagement. Strategies should be explored to boost first-year student engagement, while the higher engagement among upper-year students underscores the importance of instructor support in advanced courses.

Asking for help (help-seeking) is a recognized and effective problem-solving strategy. This study investigates students' interaction with on-demand hints (automated hints requested by students) and assesses their impact on learning progress. We conducted an A/B experiment in a third-year computer science database course, offering hints for selected SQL problems with different hint designs. We collected data on students' code submissions, grades, and hint requests, and we administered a survey to gather feedback and gauge student perception of the hints. Many students accessed hints immediately without attempting the problem first, often requesting multiple hints in quick succession. While students perceived the hints to be valuable, we did not detect an impact on student problem-solving. These insights could inform future studies on the possible impact of students' attitudes toward hints, and how different types of hints might impact uptake and perception of hints.

Prior research has shown that storytelling is an effective method for increasing comprehension of concepts. Students often find computational topics, such as data structures, to be difficult to grasp initially. To bridge this gap, we investigate whether the use of storytelling in pre-lecture videos increases students' retention and understanding. At a North American university, instructors randomly assigned students to two separate groups who watch different types of pre-lecture videos: one in a traditional format and the other where they teach a concept through storytelling. These videos were deployed as unlisted YouTube links embedded in students' quizzes. Using YouTube's Reporting API, we analyzed the audience retention data against elapsed time to compare audience retention between traditional and storytelling teaching methodologies. There were more storytelling videos with a higher average retention level, and the audience displayed less skipping behaviour than their traditional counterparts. In the future we will further analyze students' perceptions of storytelling videos to better understand higher audience retention and the effectiveness of learning through storytelling lectures.

This study explores links between attributes of computing students, such as prior programming experience (PE) and gender, with expectations for success and the perception of challenges. Using Expectancy-Value Theory (EVT), we investigate their major intentions and the impact of these factors post-CS1. Data was gathered using surveys at the beginning and end of an introductory programming course, focusing on demographics, expectations of success, and perceptions of challenges. Application status for the computing major was also recorded. Our results revealed that men and students with PE generally perceived greater potential for success and reported facing fewer challenges. In contrast, women and students without PE more often indicated concerns about intellectual ability and perceived challenges less positively. Notably, while gender appears in the preceding results, an intersectional analysis indicates that PE is the central factor. PE is also linked to persistence in the field of computing. Our results further highlight the importance of providing students with opportunities to develop experience, as it can help shape their expectations, perceived challenges, and retention in computing.

Motivation and Objectives. Productive Failure is a problem-based learning technique where students attempt to solve a problem before receiving instruction in the topic. By design, students may not find a satisfying solution. Prior studies of Productive Failure in STEM contexts have been conducted in secondary or introductory college settings. Focusing primarily on exploring appropriate analysis and modeling techniques, these studies showed that a Productive Failure approach can lead to greater conceptual knowledge acquisition and transfer capabilities compared to »traditional«Direct Instruction techniques. In this study, we build on these studies along two dimensions: First, we report on the design and evaluation of a Productive Failure intervention in a more advanced undergraduate class: third-year Operating Systems. Second, our intervention targeted a more advanced skill: applying synchronization primitives, rather than selecting appropriate modeling and analysis techniques.

Recent work has confirmed that computing students experience the Imposter Phenomenon (IP) at higher rates than reported in other disciplines. However, no work has examined what aspects of the university computing experience might lead to a higher rate of IP experiences. We aim to illustrate the IP experiences students have, identify common sources of these experiences, and document the effects of these experiences and how students respond to them. We asked undergraduate students to share recent experiences that illustrate their experiences with the IP. We conducted an inductive thematic analysis on these open-ended responses, resulting in a set of inter-connected themes. A significant fraction of students related stories about making comparisons with peers or observing peer behaviour that made them question their abilities. Students also spoke about holding unrealistic expectations learned from their peers or imposed by the environment. These experiences may be particularly acute for minority-affiliated students who may come to feel they do not belong. Ultimately, these IP experiences can lead to a loss of motivation or a cycle of failure that leads students to leave computing. The central role social comparisons play in IP experiences suggests that it is particularly important to foster communities where opportunities for comparison are reduced and where realistic expectations are explicitly set.

Generating self-explanations has been identified as a successful strategy in helping learners engage with course content and organize what they learn in a structured format. While typing an explanation may allow more structure and formality, explaining by voice can be more natural and help free cognitive resources to focus on learning goals and understanding concepts. As we investigated the effects and students' perceptions of using voice or text to self-explain new course concepts, we failed to find a tool that would meet our needs. We present our work in designing and developing VoiceEx, a submission courseware that allows text and voice input to collect data in both mediums. VoiceEx was created to support a self-explanations intervention for computer science students; however, given its features and the advantages of being able to collect spoken responses, it can be used in a variety of environments. Future refinement of this tool includes artificial intelligence features to better guide students' submissions.

In this poster, we present a pilot study investigating the impact of the medium used for self-explanation on students' performance outcomes in a databases course. We did not see notable differences in student performance based on the medium they used for self-explanation. We also note that while most students prefer using text to submit self-explanations, their preferences may differ when they use voice.

The recent advent of highly accurate and scalable large language models (LLMs) has taken the world by storm. From art to essays to computer code, LLMs are producing novel content that until recently was thought only humans could produce. Recent work in computing education has sought to understand the capabilities of LLMs for solving tasks such as writing code, explaining code, creating novel coding assignments, interpreting programming error messages, and more. However, these technologies continue to evolve at an astonishing rate leaving educators little time to adapt. This working group seeks to document the state-of-the-art for code generation LLMs, detail current opportunities and challenges related to their use, and present actionable approaches to integrating them into computing curricula.

Q&A forums are widely used in large classes to provide scalable support. In addition to offering students a space to ask questions, these forums aim to create a community and promote engagement. Prior literature suggests that the way students participate in Q&A forums varies and that most students do not actively post questions or engage in discussions. Students may display different participation behaviours depending on their comfort levels in the class. This paper investigates students' use of a Q&A forum in a CS1 course. We also analyze student opinions about the forum to explain the observed behaviour, focusing on students' lack of visible participation (lurking, anonymity, private posting). We analyzed forum data collected in a CS1 course across two consecutive years and invited students to complete a survey about perspectives on their forum usage. Despite a small cohort of highly engaged students, we confirmed that most students do not actively read or post on the forum. We discuss students' reasons for the low level of engagement and barriers to participating visibly. Common reasons include fearing a lack of knowledge and repercussions from being visible to the student community.

Large language models (LLMs) can be used to answer student questions on course discussion boards, but there is a risk of LLMs answering questions they are unable to address. We propose and evaluate an LLM-based system that classifies student questions into one of four types: conceptual, homework, logistics, and not answerable. We then prompt an LLM using a type-specific prompt. Using GPT-3, we achieve 81% classification accuracy across the four categories. Furthermore, we achieve 93% accuracy on classifying not answerable questions. This indicates that our system effectively ignores questions that it cannot address.

Previous work has reported on the advantageous effects of prior experience in CS1, but it remains unclear whether these effects fade over a sequence of introductory programming courses. Furthermore, while student perceptions suggest that prior experience remains important, studies have reported that a student's expectation of their performance is a more accurate predictor of outcome. We aim to confirm if prior experience (formal or informal) provides short-term and long-term advantages in computing courses or if the advantage fades. Furthermore, we explore whether the expectation of performance is a more accurate predictor of student success than informal and formal prior experience. To explore these questions, we deployed surveys in a CS1 course to gauge students' level of prior experience in programming, prediction of final exam grades, and self-efficacy to succeed in university. Grades from CS1 and CS2 were also collected. We observed a persistent (1-letter grade) gap between the performance of students with no prior experience and those with any experience, but we did not observe a noteworthy gap when comparing student performance based on formal or informal experience. We also observed differences in self-efficacy and retention rates between different levels of prior experience. Lastly, we confirm that success in CS1 can be better reflected and predicted by some controllable factors, such as students' perceptions of ability.

Many students are first exposed to computing in a programming course such as CS1. This course affects their understanding of computing and may affect their intention to major in the program. We investigate the intention to major in computing in relation to demographic factors and factors related to academic success. We deployed surveys at the beginning and end of a CS1 course to gauge students' level of prior experience in programming, elicit demographic factors such as gender and parental education level, and identify their intention to major in computing. Grades from CS1 and CS2 were also collected. Our results suggest that most students do not change their intention to major in computing after taking CS1. Students who were more likely to intend to major in programming at the beginning of the course were those with prior experience, those who identified as men, or students who had a parent with a bachelor's or post-grad degree. We also find that students' grades correlate to their change in program intention. This reinforces the need to change perceptions about computing early, prior to CS1.

Instructors often prefer to use email for course communication. The use of emails has been widely discussed in the fields of marketing and behavioural design, but the prevalence of email in education makes it important for instructors to collect metrics on emails to see how students engage with them. One component of emails are the subject lines, which constitute as one of the first things a receiver sees before deciding to open an email. This poster discusses a case study at deploying an email intervention in an online CS1 course. We investigate how the length of subject lines impact the rate at which students open emails of a particular type that prompts them to start their homework early. We aim to share key results to inform instructors how to design their emails to better reach students. Further, we highlight the potential benefits for instructors when collecting and analyzing email engagement data.

Writing skills are often considered unimportant by computer science students and were under-emphasized in our curriculum. We describe our experience embedding CS-specific writing instruction at scale in most of our large, core, first- and second-year Computer Science courses, each with 300-800+ students. Our approach is to collaborate with a writing specialist and a community of course instructors, centralize the management of writing teaching assistants, and introduce a variety of relevant genres and contexts to help students develop and apply writing skills. We outline the institutional support and organization crucial to a project of this scale. In addition, we report on a survey collecting student perception of the writing instruction/assessment. We reflect on quantitative and qualitative evidence of success, as well as the challenges that we faced. We believe that many of these challenges will be common across institutions, particularly those with large courses.

Every year, millions of students learn how to write programs. Learning activities for beginners almost always include programming tasks that require a student to write a program to solve a particular problem. When learning how to solve such a task, many students need feedback on their previous actions, and hints on how to proceed. In the case of programming, the feedback should take the steps a student has taken towards implementing a solution into account, and the hints should help a student to complete or improve a possibly partial solution. Only a limited number of learning environments for programming give feedback and hints on intermediate steps students take towards a solution, and little is known about the quality of the feedback provided. To determine the quality of feedback of such tools and to help further developing them, we create and curate data sets that show what kinds of steps students take when solving programming exercises for beginners, and what kind of feedback and hints should be provided. This working group aims to 1) select or create several data sets with steps students take to solve programming tasks, 2) introduce a method to annotate students' steps in these data sets, 3) attach feedback and hints to these steps, 4) set up a method to utilize these data sets in various learning environments for programming, and 5) analyse the quality of hints and feedback in these learning environments.

Motivation Bots can alleviate the workload of instructors supporting students in large course Q&A platforms, but it's not clear whether students will be receptive to the use of automated assistants in this setting. Objectives We aim to observe student reactions when they encounter bot-generated follow-ups to Q&A board posts. We investigate the effect of revealing that a bot, rather than a human, is suggesting that the current post is a duplicate. Methods Our bot revealed or hid its bot identity when suggesting duplicate posts, with the condition selected randomly. We observed students' reactions in both conditions. A post-course survey was distributed to collect students' demographic data, previous experiences with bots, and attitudes toward our bot. Results We observed a slight increase in students' response rate when the bot hid its identity. We compared the positive response rate in both conditions and did not find evidence suggesting that students had less trust in bot-generated answers. From the survey, we only saw minimal direct evidence that students might mistrust the bot: 7 of 59 students reported worries about receiving an inaccurate bot-generated answer. Other students were concerned that they would not receive attention from an instructor. Discussion We did not find evidence that revealing the bot's identity has a negative impact on student reactions. However, future bot design should consider the emotional impact of deploying a bot as there may be negative emotional effects to receiving a bot-generated response.

Email communication between instructors and students is ubiquitous, and it could be valuable to explore ways of testing out how to make email messages more impactful. This paper explores the design space of using emails to get students to plan and reflect on starting weekly homework earlier. We deployed a series of email reminders using randomized A/B comparisons to test alternative factors in the design of these emails, providing examples of an experimental paradigm and metrics for a broader range of interventions. We also surveyed and interviewed instructors and students to compare their predictions about the effectiveness of the reminders with their actual impact. We present our results on which seemingly obvious predictions about effective emails are not borne out, despite there being evidence for further exploring these interventions, as they can sometimes motivate students to attempt their homework more often. We also present qualitative evidence about student opinions and behaviours after receiving the emails, to guide further interventions. These findings provide insight into how to use randomized A/B comparisons in everyday channels such as emails, to provide empirical evidence to test our beliefs about the effectiveness of alternative design choices.

Motivation Despite the widespread belief that computing practitioners frequently experience the Imposter Phenomenon (IP), little formal work has measured the prevalence of IP in the computing community despite its negative effect on achievement.

In online asynchronous learning environments, students are assigned exercises, but it is not clear how to incorporate the kinds of actions an in-person tutor might take such as explaining, providing more practice, prompting for reflection, and motivating. We explore approaches to adding "Drop-Downs'' that appear after a student submits an answer and that contain additional information to support learning. We conducted randomized A/B experiments exploring the impact of these Drop-Downs on student learning in the online portion of a flipped CS1 course. The deployed Drop-Downs in this course provided explanations, reflective prompts, additional problems, and motivational messages. The results suggest that students benefit from various Drop-Downs in different contexts, indicating the possibility of personalizing content based on the student's state. We discuss the resulting design implications of Drop-Downs in online learning systems.

Procrastination by students may lead to adverse outcomes such as a focus on completion rather than learning or even a failure to complete learning tasks. One common method for motivating students and reducing procrastination is to send reminders with hints and study strategies, but it's not clear if these messages are effective or when is the best time to send them. Randomized A/B comparisons could be used to try different reminders or alternative ideas about how best to get students to start work earlier and, crucially, to measure the impact of these interventions on behaviour. This paper describes an A/B comparison of reminder emails set in a large CS1 course at a research-focused North American university. We found evidence that the email interventions caused a higher proportion of students to attempt the online homework but did not see evidence that these particular emails got students to start early, irrespective of changes to the timing of the reminder. More broadly, these findings illustrate how to use A/B comparisons in educational settings to test ideas about how to help students, and demonstrate the value of using randomized A/B comparisons, even when evaluating actions that seem obviously beneficial, such as reminder emails.

Engaged preparation and study in combination with lectures are important for all courses but are particularly critical for online, hybrid, and inverted classrooms. Many instructors use online systems to deliver new course content and exercises, but students often delay assignments or game these systems (e.g., guessing on multiple-choice questions), often to the detriment of their learning. In an inverted CS1 course, many students self-reported high rates of gaming-the-system behavior, so we examine survey data to identify factors that contribute to engagement in these maladaptive behaviours. We supplement that analysis with interview data to gain a deeper understanding of the situation. We also implemented and evaluated a previously reported online intervention aimed at reducing gaming behavior. Unlike prior work, our intervention did not have a significant effect on guessing behavior. We discuss why the factors we identified might explain this result, as well as suggest future work to improve our understanding of gaming behaviours and inform the design of systems that encourage effective learning.

The number of online classroom environments in universities has been steadily rising. As a result, it has become more important to understand how students utilize these online environments, as well as to identify the most effective learning strategies. This research identifies strategies used by students in an online tool associated with a first-year university computer science class. Utilizing sequential pattern mining, we identify common strategies used by high- and low-performing students. The goal is to understand how higher performing students utilize an unsupervised online learning environment to identify effective strategies in this setting. This will allow instructors to direct students toward effective methods of using online classroom tools.

We report our experience of preparing and conducting unplanned online exams in the unique half-physical, half-virtual semester of Winter 2020. The report covers four courses in a large university's computer science program, ranging from first-year to third-year. With the data generated by students taking both in-person and online exams in multiple courses, we perform analyses to evaluate the validity of the online exams (especially the unproctored ones) as an assessment of student understanding. With the fine-grained student activity data provided by the online exam platform, we are also able to investigate the patterns in student exam-taking behaviours and their correlations with student performance on the exam. In addition, we share, in detail, the tips and lessons that were learned throughout the process of designing, implementing, and hosting the online exams.

For many years, the C++-based Computer Programming Self-Efficacy Scale by Ramalingam and Wiedenbeck has been the de facto standard for assessing self-efficacy in introductory programming. Since the development of this instrument, however, both the landscape as well as the intended audience of introductory programming courses has changed beyond the use of a particular programming language. We revisit this instrument and its factorization in light of curricular developments and research results regarding concepts and competences taught in introductory courses. We report on the development and validation of a new instrument that covers most paradigms and languages used in CS1 and present exploratory and confirmatory factor analyses across different populations. Our validation and factor analyses suggest that the new instrument indeed measures self-efficacy with an acceptable fit of the model. In contrast, the factorization of the Computer Programming Self-Efficacy Scale was found to be less robust. Nonetheless, and in line with self-efficacy theory, our analyses suggest that researchers should take into account the educational context of the study population when reporting or comparing results at the level of factors.

A notional machine is a pedagogic device to assist the understanding of some aspect of programs or programming. It is typically used to support explaining a programming construct, or the user-understandable semantics of a program. For example, a variable is like a box with a label, and assignment copies or moves a value into that box. This working group will capture examples of notional machines from actual pedagogical practice, as expressed in textbooks (or other teaching materials) or used in the classroom. We will interview at least 30 teachers about their experience with, and perceptions of, the use of notional machines in teaching. Using the interviews, we will work on devising and refining a form to characterise essential features of notional machines. We will also attempt to relate them to each other to describe potential learning sequences or progressions. The working group report will contain descriptions of notional machines used at different levels in education, in different countries, by many teachers. Capturing and Characterising Notional Machines Sally Fincher, Johan Jeuring, Craig S Miller Permission to make digital or hard copies of part or all of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for third-party components of this work must be honored. For all other uses, contact the owner/author(s). ITiCSE 2020,,Trondheim, Norway © 2020 Copyright held by the owner/author(s). 978-1-4503-0000-0/18/06...$15.00 https://doi.org/10.1145/1234567890 The resulting catalogue of notional machines will allow a teacher to select a machine for a particular use, permit comparison between them, and provide a starting point for further categorization and analysis of notional machines. Additionally, we will make more theoretical explorations. We will explore a variety of presentational formats, examining what is necessary and what superfluous; we will look for dimensions of comparison and will examine how notional machines are instantiated across the discipline. We argue that the creation and use of notional machines is potentially a signature pedagogy for computing [1] and that creating and using notional machines represents a certain level of pedagogic sophistication that might be an indicator of pedagogic content knowledge (PCK).

In this paper, we introduce ProgSnap2, a standardized format for logging programming process data. ProgSnap2 is a tool for computing education researchers, with the goal of enabling collaboration by helping them to collect and share data, analysis code, and data-driven tools to support students. We give an overview of the format, including how events, event attributes, metadata, code snapshots and external resources are represented. We also present a case study to evaluate how ProgSnap2 can facilitate collaborative research. We investigated three metrics designed to quantify students' difficulty with compiler errors - the Error Quotient, Repeated Error Density and Watwin score - and compared their distributions and ability to predict students' performance. We analyzed five different ProgSnap2 datasets, spanning a variety of contexts and programming languages. We found that each error metric is mildly predictive of students' performance. We reflect on how the common data format allowed us to more easily investigate our research questions.

The struggles of first year computer science students has been extensively studied. Work has focused on students who choose not to complete introductory programming courses (CS1, CS2), with the decision to not complete attributed to a lack of prior programming experience, difficulty in grasping course content, and changes in personal priorities. The experience of students in the second year, however, is less well studied. We conducted an interview-based study of students in the second year who have dropped a second year compute science course. Our goal is to better understand student challenges that lead to the decision to drop in the second year and to compare the factors that are identified to those that are relevant to non-completion of the first year.

Computing education research is built on the use of suitable methods within appropriate theoretical frameworks to provide guidance and solutions for our discipline, in a way that is rigorous and repeatable. However, the scale of theory covered extends well beyond the CS discipline and includes educational theory, behavioural psychology, statistics, economics, and game theory, among others. A computing education researcher's journey towards appropriate and discipline relevant theory can be challenging and, when a researcher has learned one area of theory, it can be easy to return to familiar theory, as it may not be clear what the next step could be. The periodic table is a visual arrangement of the elements to group like with like, providing insight into how families of elements will react. Could we do the same with learning theories located in the domain of computer science education, and would it be useful? The working group will identify and survey existing literature on relationships between key areas of theory in computing education, identify ways of organising these research areas to show how knowledge of one could assist another, and produce initial graphical representations of theory and their relationship groupings to assist researchers in understanding how computing theory is currently used in the discipline and what theories might become of interest.

This report documents the implementation of a self-paced, mastery learning inspired CS1 course. The course was designed to increase the completion rates observed in flipped and online CS1 formats already offered at our institution. We explore the experience of students in the course and evaluate performance outcomes using grade data from all three CS1 formats, student survey responses, and exit interviews. Our evaluation identifies three main challenges in our implementation. First, the course requires significant resources and administering it is significantly more time consuming for instructors than a regular course. Second, students hesitated to treat mastery quizzes as formative. Finally, the flexibility that the course provided, with little structure and few incentives to help students stay on track, led to considerable procrastination. These factors combined to lead students to delay coursework until the end of the semester -- and beyond. As a result, while our data shows an increase in completion relative to the online format, we saw no change in completion relative to the flipped CS1 offering and saw no change in student performance as evaluated by a final exam. However, students reported more deep engagement with and understanding of the material, which encourages us to further develop the course.

Students learning to program often rely on feedback from the compiler and from instructor-provided test cases to help them identify errors in their code. This feedback focuses on functional correctness, and the output, which is often phrased in technical language, may be difficult to for novices to understand or effectively use. Static analyses may be effective as a complementary aid, as they can highlight common errors that may be potential sources of problems. In this paper, we introduce PyTA, a wrapper for pylint that provides custom checks for common novice errors as well as improved messages to help students fix the errors that are found. We report on our experience integrating PyTA into an existing online system used to deliver programming exercises to CS1 students and evaluate it by comparing exercise submissions collected from the integrated system to previously collected data. This analysis demonstrates that, for students who chose to read the PyTA output, we observed a decrease in time to solve errors, occurrences of repeated errors, and submissions to complete a programming problem. This suggests that PyTA, and static analyses in general, may help students identify functional issues in their code not highlighted by compiler feedback and that static analysis output may help students more quickly identify debug their code.

Experts and practitioners have long published reflections on how mathematics connects with computer science and how best to integrate mathematics within computing education. However, little literature exists that investigates this area empirically. One recent study used in-depth interviews to generate a theory of the relationship between student perceptions of mathematics and their attitude toward computer science. However, this theory is based on interactions with a limited number of students. Our study tests this theory in a larger population. We present a survey-based exploration of the attitudes 218 undergraduate computer science students hold on mathematics. We present quantitative evidence that supports the theory that student career inclinations are correlated with beliefs about mathematics. We also present additional qualitative data, obtained from free-response sections of the survey, that support the idea that student inclinations affect both their perception of the value of mathematics and academic decisions such as course selection.

Preparing for exams is an important yet stressful time for many students. Self-testing is known to be an effective preparation strategy, yet some students lack motivation to engage or persist in self-testing activities. Adding game elements to a platform supporting self-testing may increase engagement and, by extension, exam performance. We conduct a randomized controlled experiment (n=701) comparing the effect of two game elements -- a points system and a badge system -- used individually and in combination. We find that the badge system elicits significantly higher levels of voluntary self-testing activity and this effect is particularly pronounced amongst a relatively small cohort. Importantly, this increased activity translates to a significant improvement in exam scores. Our data supports a causal relationship between gamification and learning outcomes, mediated by self-testing behavior. This provides empirical support for Landers' theory of gamified learning when the gamified activity is conducted prior to measuring learning outcomes.

Online coding tools are an increasingly common feature of programming courses, providing students with rapid feedback and flexible practice opportunities and providing instructors with useful analytics. However, little research has explored the complexity of online exercises provided to students and the order in which students are exposed to new ideas. In this paper, we investigate the benefits of using a short sequence of practice exercises, each targeting a distinct topic, prior to having students solve a goal task that combines the concepts. As expected, we find students solve the goal task with fewer errors and in less time after completing the practice tasks. However, we also find that the practice tasks reduce the likelihood of students delaying work on the goal task, and these effects are particularly large for less-experienced students.

Since computing education began, we have sought to learn why students struggle in computer science and how to identify these at-risk students as early as possible. Due to the increasing availability of instrumented coding tools in introductory CS courses, the amount of direct observational data of student working patterns has increased significantly in the past decade, leading to a flurry of attempts to identify at-risk students using data mining techniques on code artifacts. The goal of this work is to produce a systematic literature review to describe the breadth of work being done on the identification of at-risk students in computing courses. In addition to the review itself, which will summarize key areas of work being completed in the field, we will present a taxonomy (based on data sources, methods, and contexts) to classify work in the area.

Computing educators have used code review to provide opportunities for students to engage with code artifacts and to learn about aspects of programming, such as design and style, that are difficult to appreciate when working individually. However, most implementations of code review have been implemented in relatively small courses. For the past two years, we have asked students in our 500-950 person first year courses to engage in code reviews of submissions to course assignments. This short paper describes our approach for implementing code reviews that scale to large first-year courses and describes a few lessons we have learned from deploying these learning opportunities in our courses.

Peer Instruction (PI) is an active learning pedagogy that has been shown to improve student outcomes in computing, including lower failure rates, higher exam scores, and better retention in the CS major. PI's key classroom mechanism is the PI question: a formative multiple choice question on which students vote, then discuss, then vote again. While research indicates that PI questions lead to learning gains for students, relatively little is known about the questions themselves and how faculty employ them. Additionally, much of the work has examined PI data collected by researchers operating in a quasi-experimental setting. We examine data collected incidentally by multiple instructors using PI as a pedagogical technique in their classroom. We look at how many questions instructors use in their courses, the difficulty level of the questions, and normalized gain, a metric that looks at increases in student correctness between individual and group votes. We find normalized gain levels similar to those in existing literature, indicating that students are learning, and that most questions, even those developed by instructors new to PI, fall within recommended difficulty levels, indicating instructors can create good PI questions with little training. We also find that instructors add PI questions over the first several iterations of a new PI course, showing that they find PI questions valuable and suggesting that full development of PI materials for a course may take multiple semesters.

Peer Instruction is an active learning pedagogy that has been shown to improve student outcomes in computing, including lower failure rates, higher exam scores, and better retention in the CS major. A key classroom mechanism for Peer Instruction is the "clicker question": a formative multiple-choice question on which students vote, then discuss, then vote again. While research indicates that clicker questions lead to learning gains for students, relatively little is known about the questions themselves and how faculty employ them. Additionally, much of the work has examined clicker data collected by CS Education researchers operating in a quasi-experimental setting. In this project, we examine clicker data collected incidentally by multiple instructors using clickers as a pedagogical technique in their classroom. This work represents a first effort to systematically evaluate how instructors use clicker questions, including how many clicker questions are used in a course, how difficult the questions used are, and whether instructors add or modify questions over time.

Students and instructors expend significant effort, respectively, preparing to be examined and preparing students for exams. This paper investigates question authoring, where students create practice questions as a preparation activity prior to an exam, in an introductory programming context. The key contribution of this study as compared to previous work is an improvement to the design of the experiment. Students were randomly assigned the topics that their questions should target, removing a selection bias that has been a limitation of earlier work. We conduct a large-scale between-subjects experiment (n = 700) and find that students exhibit superior performance on exam questions that relate to the topics they were assigned when compared to those students preparing questions on other assigned topics.

Current CS1 learning outcomes are relatively general, specifying tasks such as designing, implementing, testing and debugging programs that use some fundamental programming constructs. These outcomes impact what we teach, our expectations, and our assessments. Although prior work has demonstrated the utility of single concept assessments, most assessments used in formal examinations combine numerous heterogeneous concepts, resulting in complex and difficult tasks.

Course instructors need to be able to identify students in need of assistance as early in the course as possible. Recent work has suggested that machine learning approaches applied to snapshots of small programming exercises may be an effective solution to this problem. However, these results have been obtained using data from a single institution, and prior work using features extracted from student code has been highly sensitive to differences in context. This work provides two contributions: first, a partial reproduction of previously published results, but in a different context, and second, an exploration of the efficacy of neural networks in solving this problem. Our findings confirm the importance of two features (the number of steps required to solve a problem and the correctness of key problems), indicate that machine learning techniques are relatively stable across contexts (both across terms in a single course and across courses), and suggest that neural network based approaches are as effective as the best Bayesian and decision tree methods. Furthermore, neural networks can be tuned to be reliably pessimistic, so they may serve a complementary role in solving the problem of identifying students who need assistance.

Recent years have seen increasing interest in using programming snapshot data for education research. One barrier to such research, especially for studies involving data from multiple institutions, is that the data is in a wide variety of native formats, and those formats may not be conducive to automated analysis. To overcome this barrier, we propose a structured data model and archival data format called Progsnap (https://cloudcoderdotorg.github.io/progsnap-spec/). Progsnap is designed to be a neutral export format, is currently supported by two open-source programming exercise systems, and we believe will be an easy target for data export from other systems. An open source Python library makes it easy to automate analysis of Progsnap datasets.

The abstraction of student code for use in automated analysis is a key challenge. The code must be processed in a manner that reveals interesting properties while reducing the "noise" introduced by less important details. In this work, we investigate the importance of control flow as a property in the analysis of students' programming processes.

Increasing enrollments and adoption of online resources have encouraged the use of multiple choice questions as a means of providing scalable assessment. However, in contexts where formative feedback is desired, standard multiple choice questions may lead students to a false sense of confidence -- a result of their small solution space and the temptation to guess. We propose the use of multiple-answer multiple choice questions in situations where formative feedback is desired and present evidence that these questions are well suited for that role.

The subject of this work is evaluation methods of student software development team projects. Over a two year period in three courses, we study student preferences for team evaluation methods, the reasons behind their preferences, the changes in preference that occur within a single course and across courses, and misconceptions that educators hold about student preferences.

Many database courses rely on relational algebra (RA) to provide a theoretical foundation for database query languages such as SQL. However, few tools exist to support students in learning RA. To fill this need, we created RAPT. RAPT uses a syntactic and semantic understanding of RA to transform input statements into a variety of outputs, including LATEX formatted queries, parse tree diagrams, and executable SQL statements. The translation to SQL is particularly important, as it enables the creation of automatically tested exercises and allows students to view the result of executing an RA statement. RAPT-supported exercises have been integrated into PCRS, a system for creating online learning modules, and deployed to a third year databases course with over 350 students.

The C programming language is an important piece of many undergraduate CS programs, as it provides an environment for interacting directly with memory and exploring systems-programming concepts. However, while many common introductory languages have rich tools that support instruction, C has received relatively little attention [2, 1]. To provide students with rapid feedback and tools for understanding C, we have extended PCRS, a web-based platform for deploying programming exercises and content such as videos. Students submit C code to solve programming exercises and receive immediate feedback generated by running the submission against a set of instructor-defined testcases. Students also have access to graphical traces of execution, so they can explore how their code manipulates memory. The system has been deployed to two second-year systems-programming courses with a total enrollment over 600, and a set of modules, consisting of videos and exercises, is being developed for use by the community.

Project experience and teamwork have been identified as two of the most important deficiencies of recent graduates, so experiences with team projects are a critical component of computer science and software engineering education. However, evaluating these projects is difficult, as it requires a balance between rewarding the team's effort (and the development of skills that enable the team to work effectively) and recognizing individual contributions. We report on an investigation of the student perspective on evaluation of software development team projects. We find that (1) computer science and software engineering educators hold several misconceptions about students' preferences in teamwork evaluation methods, (2) that students' preferences with regard to evaluation change dramatically as they advance through the program and even in the course of a single term, and (3) that the change in preference occurs early in the term, before they complete any work as a team, with most students shifting their preference towards putting more weight on the team's effort.

We examine student difficulties with CS1 concepts by analyzing a dataset containing 266,852 student responses to weekly code-writing problems. We find that conditionals and loops prove particularly problematic, even when considering 'second chance' data; and that, while we observe some evidence of improvement, certain straightforward applications of loops continue to be problematic at the end of the term. Our contribution is the corroboration of earlier findings, and a call to use online repositories of student submissions as rich sources of data on the student learning experience.

Short, automatically-assessed programming exercises, and other types of short practice problems, are a useful way to introduce and reinforce concepts and techniques in introductory programming courses. When delivered over the web, they allow students to learn and practice, with immediate feedback, at any time and place where they have access to a web browser. However, such exercises do not seem to be as widely used as they could be. Similarly, there is not a lot of literature on the effectiveness of these types of problems. The purpose of this BOF is to bring together users (and potential users) of programming exercises with developers of programming exercise systems to discuss how exercises could be used more widely and effectively. Possible discussion topics include: What features are absolutely essential for faculty to consider adoption? What are the major obstacles preventing more widespread adoption? Are faculty willing to share their exercises under an open/non-commercial license? Should exercises best used for extra practice, as graded assignments, or both?

This work explores the effects of prior exposure to programming on student experiences in an introductory computer programming course CS1). We take a student-focused approach: using a combination of surveys and semi-structured interviews, we obtain information on the student experiences in CS1 and their relationship to prior programming experience (PE). The analysis of the results provides insight into the following two questions: a) How does PE affect peer interaction: pair programming sessions, assignment partnerships, and online and in-class interactions? b) What are the students' beliefs on the relationship between PE and success in the course?

We present the Python Classroom Response System, a web-based tool that enables instructors to use code-writing and multiple choice questions in a classroom setting. The system is designed to extend the principles of peer instruction, an active learning technique built around discussion of multiple- choice questions, into the domain of introductory programming education. Code submissions are evaluated by a suite of tests designed to highlight common misconceptions, so the instructor receives real-time feedback as students submit code. The system also allows an instructor to pull specific submissions into an editor and visualizer for use as in-class examples. We motivate the use of this system, describe its support for and extension of peer instruction, and offer use cases and scenarios for classroom implementation.

This paper presents observations about teaching program tracing to novices drawn from a study of knitting patterns. Due to changes in audience, knitting patterns have evolved from vague, chatty discourse written for experts to precise, line-by-line procedures that are akin to programs. The modern knitting community has developed numerous conventions for articulating iteration, expressing conditions, and documenting design decisions. "Executing" one of these patterns is analogous to tracing, since the knitter must demonstrate understanding of the instructions. We argue that the conventions adopted by knitters to make their patterns more understandable to non-experts provide useful insight to computer scientists teaching novices. Our observations suggest that phrasing conditions as termination cases ("until" instead of "while") and partially unrolling loops may help beginners understand code and that some structures, like parameters to functions, may be unfamiliar because they have no common analog.

In this paper, we explore the use of sequences of small code writing questions ("concept questions") designed to incrementally evaluate single programming concepts. We report on a study of student performance on a CS1 final examination that included a traditional code-writing question and four intentionally corresponding concept questions. We find that the concept questions are significant predictors of performance on both the corresponding code-writing question and the final exam as a whole. We argue that concept questions provide more accurate formative feedback and simplify marking by reducing the number of variants that must be considered. An analysis of responses categorized by the students' previous programming experience suggests that inexperienced students have the most to gain from the use of concept questions.

Many factors have been cited for poor performance of students in CS1. To investigate how assessment mechanisms may impact student performance, nine experienced CS1 instructors reviewed final examinations from a variety of North American institutions. The majority of the exams reviewed were composed predominantly of high-value, integrative code-writing questions, and the reviewers regularly underestimated the number of CS1 concepts required to answer these questions. An evaluation of the content and cognitive requirements of individual questions suggests that in order to succeed, students must internalize a large amount of CS1 content. This emphasizes the need for focused assessment techniques to provide students with the opportunity to demonstrate their knowledge.

Silicon technology will continue to provide an exponential increase in the availability of raw transistors. Effectively translating this resource into application performance, however, is an open challenge that conventional superscalar designs will not be able to meet. We present WaveScalar as a scalable alternative to conventional designs. WaveScalar is a dataflow instruction set and execution model designed for scalable, low-complexity/high-performance processors. Unlike previous dataflow machines, WaveScalar can efficiently provide the sequential memory semantics that imperative languages require. To allow programmers to easily express parallelism, WaveScalar supports pthread-style, coarse-grain multithreading and dataflow-style, fine-grain threading. In addition, it permits blending the two styles within an application, or even a single function. To execute WaveScalar programs, we have designed a scalable, tile-based processor architecture called the WaveCache. As a program executes, the WaveCache maps the program's instructions onto its array of processing elements (PEs). The instructions remain at their processing elements for many invocations, and as the working set of instructions changes, the WaveCache removes unused instructions and maps new ones in their place. The instructions communicate directly with one another over a scalable, hierarchical on-chip interconnect, obviating the need for long wires and broadcast communication. This article presents the WaveScalar instruction set and evaluates a simulated implementation based on current technology. For single-threaded applications, the WaveCache achieves performance on par with conventional processors, but in less area. For coarse-grain threaded applications the WaveCache achieves nearly linear speedup with up to 64 threads and can sustain 7--14 multiply-accumulates per cycle on fine-grain threaded versions of well-known kernels. Finally, we apply both styles of threading to equake from Spec2000 and speed it up by 9x compared to the serial version.

In recent years, computer architects have proposed tiled architectures in response to several emerging problems in processor design, such as design complexity, wire delay, and fabrication reliability. One of these architectures, WaveScalar, uses a dynamic, tagged-token dataflow execution model to simplify the design of the processor tiles and their interconnection network and to achieve good parallel performance. However, using a dataflow execution model reawakens old problems, including the instruction overhead required for control flow. Previous work compiling the functional language Id to the Monsoon Dataflow System found this overhead to be 2–3× that of programs written in C and targeted to a MIPS R3000.In this paper, we present and analyze three compiler optimizations that significantly reduce control overhead with minimal additional hardware. We begin by describing how to translate imperative code into dataflow assembly and analyze the resulting control overhead. We report a similar 2–4× instruction overhead, which suggests that the execution model, rather than a specific source language or target architecture, is responsible. Then, we present the compiler optimizations, each of which is designed to eliminate a particular type of control overhead, and analyze the extent to which they were able to do so. Finally, we evaluate the effect using all optimizations together has on program performance. Together, the optimizations reduce control overhead by 80% on average, increasing application performance between 21–37%.

This paper explores hierarchical instruction scheduling for a tiled processor. Our results show that at the top level of the hierarchy, a simple profile-driven algorithm effectively minimizes operand latency. After this schedule has been partitioned into large sections, the bottom-level algorithm must more carefully analyze program structure when producing the final schedule.Our analysis reveals that at this bottom level, good scheduling depends upon carefully balancing instruction contention for processing elements and operand latency between producer and consumer instructions. We develop a parameterizable instruction scheduler that more effectively optimizes this trade-off. We use this scheduler to determine the contention-latency sweet spot that generates the best instruction schedule for each application. To avoid this application-specific tuning, we also determine the parameters that produce the best performance across all applications. The result is a contention-latency setting that generates instruction schedules for all applications in our workload that come within 17% of the best schedule for each.

Tiled architectures, such as RAW, SmartMemories, TRIPS, and WaveScalar, promise to address several issues facing conventional processors, including complexity, wire-delay, and performance. The basic premise of these architectures is that larger, higher-performance implementations can be constructed by replicating the basic tile across the chip. This paper explores the area-performance trade-offs when designing one such tiled architecture, WaveScalar. We use a synthesizable RTL model and cycle-level simulator to perform an area/performance pareto analysis of over 200 WaveScalar processor designs ranging in size from 19mm2to 575mm2and having a 22 FO4 cycle time. We demonstrate that, for multi-threaded workloads, WaveScalar performance scales almost ideally from 19 to 101mm2when optimized for area efficiency and from 44 to 202mm2when optimized for peak performance. Our analysis reveals that WaveScalar's hierarchical interconnect plays an important role in overall scalability, and that WaveScalar achieves the same (or higher) performance in substantially less area than either an aggressive out-of-order superscalar or Sun's Niagara CMP processor

In response to current technology scaling trends, architects are developing a new style of processor, known as spatial computers. A spatial computer is composed of hundreds or even thousands of simple, replicated processing elements (or PEs), frequently organized into a grid. Several current spatial computers, such as TRIPS, RAW, SmartMemories, nanoFabrics and WaveScalar, explicitly place a program's instructions onto the grid. Designing instruction placement algorithms is an enormous challenge, as there are an exponential (in the size of the application) number of different mappings of instructions to PEs, and the choice of mapping greatly affects program performance. In this paper we develop an instruction placement performance model which can inform instruction placement. The model comprises three components, each of which captures a different aspect of spatial computing performance: inter-instruction operand latency, data cache coherence overhead, and contention for processing element resources. We evaluate the model on one spatial computer, WaveScalar, and find that predicted and actual performance correlate with a coefficient of -0.90. We demonstrate the model's utility by using it to design a new placement algorithm, which outperforms our previous algorithms. Although developed in the context of WaveScalar, the model can serve as a foundation for tuning code, compiling software, and understanding the microarchitectural trade-offs of spatial computers in general.