University of Cambridge > Talks.cam > Women@CL Events > Women@CL Talklet Event

Women@CL Talklet Event

Add to your list(s) Download to your calendar using vCal

  • UserCoral Westoby and Alessia Angeli and Smita Vijaya Kumar
  • ClockThursday 03 February 2022, 13:00-14:00
  • HouseZoom.

If you have a question about this talk, please contact Lorena Qendro.

Zoom link: https://cl-cam-ac-uk.zoom.us/j/91801402497?pwd=a293N3dHcVhMcEJyVGlHa1gzWE5sZz09

———————————————————————-

Speaker: Coral Westoby

Title: The POETS Project: Partially Ordered Event Triggered Systems for future HPC applications.

Abstract: The physical sciences have been intrigued by event driven computing systems since the first analog integrators. The POETS project (Cambridge lead S. Moore) provides a hardware platform to observe convincing weak scaling for simulations running under a Partially Ordered Event Triggered System, with the aim of motivating future HPC architectural developments. Current research activities include applications investigation in molecular mechanics, neural biology, and ML alongside the programming and architectural support required to use tens of thousands of threads where 4 instructions take longer than a message passing operation. Within the CL, a cluster of 72 Intel DE10 SX280 boards with 100Gb/s interconnect is currently under construction, providing a fabric an order of magnitude larger than currently available.

———————————————————————-

Speaker: Alessia Angeli

Title: Kinematics of reaching movements to inhibit a prepotent response: A wearable 3-axis accelerometer analysis.

Abstract: The aim of our research is to explore the distinctive contribution of motor planning and control to human reaching movements. In particular, the movements were triggered by the selection of a prepotent response (Dominant) or, instead, by the inhibition of the prepotent response, which required the selection of an alternative one (Non-dominant). To this end, we adapted a Go/NoGo task to investigate both the dominant and non-dominant movements of a cohort of 19 adults, utilizing kinematic measures to discriminate between the planning and control components of the two actions. In this experiment, a low-cost, easy to use, 3-axis wrist-worn accelerometer was put to good use to obtain raw acceleration data and to compute and break down its velocity components. The values obtained with this task indicate that with the inhibition of a prepotent response, the selection and execution of the alternative one yields both a longer reaction time and movement duration. Moreover, the peak velocity occurred later in time in the non-dominant response with respect to the dominant response, revealing that participants tended to indulge more in motor planning than in adjusting their movement along the way. Finally, comparing such results to the findings obtained by other means in the literature, we discuss the feasibility of an accelerometer-based analysis to disentangle distinctive cognitive mechanisms of human movements. In addition, the development of motor skills is strictly connected to the optimisation of cognitive abilities and difficulties to inhibit motor behaviours are common to neurodevelopmental disorders such as Attention Deficit and Hyperactivity Disorder (ADHD). However, the motor and cognitive processes beneath the profound interindividual differences that characterize this population are still unclear. For these reasons, we are working in this direction, considering the same experimental setting, in order to investigate both the dominant and non-dominant movements of a cohort of ADHD children.

———————————————————————-

Speaker: Smita Vijaya Kumar

Title: How To Simulate Standard Workload Traces on Kubernetes Cluster.

Abstract: Many companies have made available traces of workloads running on their compute clusters. The well-known ones include workloads from Google, Alibaba, Microsoft, and Yahoo. The workloads are a collection of requests, and each request is described by parameters such as the arrival time, number of tasks and task durations. How can one use these workloads and simulate them on a Kubernetes cluster? This talk will walk us through the steps involved in converting a large textual workload file consisting of tens of thousands of such requests into actual jobs that are consumed by Kubernetes.

———————————————————————-

This talk is part of the Women@CL Events series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2022 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity